Greg Shah wrote:

Constantin: Please evaluate the gaps that need to be resolved to convert all the OO built-in classes in the ADE project and document that work here.

I assume you refer to parse and run reports on it. I'll work on creating a separate project and parse these.

BTW, there are some ~6000 files in this project. I propose to create a separate project, outside of the main FWD runtime, and include this as a separate jar. FWD will remain with only progress.lang parts and whatever else we need to 'hand write' maybe.

There are less than 450 classes in src/corelib/ and src/netlib/. Isn't that all we're worried about? I don't think the src/businesscomponents/ is used much but could be wrong.

Anyway, we just care about the OO built-ins. I was expecting them to be part of FWD itself.

Greg Shah wrote:

There are less than 450 classes in src/corelib/ and src/netlib/. Isn't that all we're worried about? I don't think the src/businesscomponents/ is used much but could be wrong.

Anyway, we just care about the OO built-ins. I was expecting them to be part of FWD itself.

Ah, then makes sense, sorry for misunderstanding.

To make this work, we will have to implement anything missing in progress.lang.* since those are not written in 4GL code but instead are implemented in OE itself (C code).

It is more than just progress.lang.*, I think we will need to also implement:

• progress.applicationserver.*
• progress.io.*
• progress.json.*
• progress.reflect.*
• progress.util.*
• progress.web.*

I'm not as sure if we need these:

• progress.data.* (seems .NET related)
• progress.database.*
• progress.security.*
• progress.windows.* (.NET related)
• system.exception (.NET related)

• generic usage
  

  constructor private List (input pBackingList as Progress.Collections.IList<Object>):
  
  method public List<String> GetQueryValueList (input pcName as character):
  
  return (new List<String>()). // At least return an empty List.
  

• var defined without 'def var'
  

  var serializable private Progress.Collections.IList<Object> mBackingList.
  

• tenant-name(<dbname>) and tenant-id(<dbname>) handle methods
• final prop definition:
  

      define final override public property Request as IHttpRequest no-undo
          get():
              return NewRequest().
          end get.
  

• syntax like this - is this 'null protection':
  

  if not valid-object(pData?:ReturnValue?:OutputElement[loop]?:ElementType) then next RETURNLOOP.
  

We need these skeletons:

Cannot find class/interface Progress.Collections.HashMap in PROPATH.
Cannot find class/interface Progress.Collections.IComparable in PROPATH.
Cannot find class/interface Progress.Collections.IEqualityComparer in PROPATH.
Cannot find class/interface Progress.Collections.IHashable in PROPATH.
Cannot find class/interface Progress.Collections.IIterator in PROPATH.
Cannot find class/interface Progress.Collections.IList in PROPATH.
Cannot find class/interface Progress.Collections.ISet in PROPATH.
Cannot find class/interface Progress.Collections.KeyValuePair in PROPATH.
Cannot find class/interface Progress.Collections.List in PROPATH.
Cannot find class/interface Progress.Collections.SortedSet in PROPATH.
Cannot find class/interface Progress.IO.JsonSerializer in PROPATH.
Cannot find class/interface Progress.IO.MemoryInputStream in PROPATH.
Cannot find class/interface Progress.IO.MemoryOutputStream in PROPATH.
Cannot find class/interface Progress.Lang.Stop in PROPATH.
Cannot find class/interface Progress.Lang.StopAfter in PROPATH.
Cannot find class/interface Progress.Lang.UserInterrupt in PROPATH.
Cannot find class/interface Progress.OpenTelemetry.SpanContext in PROPATH.
Cannot find class/interface Progress.Web.AbstractWebRouter in PROPATH.

There may be more issues after these are fixed.

These are only for the corelib and netlib folders from the ADE project. The skeleton/oo4gl/OpenEdge/ was removed for this test.

• unsupported handle attributes: ROLES
• unsupported methods: get-cgi-value
• OO skeletons:
  

  Progress.Lang.Object.getclass()
  Progress.Lang.Object.tostring()
  Progress.Lang.Object.equals(KW_INPUT object<? extends progress.lang.object>)
  Progress.Lang.Class.isa(KW_INPUT object<? extends progress.lang.class>)
  Progress.Lang.Class.getclass(KW_INPUT character)
  Progress.Lang.Class.isinterface()
  Progress.Lang.Class.isabstract()
  Progress.Lang.Class.isfinal()
  Progress.Lang.Class.isserializable()
  Progress.Lang.Class.new()
  Progress.Lang.Object.clone()
  Progress.Lang.Class.new(KW_INPUT object<? extends progress.lang.parameterlist>)
  Progress.Lang.Class.invoke(KW_INPUT character, KW_INPUT object<? extends progress.lang.parameterlist>)
  Progress.Lang.Class.getpropertyvalue(KW_INPUT object<? extends progress.lang.object>, KW_INPUT character)
  Progress.Lang.Class.invoke(KW_INPUT object<? extends progress.lang.object>, KW_INPUT character, KW_INPUT object<? extends progress.lang.parameterlist>)
  Progress.Lang.Error.getmessage(KW_INPUT integer)
  Progress.Lang.Error.getmessagenum(KW_INPUT integer)
  Progress.Lang.Class.getproperty(KW_INPUT character)
  Progress.Reflect.Variable.set(KW_INPUT object<? extends progress.lang.object>, KW_INPUT Object)
  Progress.Lang.Class.getconstructor(KW_INPUT object<? extends progress.lang.parameterlist>)
  
  progress.lang.object
  progress.lang.class
  progress.lang.error
  progress.lang.enum
  progress.lang.stop
  progress.reflect.property
  progress.lang.oerequestinfo
  progress.reflect.constructor
  
  

• serialization, use-widget-pool, final, abstract classes
• undo, throw ; on ... undo, throw
• put stream-handle - but this already works.
• web stream

Please let me know if you want to upload the ADE project I used.

Please let me know if you want to upload the ADE project I used.

Yes, we should have it located in ~/secure/code/p2j_repo/ade/.

OK, is there. I've included the abl/ADE checkout, but couldn't include the .git folder, too. The commit message is:

Initial version. Contains setup for parsing and running reports of ADE release 12.8.4 from Sep. 9 2024 (hash 2e350f0a48a6468b0a89254a1e21c0a98d6b2604).

• setup p2j/ and skeleton/
• remove skeleton/oo4gl/OpenEdge
• run ant clean convert.front.list rpt-no-front

Please checkout again the project fully, there was some references from the source project I used to build it, which I didn't clean up properly, and I've commited rev 1 again.

I've tried parsing the web-disp.p program from ADE and this has hard-coded include-file references like src/web/method/webutils.i - I had to rename the abl/ADE folder to abl/src and adjust p2j.cfg.xml and file-cvt-list.txt accordingly.

The .cls files don't show any include file issue. Should I commit this change? We will need to create a profile/conversion list of files to be converted for webspeed, and that may show other issues (like propath).

Constantin Asofiei wrote:

I've tried parsing the web-disp.p program from ADE and this has hard-coded include-file references like src/web/method/webutils.i - I had to rename the abl/ADE folder to abl/src and adjust p2j.cfg.xml and file-cvt-list.txt accordingly.

Yes, these references are deep inside the 4GL code. I don't know why PSC messed with the structure of the zip in this version.

If you saw my recent wiki page using this zip, it had these instructions:

unzip ~/shared/projects/posse/ade_tools_apache_license_12_8_4.zip -d abl/ade/
mv abl/ade/ADE-release-12.8.x/ abl/ade/src/

The .cls files don't show any include file issue. Should I commit this change? We will need to create a profile/conversion list of files to be converted for webspeed, and that may show other issues (like propath).

Yes, we should build the proper unzip/rename into the bzr repo.

We should separate the ADE source out into its own repo (as is being discussed in #9516).

Why? Because:

• We will be using it in many projects even though some of the code is included in FWD.
• We will potentially need to process later versions of the ADE code (12.8.5? 12.8.6?).

Greg Shah wrote:

We should separate the ADE source out into its own repo (as is being discussed in #9516).

Hm... thanks, I understand now, we need a new project.

Please point me to the wiki page, I can't find it.

Constantin Asofiei wrote:

Please point me to the wiki page, I can't find it.

Nevermind, found it.

Marian, please let me know if you started any work on this.

Constantin Asofiei wrote:

Marian, please let me know if you started any work on this.

Constantin, sorry I didn't even saw this one was assigned to me so no didn't started anything on it... I was under the impression you started to work on it converting all ADE code and ran some code analysis on it.

Normally we can try to convert and see how the converted code works agains existing test cases, problem is the test cases were written against 12.2(.0) version not 12.8(.4) and things surely have changed even if only slightly. The only thing I could think from the top of my head that it makes sense to keep - maybe just as an alternative solution - is the Apache based HttpClient instead of using the socket based ADE implementation.

What we can do is to run all OO tests we have against 12.8.4 ADE, adapt those as necessary and then attempt the conversion and see what needs to be done on the FWD side - some stuff we had will simply have to move in the 'Progress' package, like collections that will still need to be wrappers on top of corresponding Java implementation.

Marian Edu wrote:

What we can do is to run all OO tests we have against 12.8.4 ADE, adapt those as necessary

Please run the existing unit tests in OE, with 12.8.4 ADE and first lets document any differences.

and then attempt the conversion and see what needs to be done on the FWD side

We'll work on the ADE conversion part.

- some stuff we had will simply have to move in the 'Progress' package

We may need help with the 'Progress' package, I'll let you know.

This is a comment I meant to post yesterday:

For the ADE support, first we need to focus on these classes:

abl/src/corelib/OpenEdge/Core/Assert.cls
abl/src/corelib/OpenEdge/Core/ByteBucket.cls
abl/src/corelib/OpenEdge/Core/Collections/Collection.cls
abl/src/corelib/OpenEdge/Core/Collections/ICollection.cls
abl/src/corelib/OpenEdge/Core/Collections/IIterable.cls
abl/src/corelib/OpenEdge/Core/Collections/IIterator.cls
abl/src/corelib/OpenEdge/Core/Collections/IList.cls
abl/src/corelib/OpenEdge/Core/Collections/IListIterator.cls
abl/src/corelib/OpenEdge/Core/Collections/ISet.cls
abl/src/corelib/OpenEdge/Core/Collections/Iterator.cls
abl/src/corelib/OpenEdge/Core/Collections/List.cls
abl/src/corelib/OpenEdge/Core/Collections/ListIterator.cls
abl/src/corelib/OpenEdge/Core/Collections/Map.cls
abl/src/corelib/OpenEdge/Core/Collections/ObjectStack.cls
abl/src/corelib/OpenEdge/Core/Integer.cls
abl/src/corelib/OpenEdge/Core/IOModeEnum.cls
abl/src/corelib/OpenEdge/Core/Memptr.cls
abl/src/corelib/OpenEdge/Core/String.cls
abl/src/corelib/OpenEdge/Core/System/ApplicationError.cls
abl/src/corelib/OpenEdge/Core/System/InvalidValueSpecifiedError.cls
abl/src/corelib/OpenEdge/Core/System/NotFoundError.cls
abl/src/corelib/OpenEdge/Core/System/UnsupportedOperationError.cls
abl/src/corelib/OpenEdge/Core/TimeStamp.cls
abl/src/corelib/OpenEdge/Logging/LogEvent.cls
abl/src/corelib/OpenEdge/Logging/LogMessage.cls

Parsing shows problems as noted in #9488-9. From the list of the above classes, please start with the ones which do not yet exist in p2j.oo (and all their dependencies), and also see the #9436-7 for the support. From the front phase, there are lots of dependencies, please determine the full set which needs to be converted.

Attached are the converted skeletons (not ADE code) for the above and #9436 ones (I'm using this to go through parsing issues for the #9457 app). The FWD changes I used to generate these and the configured ADE project to convert the same files from ADE is on devsrv01:/tmp/ade_project.zip. This includes a p2j/ project of 9198a branch with changes which may or may not end up in trunk. For now, there is no branch for these changes. Please use a trunk branch to work on parsing/conversion issues, and included the p2j/ changes from ade_project.zip as necessary.

Also, if you encounter non-ADE builtin 4GL classes (which appear or not in skeleton/ - look for Builtin class ... is not implemented in FWD! in the parse log) please post here and we'll decide if they are needed or not for the first phase.

• determine the minimum list of .cls files from ADE which need to be converted and are not yet implemented in FWD, using as a start the list above.
• this will require to determine dependencies from the hierarchy or usage from ADE code; so the list above will expand
• any progress. package dependencies which are not yet implemented will need to be worked in #9436.
  • Determine any .cls file which does not exist in the skeleton/ project and please work with Octavian to get the skeleton .cls
  • once you have these, you can use FWD conversion of the new skeleton .cls files to get the 'template' of the Java code which needs to be implemented by us
  • check 6410b for any already implemented progress. package classes

When converting both ADE and standalone skeleton/ .cls files, this is required to have the proper Java package emitted; in com.goldencode.p2j.convert.NameConverter, method convertPackage, the return can be changed to this:

      return (false && builtin ? "com.goldencode.p2j.oo." : "") + sb.toString();

We should think of a better approach for this, but for now it works.

• @LegacyResourceSupport(supportLvl = CVT_LVL_FULL|RT_LVL_FULL) for ADE stuff
• @LegacyResourceSupport(supportLvl = CVT_LVL_FULL|RT_LVL_STUBS) for cases when skeleton/ .cls files are converted (i.e. Progress. package).

• not emit @LegacyResourceSupport - if the config param is missing
• emit whatever string is set at the param (i.e. CVT_LVL_FULL|RT_LVL_FULL, CVT_LVL_FULL|RT_LVL_STUBS or whatever combination).

Another config param is needed to emit or not LegacyResource annotation - the change is already in the ADE/p2j/ project, but this emits it always.

From testcases/oo/skeleton/12.8 I copied the following classes into ADE skeleton:

Progress.Collections.IList
Progress.Lang.Stop
Progress.Collections.IComparable
Progress.Collections.IHashable
Progress.Collections.HashMap
Progress.Collections.KeyValuePair
Progress.Collections.List
Progress.Collections.ISet
Progress.Collections.SortedSet
Progress.Collections.ICollection
Progress.Collections.IIterable
Progress.Collections.IMap
Progress.Collections.IEqualityComparer
Progress.Collections.IIterator
Progress.Collections.IComparer

com.goldencode.p2j.oo.collections.Iiterable
com.goldencode.p2j.oo.collections.Iiterator 
com.goldencode.p2j.oo.collections.Icollection 
com.goldencode.p2j.oo.collections.Ilist 
com.goldencode.p2j.oo.lang.Stop 
com.goldencode.p2j.oo.collections.Icomparable 
com.goldencode.p2j.oo.collections.Ihashable 
com.goldencode.p2j.oo.collections.Imap 
com.goldencode.p2j.oo.collections.Iset 
com.goldencode.p2j.oo.collections.HashMap 
com.goldencode.p2j.oo.collections.IequalityComparer
com.goldencode.p2j.oo.collections.KeyValuePair 
com.goldencode.p2j.oo.collections.List 
com.goldencode.p2j.oo.collections.SortedSet 
com.goldencode.p2j.oo.collections.Icomparer 

Did you check 6410b if they are there?

Constantin Asofiei wrote:

Marian Edu wrote:

What we can do is to run all OO tests we have against 12.8.4 ADE, adapt those as necessary

Please run the existing unit tests in OE, with 12.8.4 ADE and first lets document any differences.

There aren't that many differences, mainly changes in error messages/numbers.

• tests.oo.json.array.TestSetGetRaw (diff error number)
• tests.oo.openedge.net.server_connection.client_socket.TestMethodStartTls (diff error msg)
• tests.oo.progress.lang.class.TestMethodNew (diff error number)
• tests.oo.progress.lang.parameter_list.TestMethodSetParameterExtent (diff error number)
• tests.oo.progress.lang.parameter_list.TestMethodSetParameterParamErrors (diff error number)
• tests.oo.progress.reflect.variable.TestPropertyExtent (extent property returns different values for variable length extents with default value set)

  Beside this we already changed the code because of some compile time error - numeric variables/properties used to allow string values as extent initial values, this is a syntax error now so we had to update the code.

Constantin Asofiei wrote:

Did you check 6410b if they are there?

At the moment 6410b does not compile, I am not sure about the state of this branch.

From the list above, this is the support level in 6410b:

From the list above, HashMap is in ADE at corelib/OpenEdge/Core/Collections/HashMap.cls, so this need to be added to the classes we convert, not implement.

I think I made an error in #9488-28 and #9488-29. We need the classes in com.goldencode.p2j.oo.collections.* (Progress.collections) not the ones in com.goldencode.p2j.oo.core.collections.* (OpenEdge.core.collections). The classes needed mentioned in #9488-25, all need their skeleton converted and implemented in FWD, they are not in 6410b.

• rev. 15716: Added support for generating LegacyResource and LegacyResourceSupport annotations and correcting the package name for converting legacy oo package.
• rev. 15717: A change from Constantin's original branch needed for the conversion to successfully complete.

I will now convert all the needed skeletons. I understand I should add them to p2j oo package and commit them on another branch. What should that branch be? 9488b?

Ioana-Cristina Prioteasa wrote:

I understand I should add them to p2j oo package and commit them on another branch. What should that branch be? 9488b?

No, use 9436a and lets discuss them there. The plan is to commit the skeletons first and after that work on implementations in 9436b. Please add copyright/history/etc to all converted files.

OK, this is a 'chicken and the egg' problem I think. Get the .java converted skeletons as close as possible to 'good' and do manually adjustments where 'n/a' appears; also check if anything has been dropped - i.e. all methods/properties/vars are there, inherits/implements clauses, etc.

I will do that for the classes that are converted, but I am not exactly sure why the rest are not ending in java classes.

Ioana-Cristina Prioteasa wrote:

... but I am not exactly sure why the rest are not ending in java classes.

Please post the full log.

Did you add all files to the conversion list? For example Iiterator.cls is not picked up in Scanning Progress Source. Something is wrong with the config used as input.

I have a file-cvt-list.txt with the above contents and i run ant clean convert.list

• running ant convert:
  

  Buildfile: /home/icp/projects/ADE/build.xml
  
  init:
  
  check-file-cvt-list:
  
  check-file-ignore-list:
  
  convert:
  
  BUILD SUCCESSFUL
  Total time: 0 seconds
  

• running ant deploy.all - i still get the same files in the java classes.

It is weird because it seems they are parsed, these are the contents of by abl folder:

✨icp@icp:~/projects/ADE/abl$ tree -a
.
├── fwd-embedded-driver.p
└── Progress
    ├── Collections
    │   ├── HashMap.cls
    │   ├── HashMap.cls.ast
    │   ├── HashMap.cls.cache
    │   ├── HashMap.cls.jast
    │   ├── HashMap.cls.lexer
    │   ├── HashMap.cls.parser
    │   ├── HashMap.cls.pphints
    │   ├── ICollection.cls
    │   ├── ICollection.cls.ast
    │   ├── ICollection.cls.cache
    │   ├── ICollection.cls.lexer
    │   ├── ICollection.cls.parser
    │   ├── ICollection.cls.pphints
    │   ├── IComparable.cls
    │   ├── IComparable.cls.ast
    │   ├── IComparable.cls.cache
    │   ├── IComparable.cls.jast
    │   ├── IComparable.cls.lexer
    │   ├── IComparable.cls.parser
    │   ├── IComparable.cls.pphints
    │   ├── IComparer.cls
    │   ├── IComparer.cls.ast
    │   ├── IComparer.cls.cache
    │   ├── IComparer.cls.jast
    │   ├── IComparer.cls.lexer
    │   ├── IComparer.cls.parser
    │   ├── IComparer.cls.pphints
    │   ├── IEqualityComparer.cls
    │   ├── IEqualityComparer.cls.ast
    │   ├── IEqualityComparer.cls.cache
    │   ├── IEqualityComparer.cls.jast
    │   ├── IEqualityComparer.cls.lexer
    │   ├── IEqualityComparer.cls.parser
    │   ├── IEqualityComparer.cls.pphints
    │   ├── IHashable.cls
    │   ├── IHashable.cls.ast
    │   ├── IHashable.cls.cache
    │   ├── IHashable.cls.jast
    │   ├── IHashable.cls.lexer
    │   ├── IHashable.cls.parser
    │   ├── IHashable.cls.pphints
    │   ├── IIterable.cls
    │   ├── IIterable.cls.ast
    │   ├── IIterable.cls.cache
    │   ├── IIterable.cls.lexer
    │   ├── IIterable.cls.parser
    │   ├── IIterable.cls.pphints
    │   ├── IIterator.cls
    │   ├── IIterator.cls.ast
    │   ├── IIterator.cls.cache
    │   ├── IIterator.cls.lexer
    │   ├── IIterator.cls.parser
    │   ├── IIterator.cls.pphints
    │   ├── IList.cls
    │   ├── IList.cls.ast
    │   ├── IList.cls.cache
    │   ├── IList.cls.lexer
    │   ├── IList.cls.parser
    │   ├── IList.cls.pphints
    │   ├── IMap.cls
    │   ├── IMap.cls.ast
    │   ├── IMap.cls.cache
    │   ├── IMap.cls.lexer
    │   ├── IMap.cls.parser
    │   ├── IMap.cls.pphints
    │   ├── ISet.cls
    │   ├── ISet.cls.ast
    │   ├── ISet.cls.cache
    │   ├── ISet.cls.lexer
    │   ├── ISet.cls.parser
    │   ├── ISet.cls.pphints
    │   ├── KeyValuePair.cls
    │   ├── KeyValuePair.cls.ast
    │   ├── KeyValuePair.cls.cache
    │   ├── KeyValuePair.cls.jast
    │   ├── KeyValuePair.cls.lexer
    │   ├── KeyValuePair.cls.parser
    │   ├── KeyValuePair.cls.pphints
    │   ├── List.cls
    │   ├── List.cls.ast
    │   ├── List.cls.cache
    │   ├── List.cls.jast
    │   ├── List.cls.lexer
    │   ├── List.cls.parser
    │   ├── List.cls.pphints
    │   ├── SortedSet.cls
    │   ├── SortedSet.cls.ast
    │   ├── SortedSet.cls.cache
    │   ├── SortedSet.cls.jast
    │   ├── SortedSet.cls.lexer
    │   ├── SortedSet.cls.parser
    │   └── SortedSet.cls.pphints
    └── Lang
        ├── Stop.cls
        ├── Stop.cls.ast
        ├── Stop.cls.cache
        ├── Stop.cls.jast
        ├── Stop.cls.lexer
        ├── Stop.cls.parser
        └── Stop.cls.pphints

4 directories, 100 files

I am trying to add the LegacyResourceSupport at class level, right after the LegacyResource.
In /rules/convert/base_structure.xml:

            <action>isLegacyOo = getConfigParameter("is-legacy-oo")</action>
            <rule>isLegacyOo.equalsIgnoreCase("true")
               <action>annoref = java.createJavaAst(java.annotation, "LegacyResource", jclsdef.parent, jclsdef.indexPos)</action>
               <action>annoref = createJavaAst(java.assign, "=", annoref)</action>
               <action>createJavaAst(java.reference, "resource", annoref)</action>
               <action>createJavaAst(java.string, getNoteString("qualified_oo_name"), annoref)</action>
            </rule>

            <action>
               legacyResSupportType = getConfigParameter("legacy-res-support-type")
            </action>
            <rule> legacyResSupportType != null
               <action>
                  annoref = java.createJavaAst(java.annotation, "LegacyResourceSupport", jclsdef.parent, jclsdef.indexPos)
               </action>
               <action>hasLRSAnn = true</action>
               <action>
                  execLib("add_annotation_setting", "supportLvl", java.reference, legacyResSupportType, annoref)
               </action>
            </rule>
      </rule>

     [java] execLib("add_annotation_setting", "supportLvl", java.reference, legacyResSupportType, annoref)
     [java] ^  { Cannot set value for variable 'lastid' (java.lang.Long) to type com.goldencode.p2j.uast.JavaAst }

You need to use annoref.id as last argument.

• rev. 15719: Added lang.Stop, collections.Ihashable, collections.Icomparable, collections.Icomparer, collections.IequalityComparer.
• rev. 15718: Conversion changes for adding LegacyResourceSupport at the class level.

In 9488a rev. 15720 I also added collection.KeyValuePair.
Next I have HashMap, List and SortedSet to correct.

As disscused in the office today, the skeletons needed for #8904 (Progress.Lang.UserInterrupt, Progress.Lang.Stop, Progress.Lang.StopError, Progress.Lang.LockConflict, Progress.Lang.StopAfter) will also be converted and added in 9488a. After a review these can be merged and Eduard would pick them up in a rebase.

• Stop is final, but UserInterrupt, LockConflict and StopAfter extend it. This can be easily fixed by Stop not being final, but in the skeleton it is and in the documentation: Progress.Lang.Stop is also the immediate super class for the following built-in objects ..(mentioned above). You cannot inherit from this class.

• I get a conversion mismatch in the mentioned classes that extend Stop.
  The constructor of Stop is called like:
  

  __lang_Stop_constructor__();
  

  
  instead of:
  

  __lang_stop__constructor__();
  

Ioana, sorry I've not noticed sooner - the progress.* Java stubs for the skeletons need to be in 9436a, so please move them there. I just want to do some tests with them. 9488a has some other changes, we need to discuss those.

Otherwise, if you have changes in the skeleton bzr project, please make sure they are committed.

• Progress.Data.ChildrenAllowEdit
• Progress.Data.ChildrenAllowNew
• Progress.Data.ChildrenAllowRemove
• Progress.Data.ChildrenInputValue
• Progress.Data.InputValue
• Progress.Data.CancelCreateRowEventArgs
• Progress.Data.CreateRowEventArgs
• Progress.Data.SortRequestEventArgs
• System.ComponentModel.ComponentResourceManager

I found all of them, except the last System.ComponentModel.ComponentResourceManager, this is not in skeleton/dotnet/System/ComponentModel/ and in testcases/oo there is nothing regarding dotnet.
Marian, do you know if this skeleton exists of how can I generate it?

Sorry, I meant to post the above on #9436.

I'm encountering recurring issues while converting ADE, so I'm currently focusing on converting a single class: OpenEdge/Core/TimeStamp.cls

The conversion fails due to a parsing error related to generics:

pre_scan_class failure with text '<', error 'unexpected token: <' at line 26, column 23 in file ./abl/OpenEdge/Core/TimeStamp.cls

This happens because the original code includes:

using Progress.Collections.IComparable.
using Progress.Collections.IHashable.
class OpenEdge.Core.TimeStamp
serializable
implements IComparable<TimeStamp>
         , IHashable

As an initial fix, I plan to ignore the < > syntax when parsing implements clauses. The error occurs in ProgressParser, which is a generated file, so I'm currently investigating progress.g to adjust the grammar accordingly.

Committed to 9488a rev 15724: Ignore generics in symbol-based class names.
With this change the conversion passes the previous error but other issues are found:

     [java] ./abl/OpenEdge/Core/TimeStamp.cls:256:27: unexpected token: *
     [java]     at com.goldencode.p2j.uast.ProgressParser.assign(ProgressParser.java:39897)
     [java]     at com.goldencode.p2j.uast.ProgressParser.assign_stmt(ProgressParser.java:28477)
     [java]     at com.goldencode.p2j.uast.ProgressParser.stmt_list(ProgressParser.java:27480)

assign tz *= -1.

Committed to 9488a rev. 15725: Added support for compound assignment operators (*=, +=, -=, /=).
The next parsing error to investigate:

     [java] ./abl/OpenEdge/Core/TimeStamp.cls:478:16: unexpected token: hash-code
     [java]     at com.goldencode.p2j.uast.ProgressParser.lvalue(ProgressParser.java:18638)
     [java]     at com.goldencode.p2j.uast.ProgressParser.primary_expr(ProgressParser.java:59609)
     [java]     at com.goldencode.p2j.uast.ProgressParser.chained_object_members(ProgressParser.java:23377)
     [java]     at com.goldencode.p2j.uast.ProgressParser.un_type(ProgressParser.java:59311)
     [java]     at com.goldencode.p2j.uast.ProgressParser.prod_expr(ProgressParser.java:59176)

This is coming from this line:

return hash-code(this-object:GetClass():TypeName, this-object:Format, mtDate, miTime, miTZOffset).

I'm having some trouble with the hash-code() function.
I modified progress.g and now the token is found, the new error is:

{ ## Error:: Could not process function 'hash-code', not found in super or local definitions (hmfuncs/reference)! [BLOCK id <12884901889> 0:0] }

I tried to create a testcase but my it seems these are not available in my VM, they must be from a newer version.

I think the problem is that it is searching for it in local or super definitions when it should be seen as a built in function. Tried mapping it to a new annotated BaseDataType.hashCode but I must be missing something.

hash-code is a new builtin function, see https://docs.progress.com/bundle/abl-reference/page/HASH-CODE-function.html . You need to treat it as any other builtin function like ENCODE.

Committed to 9488a rev. 15726: Added conversion support and basic runtime support for the new hash-code builtin function.

• the n/a issue, the output is:
  

  public class TimeStamp
  extends BaseObject
  implements Icomparable,
             n/a
  

  
  for
  

  using Progress.Collections.IComparable.
  using Progress.Collections.IHashable.
  
  class OpenEdge.Core.TimeStamp
  serializable
  implements IComparable<TimeStamp>
           , IHashable
  

  
  Not exactly sure why because IHashable exists in 9488a, like IComparable.
  n/a also appears in other places:
  

  import n/a;
  storeReturnValue(DateOps.interval(thisDate, n/a.oo_references.rules_FIXME4(), "seconds"));
  storeReturnValue(DateOps.interval(thisDate, n/a.oo_references.rules_FIXME4(), "milliseconds"));
  n/a.oo_references.rules_FIXME6(pOther, new character("CompareTo object"));
  

• assign tz *= -1. convert to:
  

  tz;
  -1;
  

  
  It should be tz.assign(multiply(tz, new integer(-1)));

• Implementing Icomparable forces the compareTo method to have this signature
  public integer compareTo(final object< ? extends com.goldencode.p2j.oo.lang._BaseObject_ > _otherObject);
  but it is converted as:
  public integer compareTo(final object< ? extends TimeStamp > _pOther)
  I think Icomparable needs generics and more conversion changes are required to emit implements Icomparable<Timestamp>

Committed to 9488a rev 15727: Full conversion support for +=, -=, *=, /=.
Now assign tz *= -1. converts to tz.assign(multiply(tz, new integer(-1)));

• Conversion support for using generics in implements clause.
• Modified Icomparable to use generics, this change will need to be ported in 9436a.

We may need more changes for other places that could use generics, for now this is sufficient to correctly convert OpenEdge.Core.TimeStamp.
Now I will move to supporting the new variable definition statements: VAR INT n.

Ioana, I assume you want to create the equivalent DEFINE VAR n AS INT instead of VAR INT n, directly in the parser. Greg, I think this would be the 'right' approach, either in the parse or post-parse-fixups.

If we do it in the parser, we can create ASTs with constructs like this (see the assignment rule in the parser)

assignment! 
  :
      stmt_term
      {
         if (lstmt)
         {
            ## = #([STATEMENT, "statement"], e, k);
         }
         else
         {
            ## = #([ASSIGNMENT, "assignment"], e, k);
         }
         skipExpression = false;
         matchAssign = false;
      }

There is this doc: https://bioinformatics.ai.sri.com/biowarehouse/repos/enzyme-loader/docs/antlr_2.7.1_ref_man.pdf Look for #([. In the assignment rule, ! turns of standard tree construction, and ## is the returned AST for this rule. #() creates another AST with the wanted structure.

VAR CHAR s1, s2, s3.
VAR CHAR[10] s4.
VAR CHAR s6 = "My character string".

• final keyword use, I am not sure about this, cannot find Progress documentation for it. For now i removed it.
  

      define final non-serializable public property Size as integer no-undo
          get():
              return mBackingList:Count.
          end get.
  

• generics used, here we need some conversion modifications.
  

      constructor public List():
          this-object(new Progress.Collections.List<Object>()).
      end constructor.
  

I will try now to support the generic in the new call.

Ioana, please change progress.g to allow FINAL option at the property rule, and the conversion rules to emit it. For generics - that's lets say ... complicated again.

• a quick fix for a NPE when the p2j.cfg.xml is-legacy-oo configuration is not present.
• parser and conversion support for emitting a final property.

Note: added in rev. 15729 my changes related to the VAR statement and reverted them in rev. 15730.

Reproduced the extent conversion error with this testcase:

def var myList as Progress.Collections.IList.
myList = new Progress.Collections.List().

def var arr as integer extent.
assign extent(arr) = myList:Count.

Reproduced the error with an even simpler testcase. This works in Progress.

def var arr as integer extent.
assign extent(arr) = 10.

Ioana-Cristina Prioteasa wrote:

Reproduced the error with an even simpler testcase. This works in Progress.
[...]

Did you test with trunk?

With trunk it works, working on pinpointing what change on 9488a introduced this regression.

Committed to bzr skeleton project, rev 123: Added the following skeletons from Progress.Collections: IList, IComparable, IHashable, ISet, ICollection, IMap, IIterable, IIterator, IComparer, IEqualityComparer, List, HashMap, KeyValuePair, SortedSet.
These are only the skeletons I needed to far for the ADE conversion, ported from testcases/oo/skeleton/12.8/Progress/Collections.

The regression was caused by the changes for the *= operators, found the issue and committed the solution in 9488a rev. 15732.
Now the warnings still appear but the List conversion completes. It has a lot of issues with n/a but those are fixed when the needed classes are converted or set up for conversion.
Next I will investigate the warnings, will post what I find. I also want to add to the conversion list everything again to see if there are any other issues in other classes. For now I think I will modify where generics are used.

• The first warning:
  

       [java] annotateMethodCall FAILURE: Join from class OpenEdge.Core.String, access KW_PRIVATE, static true, at Join [OO_METH_CLASS]:<id_unavailable> @369:37
       [java]    cast [FUNC_CLASS]:<id_unavailable> @369:42
       [java]       : [OBJECT_INVOCATION]:<id_unavailable> @369:54
       [java]          poValue [VAR_CLASS]:<id_unavailable> @369:47
       [java]          ToArray [OO_METH_CLASS]:<id_unavailable> @369:55
       [java]       OpenEdge.Core.String [CLASS_NAME]:<id_unavailable> @369:66
       [java]    pcDelimiter [VAR_CHAR]:<id_unavailable> @369:89
  

  
  Seems to be emitted in this method:
  

  method public override logical Equals(o as Object):
          define variable list as OpenEdge.Core.Collections.IList no-undo.
          def var iter as OpenEdge.Core.Collections.IIterator.
          def var idx as integer.
          def var element as Object extent 2.
  
          if super:Equals(o) then
              return true.
  
          if type-of(o, OpenEdge.Core.Collections.IList) then
          do:
              assign list = cast(o, OpenEdge.Core.Collections.IList).
              if not list:Size eq this-object:Size then
                  return false.
  
              assign iter = list:Iterator().
              do while iter:HasNext():
                  assign idx       += 1
                         element[1] = mBackingList:Get(idx)
                         element[2] = iter:Next()
                         .
                  // if one item is not valid, and the other is, they cannot be the same
                  if valid-object(element[1])
                  and not valid-object(element[2])
                  then
                      return false.
  
                  if not valid-object(element[1])
                  and valid-object(element[2])
                  then
                      return false.
  
                  // if both items are not valid, then they are the same
                  // if they are both valid, check for equality
                  if valid-object(element[1])
                  and valid-object(element[2])
                  and not element[1]:Equals(element[2])
                  then
                      return false.
              end.
  
              return true.
          end.
  
          return false.
      end method.
  

  
  Specifically here:
  

  if super:Equals(o) then
              return true.
  

  
  Not exactly sure how Join is resolved as the method name.

• The second warning:
  

  annotateMethodCall FAILURE: listiterator from class OpenEdge.Core.Collections.ListIterator, access KW_PUBLIC, static false, at new [FUNC_CLASS]:<id_unavailable> @435:16
       [java]    OpenEdge.Core.Collections.ListIterator [CLASS_NAME]:<id_unavailable> @435:20
       [java]    ( [LPARENS]:<id_unavailable> @435:58
       [java]       parameter [PARAMETER]:<id_unavailable> @0:0
       [java]          mBackingList [VAR_CLASS]:<id_unavailable> @435:59
  

  
  Emits in multiple places:
  

      method public OpenEdge.Core.Collections.IIterator Iterator():
          return new OpenEdge.Core.Collections.ListIterator(mBackingList).
      end method.
  

  
  

      method public OpenEdge.Core.Collections.IListIterator ListIterator():
          return new OpenEdge.Core.Collections.ListIterator(mBackingList).
      end method.
  

  
  

      method public OpenEdge.Core.Collections.IListIterator ListIterator(pStartIdx as integer):
          Assert:IsPositive(pStartIdx, 'List index').
          if pStartIdx gt mBackingList:Count then
              return error new AssertionFailedError(substitute('Index &1 is larger than List size &2',
                                                              pStartIdx, mBackingList:Count)).
  
          return new OpenEdge.Core.Collections.ListIterator(mBackingList, pStartIdx).
      end method.
  

Continue with conversion and I had to add 2 more skeletons: Progress.IO.MemoryOutputStream, Progress.IO.ByteOrder. These need to be converted and committed to 9436a.
After adding the skeletons, for OpenEdge/Core/ByteBucket.cls:

./abl/OpenEdge/Core/ByteBucket.cls:336:39: unexpected token: GetByte
     [java]     at com.goldencode.p2j.uast.ProgressParser.primary_expr(ProgressParser.java:59724)

assign iByte                = GetByte(this-object:Position)

Context:

    method public integer GetByte ():
        define variable iByte as integer no-undo.

        assign iByte                = GetByte(this-object:Position)
               this-object:Position = this-object:Position + 1.
        return iByte.
    end method.

    method public integer GetByte (input piStartPos as int64):
        if piStartPos  <=  memoutStream:BytesWritten then do:
            return get-byte(memoutStream:Data,piStartPos).
        end.

        return 0.
    end method.

Modifying from this

assign iByte = GetByte(this-object:Position)

assign iByte = this-object:GetByte(this-object:Position)

Another issue discovered in OpenEdge/Core/Collections/StringKeyedMap.cls, has to do with generics and I think this one will be difficult to solve:

//var Progress.Collections.IIterator<KeyValuePair<ILongcharHolder, Object>> iter.
def var iter as Progress.Collections.IIterator.
def var hValue as Object.
assign hValue = iter:Current:Value.

The commented line is the original definition, I removed the generic but this way Value is unknown to the parser because it is a property from KeyValuePair:

./abl/OpenEdge/Core/Collections/StringKeyedMap.cls:225:46: unexpected token: Value

Report on remaining issues for ADE conversion¶

I added all files listed in #9488-22 to the conversion list using the latest 9488a, along with the two new required skeletons: Progress.IO.MemoryOutputStream and Progress.IO.ByteOrder. I resolved several issues by modifying the original code and compiled a comprehensive list of remaining items that need to be fixed or discussed.

VAR Statement¶

The new VAR statement is widely used in these classes with various constructs, including:

VAR INT a.
VAR INT a = 1.
VAR INT a,b.
VAR INT a, b =1.

Generics¶

Below are some use cases I encountered:

def private serializable var mBackingList as Progress.Collections.IList<Object>.
var private HashMap<ILongcharHolder, Object> mBackingHashMap.
this-object(new Progress.Collections.List<Object>()).
constructor private List (input pBackingList as Progress.Collections.IList<Object>):

var Progress.Collections.IIterator<KeyValuePair<ILongcharHolder, Object>> iter.
assign hValue = iter:Current:Value.

Overloaded class function not recognized¶

In OpenEdge/Core/ByteBucket.cls, the following construct did not work: assign iByte = GetByte(this-object:Position) However, this worked correctly: assign iByte = this-object:GetByte(this-object:Position). I applied this solution for now.

tenant-name and tenant-id built in functions not recognized by parser.¶

After discussing with Radu, I confirmed that these are added in progress.g, but further investigation is needed.
A test case fails with both trunk and 9488a with unexpected token: tenant-id and unexpected token: tenant-name, even though the code works in 4GL:

define variable client as handle no-undo.
define variable tokenValue as character no-undo.
define variable domainAccessCode as character no-undo initial "MyAccessCode".

create client-principal client.

client:user-id = "MyUser".
client:domain-name = "MyDomain".
client:session-id = "MySessionID".

client:seal(domainAccessCode).

assign tokenValue = string(client:tenant-id).
message "Tenant ID:" tokenValue view-as alert-box.

assign tokenValue = client:tenant-name.
message "Tenant Name:" tokenValue view-as alert-box.

delete object client no-error.

Issue with the Join function from OpenEdge.Core.String¶

The Join function also appears in the parser warnings posted in #9488-71. During the conversion of this class, it leads to the following error:

clsdef = loadConvertedClass(fname, false, this)
     [java]          ^  { Can not find class for accessing Join [OO_METH_CLASS] @369:37 (266287974500) }

at this line

return OpenEdge.Core.String:Join(cast(poValue:ToArray(), OpenEdge.Core.String), pcDelimiter).

A test case using a similar construct converts correctly using the LegacyString implementation, and I have not yet managed to reproduce the issue:

define variable myArray as character extent 2 no-undo.
define variable pcDelimiter as character no-undo initial ",".
define variable joinedString as character no-undo.

myArray[1] = "Item1".
myArray[2] = "Item2".

joinedString = OpenEdge.Core.String:Join(myArray, pcDelimiter).

message "Joined String:" joinedString view-as alert-box.

Parser warnings:¶

• A lot of Could not find legacy builtin method javaname for
• WARNING: more than one method def found using fuzzy lookup: GetIterator from class Progress.Collections.IList
• annotateMethodCall FAILURE: listiterator from class OpenEdge.Core.Collections.ListIterator, access KW_PUBLIC, static false, at new [FUNC_CLASS]:<id_unavailable> @435:16

Conversion warnings:¶

• Null annotation (legacy-import) for using [KW_USING] @17:1 (12884901914) - a lot of these for annotations: legacy-import, full-java-class, simple-java-class, containing-package, found-in-full-java-class
• MISSING_REFID for property Size [VAR_INT]:12884923722 @2355:22
• WARNING: non-local property javaname (get-javaname) lookup failure!
• WARNING: removing unsuported legacy annotation for since and depricated
• WARNING: no type calculated for BUFFER-VALUE in [261993005876] buffer-value [ATTR_POLY]:261993005876 @99:53
• Passing unknown value literal to builtin func! Size [OO_METH_INT64]:158913791822 @302:31

Next steps¶

• Investigate tenant-name and tenant-id parsing issues in progress.g.
• Deep dive into the Join function issue in OpenEdge.Core.String.
• Investigate parser and conversion warnings, focusing on annotateMethodCall failure and method lookups.
• Discuss about generics handling strategy.

• the tenant-id() and tenant-name() support I think is only for the builtin function, and not the handle method, in FWD. Please add conversion and runtime stubs for these, at CLIENT-PRINCIPAL.
• Null annotation and MISSING_REFID are of concern - are these because of parse failures of ADE classes?
• are there .NET dependencies from ADE?

Constantin Asofiei wrote:

Ioana, some notes:

• the tenant-id() and tenant-name() support I think is only for the builtin function, and not the handle method, in FWD. Please add conversion and runtime stubs for these, at CLIENT-PRINCIPAL.

Working on this.

• Null annotation and MISSING_REFID are of concern - are these because of parse failures of ADE classes?

No.. I am in a state now where I have no parsing or conversion errors, only warnings.

• are there .NET dependencies from ADE?

No, AFAIK.

For this.null - look in the .ast, I think ByteBucket:size may be missing?

The refid annotation points to a constructor instead to the getter - look for '137438953915' in the .ast file.

<ast col="71" id="137438957382" line="642" text="Value" type="OO_METH_MEMPTR">
<annotation datatype="java.lang.String" key="property-access-type" value="getter"/>
<annotation datatype="java.lang.Long" key="access-mode" value="2130"/>
<annotation datatype="java.lang.Long" key="peerid" value="871878365155"/>
<annotation datatype="java.lang.String" key="cls-container" value="./abl/OpenEdge/Core/ByteBucket.cls"/>
<annotation datatype="java.lang.Long" key="original-property-type" value="395"/>
<annotation datatype="java.lang.Boolean" key="override" value="true"/>
<annotation datatype="java.lang.Long" key="refid" value="137438953915"/>
<annotation datatype="java.lang.Long" key="support_level" value="16400"/>
<annotation datatype="java.lang.Boolean" key="cls-property" value="true"/>
</ast>

Use -Drules.tracing=true in build.xml for the convert task, this may help to get you who sets refid.

Something looks unbalanced, may be the same root cause why the wrong field is used in #9488-78.

The refid is set in rules/annotations/early_annotations.xml:

<!-- identify all var references and convert tempidx to real using
              the dictionary, exclude widgets that can't be frame fields
              (and thus are not backed by a variable or are not cross-ref'd
              otherwise like button or browse) -->
               <!--type == prog.wid_sub_menu or--> 
         <rule>
            !(type == prog.wid_dialog   or  type == prog.wid_frame    or
              type == prog.wid_window)   and 
            (evalLib("variables")                                                      or
             (this.type == prog.class_event and parent.type == prog.object_invocation) or 
             evalLib("widget_references"))                              and
            type != prog.sys_handle                                     and
            isNote("tempidx")                                           and
            isNote("tempidx-file")                                      and
            ooVarTempIdx.containsKey(sprintf("%s-%s", 
                                             getNoteString("tempidx-file"),
                                             getNoteLong("tempidx")))
            <action>
               putNote("refid", 
                       #(java.lang.Long) ooVarTempIdx.get(sprintf("%s-%s", 
                                                                  getNoteString("tempidx-file"), 
                                                                  getNoteLong("tempidx"))))
            </action>
            <action>removeNote("tempidx")</action>
            <action>removeNote("tempidx-file")</action>
            <action>fprintf("post_parse_fixups.log",
                            "VAR_REF  %-40s: %-25s @ %05d : %05d\n",
                            file,
                            this.getSymbolicTokenType(), 
                            line,
                            this.getColumn())
            </action>
         </rule>

The dumpTree shows:

     [java] Size [VAR_INT64]:137438955304 @306:20
     [java]       (tempidx-file=./abl/OpenEdge/Core/ByteBucket.cls, found-in-source-file=./abl/OpenEdge/Core/ByteBucket.cls, found-in-cls=OpenEdge.Core.ByteBucket, tempidx=0, cls-container=./abl/OpenEdge/Core/ByteBucket.cls, override=true, cls-property=true, access-mode=2130)

               // private will match any access mode
               boolean st = var.isStatic();
               Variable old = ssd == varDict 
                                 ? cls.lookupVariableWrapper(name, KW_PRIVATE, st)
                                 : ssd == dsetDict ? cls.lookupDataSetWrapper(name, KW_PRIVATE, st)
                                                   : cls.lookupDataSourceWrapper(name, KW_PRIVATE, st);
               if (old != null)
               {
                  var.setTempIndex(old.getTempIndex());
               }

Ioana, is there a task for the new VAR statement? If not, please create one, and lets work it in a different branch. The goal would be to get the minimal syntax to allow what you posted in #9488-75 to convert and run.

Ioana, please check something (tomorrow is OK): are the associated .cls skeletons in the FWD skeleton project, for what you've converted in ADE? If yes, I'd like to get these converted (the skeletons, not the ADE conversion) and put in a branch.

Committed in 9488a rev. 15733: Added conversion and minimal runtime support for TENANT-ID and TENANT-NAME functions in ClientPrincipal.

Constantin Asofiei wrote:

Ioana, please check something (tomorrow is OK): are the associated .cls skeletons in the FWD skeleton project, for what you've converted in ADE? If yes, I'd like to get these converted (the skeletons, not the ADE conversion) and put in a branch.

I checked and it seems that the skeletons of the classes mentioned in #9488-78 are in the skeleton bzr project. I will start converting them. Where should I commit them? 9436a?

Ioana-Cristina Prioteasa wrote:

Where should I commit them? 9436a?

Yes, 9436a

From those some are already present in fwd, should I skip them?

Ioana-Cristina Prioteasa wrote:

From those some are already present in fwd, should I skip them?

Yes, add only what's missing.

I still have the error with n/a, not sure why some classes that are in p2j are not recognized. Example:

  [java] ./abl/OpenEdge/Core/Decimal.cls
     [java] Null annotation (full-java-class) for Ccs.Common.Support.IPrimitiveHolder [CLASS_NAME] @3:17 (111669149708)
     [java] Null annotation (simple-java-class) for Ccs.Common.Support.IPrimitiveHolder [CLASS_NAME] @3:17 (111669149708)
     [java] Null annotation (containing-package) for Ccs.Common.Support.IPrimitiveHolder [CLASS_NAME] @3:17 (111669149708)
     [java] Null annotation (full-java-class) for Ccs.Common.Support.IPrimitiveHolder [CLASS_NAME] @3:17 (120259084300)
     [java] Null annotation (simple-java-class) for Ccs.Common.Support.IPrimitiveHolder [CLASS_NAME] @3:17 (120259084300)
     [java] Null annotation (containing-package) for Ccs.Common.Support.IPrimitiveHolder [CLASS_NAME] @3:17 (120259084300)
     [java] Null annotation (full-java-class) for Ccs.Common.Support.IPrimitiveArrayHolder [CLASS_NAME] @3:17 (124554051596)
     [java] Null annotation (simple-java-class) for Ccs.Common.Support.IPrimitiveArrayHolder [CLASS_NAME] @3:17 (124554051596)
     [java] Null annotation (containing-package) for Ccs.Common.Support.IPrimitiveArrayHolder [CLASS_NAME] @3:17 (124554051596)

package com.goldencode.p2j.oo.common.support;

import com.goldencode.p2j.util.*;

import static com.goldencode.p2j.util.BlockManager.*;
import static com.goldencode.p2j.report.ReportConstants.*;

/**
 * Business logic (converted to Java from the 4GL source code
 * in Ccs/Common/Support/IPrimitiveHolder.cls).
 */
@LegacyResource(resource = "Ccs.Common.Support.IprimitiveHolder")
@LegacyResourceSupport(supportLvl = CVT_LVL_FULL|RT_LVL_FULL)
public interface IprimitiveHolder
extends com.goldencode.p2j.oo.lang._BaseObject_
{
}

I tried to modify the LegacyResource to resource = "Ccs.Common.Support.IPrimitiveHolder"), with capital P but the error persists. I can modify them by hand to get them committed today, there are not that many n/a, but i don't really know why this happens.

Constantin, is the LegacyResourceSupport annotation needed for types: PROPERTY, SUBSCRIBE, PUBLISH, UNSUBSCRIBE?

• OpenEdge/Core/IMemoryPointer.cls
• OpenEdge/Core/TimeStampConstants.cls
• OpenEdge/Security/Principal.cls

I got them from the testcases/oo/skeleton and have them converted. Should I add them to the skeleton project?

• abl/OpenEdge/Security/Principal.cls
  

                                                           
  block-level on error undo, throw.
  using OpenEdge.Core.Memptrr.
  
  class OpenEdge.Security.Principal :
      constructor private Principal():
      end constructor.
  
      method static public void Import(input poCP as class Memptrr):
          poCP:Value = 0.
      end method.
  end class.
  

• abl/OpenEdge/Core/Memptrr.cls
  

  block-level on error undo, throw.
  
  using Ccs.Common.Support.IMemptrHolderr.
  
  class OpenEdge.Core.Memptrr
  implements IMemptrHolderr:
      define final public property Value as memptr no-undo get.
  
      constructor protected Memptrr():
      end constructor.
  end class.
  

• abl/Ccs/Common/Support/IMemptrHolderr.cls
  

                                                       
  INTERFACE Ccs.Common.Support.IMemptrHolderr:
      DEFINE PUBLIC PROPERTY Value AS MEMPTR NO-UNDO
      GET.
  
  END INTERFACE.
  

If I delete implements IMemptrHolderr: from IMemptrHolderr, the issue is gone. But with the above, the Import function from Principal.cls converts to:

   @LegacySignature(type = Type.METHOD, name = "Import", parameters = 
   {
      @LegacyParameter(name = "poCP", type = "OBJECT", qualified = "openedge.core.memptrr", mode = "INPUT")
   })
   @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_STUB)
   public static void import_(final object<? extends Memptrr> _poCp)
   {
      ObjectOps.load(Principal.class);
      object<? extends Memptrr> poCp = TypeFactory.initInput(_poCp);

      internalProcedure(Principal.class, "Import", new Block((Body) () -> 
      {
         poCp.ref().null(new memptr(new memptr(0)));
      }));
   }

• IInterface.cls
  

  INTERFACE IInterface:
      DEFINE PUBLIC PROPERTY Value AS int NO-UNDO
      GET.
  END INTERFACE.
  

• Cls.cls
  

  block-level on error undo, throw.
  
  class Cls
  implements IInterface:
      define final public property Value as int no-undo get.
  
      constructor protected Cls():
      end constructor.
  
      method public void SetValue():
          if(this-object:Value > 0) then
              MESSAGE "".
      end method.
  end class.
  

If I delete the final keyword from the property definition in Cls, the above converts correctly.

• IInterface.cls
  

  INTERFACE IInterface:
      DEFINE PUBLIC PROPERTY v1 AS int NO-UNDO GET.
      DEFINE PUBLIC PROPERTY v2 AS int NO-UNDO GET.
  END INTERFACE.
  

• Cls.cls
  

  class Cls
  implements IInterface:
      define final public property v1 as int no-undo get.
      define public property v2 as int no-undo get.
  
      constructor protected Cls():
      end constructor.
  
      method public void verify():
          if(v1 > 0 AND v2 > 0) then
              MESSAGE "".
      end method.
  end class.
  

This converts to:

public class Cls
extends BaseObject
implements com.goldencode.dataset.Iinterface
{
   @LegacySignature(type = Type.PROPERTY, name = "v1")
   private final integer v1 = TypeFactory.integer();

   @LegacySignature(type = Type.PROPERTY, name = "v2")
   private integer v2 = TypeFactory.integer();

   @LegacySignature(type = Type.GETTER, name = "v1", returns = "INTEGER")
   public integer getV1()
   {
      return v1.duplicate();
   }

   @LegacySignature(type = Type.GETTER, name = "v2", returns = "INTEGER")
   public integer getV2()
   {
      return v2.duplicate();
   }

   @LegacySignature(type = Type.CONSTRUCTOR, name = "Cls")
   protected void __cls_constructor__()
   {
      internalProcedure(Cls.class, this, "Cls", new Block((Body) () -> 
      {
         __lang_BaseObject_constructor__();
      }));
   }

   @LegacySignature(type = Type.METHOD, name = "verify")
   public void verify()
   {
      internalProcedure(Cls.class, this, "verify", new Block((Body) () -> 
      {
         if (_and(isGreaterThan(null(), 0), () -> isGreaterThan(getV2(), 0)))
         {
            message("");
         }
      }));
   }
}

And the conversion log shows at Core Conversion Annotation step:

     [java] ./abl/Cls.cls
     [java] WARNING: non-local property javaname (get-javaname) lookup failure! v1 [OO_METH_INT]:12884901978 @10:12
     [java]       (oldtype=2986, support_level=16400, override=true, support_level-byrule=gaps/gap_analysis_marking.xml:435, refid=12884901940, access-mode=2130, cls-container=./abl/Cls.cls, refid-byrule=annotations/early_annotations.xml:136, cls-property=true, property-access-type-byrule=annotations/oo_references.rules:895, original-property-type-byrule=include/common-progress.rules:6689, property-access-type=getter, original-property-type=391)

Please post the AST for the DEFINE PROPERTY node for v1 and v2.

Please post the entire .ast file after parse, before post-parse-fixups.

Ioana, in progress.g:define_stmt_pre_scan you need to have KW_FINAL, too:

      d:KW_DEFINE^
      (
           am:access_mode
         | KW_SERIALAB     // should only apply to dataset, temp-table, var and prop
         | KW_NON_SER      // should only apply to dataset, temp-table, var and prop
         | st:KW_STATIC    // inStaticCtxt processing is not needed in pre-scan mode
         | ab:KW_ABSTRACT  // ignored for now
         | ov:KW_OVERRIDE  // ignored for now
         | KW_FINAL
      )*

Thank you! This solves the issue. I am looking now to see what other problems there are when compiling.

When generics are detected in the parser I added an annotation:

genericType = sym.loadClass(typeParams.toString());
if (genericType != null) 
{
  #sym.putAnnotation("generic-type", genericType);
}

Then in base_structure.xml:

         <rule>relativePath(this, prog.class_def, prog.kw_class, prog.kw_implemts)
            <action>createPeerAst(java.kw_implements, "implements", jclsdef.id, 0)</action>
         </rule>
         <rule>relativePath(this, prog.kw_implemts, prog.class_name)
            <action>jname = execLib("resolveClassReference", copy, null)</action>
            <action>genType = copy.getAnnotation("generic-type")</action>
            <rule>genType != null
               <action>jname = sprintf("%s%s%s%s", jname, "&lt;", genType, "&gt;")</action>
            </rule>
            <action>createPeerAst(java.class_reference, jname, closestPeerId)</action>
         </rule>

Committed to 9488a rev. 15734: Added FINAL keyword in define_stmt_pre_scan and excluded generics from implements clause.

Now the conversion completes. Remaining issues:

At parsing¶

There is still this warning:

     [java] WARNING: more than one method def found using fuzzy lookup: GetIterator from class Progress.Collections.IList
     [java] 
     [java]    CALLER GetIterator [OO_METH_CLASS]:<id_unavailable> @313:43
     [java]       (original-token=["GetIterator",<496>,line=313,col=43])
     [java] 
     [java]    PARAMETERS: 0
     [java]    MATCH 0: METRICS 0 exact; 0 cvt; 0 extExact; 0 extCvt; 0 wildcard; [I@b7a9093 ooLvl; [Ljava.lang.String;@7beba1a8 overrides; 
     [java]         KW_PUBLIC instance member GetIterator (OO_METH_CLASS - progress.collections.iiterator), extent 0, javaname getIterator, override false, abstract false in class Progress.Collections.IIterable, tempIdx 0
     [java]    PARAMETERS: 0
     [java]  data
     [java] 
     [java]    MATCH 1: METRICS 0 exact; 0 cvt; 0 extExact; 0 extCvt; 0 wildcard; [I@4f1f569a ooLvl; [Ljava.lang.String;@6332f548 overrides; 
     [java]         KW_PUBLIC instance member GetIterator (OO_METH_CLASS - progress.collections.iiterator), extent 0, javaname getIterator, override false, abstract false in class Progress.Collections.IIterable, tempIdx 0
     [java]    PARAMETERS: 0
     [java]  data

And this failure:

     [java] annotateMethodCall FAILURE: listiterator from class OpenEdge.Core.Collections.ListIterator, access KW_PUBLIC, static false, at new [FUNC_CLASS]:<id_unavailable> @435:16
     [java]    OpenEdge.Core.Collections.ListIterator [CLASS_NAME]:<id_unavailable> @435:20
     [java]    ( [LPARENS]:<id_unavailable> @435:58
     [java]       parameter [PARAMETER]:<id_unavailable> @0:0
     [java]          mBackingList [VAR_CLASS]:<id_unavailable> @435:59
     [java] 
     [java]    PARAMETERS: 1
     [java]        object<? extends progress.collections.ilist>

At complilation¶

Generics¶

• p2j/oo/core/String_.java:58: error: repeated interface Icomparable,
  Because in the original code there is:
  

  class OpenEdge.Core.String
  serializable
  implements ILongcharHolder
           , ISupportEncoding
           // Support comparision between either LongcharHolder or other String objects
           , IComparable<ILongcharHolder>
           , IComparable<OpenEdge.Core.String>
           , IHashable
  

• p2j/oo/core/ByteBucket.java:46: error: ByteBucket is not abstract and does not override abstract method compareTo(object) in Icomparable
• 
  

  /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/ByteBucket.java:1140: error: name clash: compareTo(object<? extends ImemptrHolder>) in ByteBucket and compareTo(object) in Icomparable have the same erasure, yet neither overrides the other
      [javac]    public integer compareTo(final object<? extends ImemptrHolder> _pOther)

Constructors:¶

    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/util/TokenResolverEventArgs.java:179: error: cannot find symbol
    [javac]          __core_EventArgs_constructor__();

Others:¶

• no suitable method found for indexOf(memptr,longchar,int64): idx.assign(indexOf(memoutStream.ref().getData(), pSearchStr, pStartAt));

• p2j/oo/core/collections/Array_.java:470: error: incompatible types: unknown cannot be converted to int64
      [javac]             Array_.this.resizeValue(new unknown());

• p2j/oo/core/EventArgs.java:74: error: no suitable method found for storeReturnValue(ContextLocal<object>): storeReturnValue(EventArgs.empty);
• p2j/oo/core/LongcharArrayHolder.java:219: error: no suitable constructor found for longchar(character[]): LongcharArrayHolder.this.bulkSetValue(new longchar(pcValue[0]));
• p2j/oo/core/collections/List.java:775: error: no suitable method found for newInstance(Class<ListIterator>,object<CAP#1>): storeReturnValue(ObjectOps.newInstance(com.goldencode.p2j.oo.core.collections.ListIterator.class, mBackingList));
• p2j/oo/core/collections/List.java:815: error: no suitable method found for newInstance(Class<ListIterator>,object<CAP#1>,integer)

Ioana, my approach for investigating compile errors is to comment or otherwise 'remove' the offending code that doesn't compile, and run ant compile again - until it completes. This way, you can get a full list of compile errors. Please do so, if you haven't already.

• the ADE conversion project I used, with a clean 9488a: ADE_project.zip
• 9488a modified with the ADE converted code inside: 9488_withADE.zip

I think the problem with the ObjectOps.newInstance missing the parameter type mode is that the target method or constructor call has parameters defined with generics - so the target can't be resolved properly.

See these errors during parsing:

     [java] annotateMethodCall FAILURE: listiterator from class OpenEdge.Core.Collections.ListIterator, access KW_PUBLIC, static false, at new [FUNC_CLASS]:<id_unavailable> @435:16
     [java]    OpenEdge.Core.Collections.ListIterator [CLASS_NAME]:<id_unavailable> @435:20
     [java]    ( [LPARENS]:<id_unavailable> @435:58
     [java]       parameter [PARAMETER]:<id_unavailable> @0:0
     [java]          mBackingList [VAR_CLASS]:<id_unavailable> @435:59
     [java]
     [java]    PARAMETERS: 1
     [java]        object<? extends progress.collections.ilist>

Considering that we can't define our own types with <T> generic, to test this properly (please confirm), I'm not sure what's the best approach here.

For now, please try to debug while annotateMethodCall fails. And also open a separate task for the generic discussions.

Greg: considering that the #9457 project doesn't use generics in the app code, I want to find a way to delay this implementation. One of the 'weird' parts is:

class OpenEdge.Core.String
serializable
implements ILongcharHolder
         , ISupportEncoding
         // Support comparision between either LongcharHolder or other String objects
         , IComparable<ILongcharHolder>
         , IComparable<OpenEdge.Core.String>
         , IHashable

Note how you can implement the same interface but with different generic types. This is not allowed in Java.

Created issue #9868 for the discussion about generics. I will try to gather some examples of generics usage and add the description.

The initial idea was to convert everything we need from ADE and later choose of we keep the current implementation(existent in p2j.oo package) or if we will keep the converted class. After discussing with Constantin today, we decided to get a minimum list of what is needed to be converted and only keep what is not existent in p2j.oo with full support.
I obtained this list:

abl/OpenEdge/Core/Collections/List.cls
abl/OpenEdge/Core/Collections/ListIterator.cls
abl/OpenEdge/Core/Collections/ObjectStack.cls
abl/OpenEdge/Core/Integer.cls
abl/OpenEdge/Core/Memptr.cls
abl/OpenEdge/Core/String.cls
abl/OpenEdge/Core/System/InvalidValueSpecifiedError.cls
abl/OpenEdge/Core/System/NotFoundError.cls
abl/Ccs/Common/Support/IInt64Holder.cls
abl/OpenEdge/Core/StringConstant.cls

I created a new /abl folder where I included only these classes (some classes not included in the conversion where parsed from the /abl folder if I did not do that).
I converted with 9488a. What I needed was to add ImemoryPointer in p2j.oo.core (this was merged to trunk with 9436a but did not reach 9488a) and also the skeleton for this class.
I still have some annotateMethodCall FAILURE that I will start investigating but these are all the compilation errors I have:

    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/Memptr.java:156: error: '(' expected
    [javac]    public void n/a(final int64 _piSize)
    [javac]                 ^
    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/Memptr.java:156: error: invalid method declaration; return type required
    [javac]    public void n/a(final int64 _piSize)
    [javac]                  ^
    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/Memptr.java:611: error: not a statement
    [javac]             Memptr.this.n/a(piSize);
    [javac]                          ^
    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/String_.java:121: error: '(' expected
    [javac]    public void n/a(character var)
    [javac]                 ^
    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/String_.java:121: error: invalid method declaration; return type required
    [javac]    public void n/a(character var)
    [javac]                  ^
    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/String_.java:166: error: not a statement
    [javac]          String_.this.n/a(new character(I18nOps.getCPInternal()));
    [javac]                        ^
    [javac] /home/icp/projects/ADE/src/com/goldencode/p2j/oo/core/String_.java:186: error: not a statement
    [javac]          String_.this.n/a(new character(longchar.getCodePage(pcString)));

• Progress.IO.MemoryOutputStream, Progress.IO.ByteOrder, OpenEdge.Core.IMemoryPointer, OpenEgde.Core.TimeStampConstants, OpenEdge.Security.Principal.

Greg: considering that the #9457 project doesn't use generics in the app code, I want to find a way to delay this implementation. One of the 'weird' parts is:
[...]

Note how you can implement the same interface but with different generic types. This is not allowed in Java.

Yuck. Some things to record for posterity:

• OO 4GL currently (as of the latest 12.8 LTS) doesn't have generics support that can be used by end user code. These references are only in the built-ins. Presumeably, they implement this limited special support in the native code of the OE runtime.
• OE is shrinking (dying actually) annually and the next LTS release is not expected until at least 2027 (best case) as explained by Progress Software Corp's (PSC) themselves. Most (all?) commercial ISVs would never deploy an OE release that is not LTS. That means that no one will be deploying the post-12.8 LTS until 2027 or possibly long after that.
• Supposedly PSC is working on generics for OO 4GL but generics can't appear in real OO 4GL code that goes to customers until that next LTS release is available so we have years to go before end user code will ever have direct access to generics.
• I wonder if OE will still exist in 2028 or if it will just be in maintenance mode. Actually, it is already kind of in maintenance mode since the amount of development on OE itself is very limited. But for the sake of argument, let's say that maintenance mode is in the future somewhere. My suggestion is that given the pace of decline, maintenance mode is possibly not that far in the future. Under those conditions we may never have to implement generics in the FWD implementation of the OO 4GL.
• Nevertheless, the current approach chosen by PSC seems to match the C# concept of generics which uses runtime information and tends to be much slower than generics in Java which use type erasure and are fast but have more type limitations on the implementation. As you note, we can't directly implement two parameterized interfaces using Java generics because of the type erasure implementation.

If it is feasible to defer the work on generics to another task, I am OK with that so long as that work can be deferred to after the current projects are complete. My point here is that if we defer the work but then just have to work it again in order to finish some current projects, then the deferral doesn't help too much. So, I'd like to know if we need it in the short term or not. If not, then we should defer it.

Greg Shah wrote:

So, I'd like to know if we need it in the short term or not. If not, then we should defer it.

The only concern is the ADE code - if we want to convert it fully, then we need generics. Ioana has a short list of files which we urgently need converted - if those can be solved without generics, then I think we are OK on the very short term.

Constantin Asofiei wrote:

Greg Shah wrote:

So, I'd like to know if we need it in the short term or not. If not, then we should defer it.

The only concern is the ADE code - if we want to convert it fully, then we need generics. Ioana has a short list of files which we urgently need converted - if those can be solved without generics, then I think we are OK on the very short term.

Agreed.

Minimized the testcase where the setter of a property is not correctly emitted and ends up as n/a.
For this:

using OpenEdge.Core.IMemoryPointer.

class OpenEdge.Core.Memptr
implements IMemoryPointer
:
    define final public property Size as int64 no-undo
        get():
            return 0. 
        end get.
        set(input piSize as int64):
        end set.
end class.

public class Memptr
extends BaseObject
implements ImemoryPointer
{
// ..
   @LegacySignature(type = Type.SETTER, name = "Size", parameters = 
   {
      @LegacyParameter(name = "piSize", type = "INT64", mode = "INPUT")
   })
   @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_STUB)
   public void n/a(final int64 _piSize)
   {
      int64 piSize = TypeFactory.initInput(_piSize);

      internalProcedure(Memptr.class, this, "Size", new Block());
   }
// .. 
}

If I add IMemoryPointer in the /abl folder, but not include it in the file-cvt-list, the behavior changes. It converts to:

public class Memptr
extends BaseObject
implements n/a
{
// ..
// the setter is emmitted correctly here
   @LegacySignature(type = Type.SETTER, name = "Size", parameters = 
   {
      @LegacyParameter(name = "piSize", type = "INT64", mode = "INPUT")
   })
   @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_STUB)
   public void setSize(final int64 _piSize)
   {
      int64 piSize = TypeFactory.initInput(_piSize);

      internalProcedure(Memptr.class, this, "Size", new Block());
   }
// ..
}

I've added my changes to 9488a rev 15736. Sorry, they are not fully tested/complete. Please check with your test now.

Also: please rebase the branch. The skeletons you added are already in trunk.

Rebased 9488a and tested again the conversion for only these files (needed from ADE and not existent in p2j with full support reported):

abl/OpenEdge/Core/Collections/List.cls
abl/OpenEdge/Core/Collections/ListIterator.cls
abl/OpenEdge/Core/Collections/ObjectStack.cls
abl/OpenEdge/Core/Integer.cls
abl/OpenEdge/Core/Memptr.cls
abl/OpenEdge/Core/String.cls
abl/OpenEdge/Core/System/InvalidValueSpecifiedError.cls
abl/OpenEdge/Core/System/NotFoundError.cls
abl/Ccs/Common/Support/IInt64Holder.cls
abl/OpenEdge/Core/StringConstant.cls

LegacyList(OpenEdge.Core.Collections.List)¶

• return new OpenEdge.Core.Collections.ListIterator(mBackingList).
  

  
  converts to:
  

  storeReturnValue(ObjectOps.newInstance(com.goldencode.p2j.oo.core.collections.LegacyListIterator.class, mBackingList));  
  

  
  Compilation error: Cannot resolve method 'newInstance(Class<LegacyListIterator>, object<capture of ? extends Ilist>)'

• return new OpenEdge.Core.Collections.ListIterator(mBackingList, pStartIdx).
  

  
  converts to:
  

  storeReturnValue(ObjectOps.newInstance(com.goldencode.p2j.oo.core.collections.LegacyListIterator.class, mBackingList, pStartIdx));   
  

  
  Compilation error: Cannot resolve method 'newInstance(Class<LegacyListIterator>, object<capture of ? extends Ilist>, integer)'

LegacyInteger(OpenEdge.Core.Integer)¶

Generics issue:
LegacyInteger implements Icomparable, this interface uses generics now:

@LegacyResource(resource = "Progress.Collections.IComparable")
@LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_STUB)
public interface Icomparable<T extends _BaseObject_>
extends com.goldencode.p2j.oo.lang._BaseObject_
{
   @LegacySignature(type = Type.METHOD, name = "CompareTo", returns = "INTEGER", parameters = 
   {
      @LegacyParameter(name = "otherObject", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT")
   })
   @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_STUB)
   public integer compareTo(final object<? extends T> _otherObject);
}

public class LegacyInteger
extends BaseObject
implements Iint64Holder,
           Icomparable,
           Ihashable
{
   // ..
   public integer compareTo(final object<? extends Iint64Holder> _pOther)
   {
      // ..
   }
   // ..
}

Memptr(OpenEdge.Core.Memptr)¶

The same issue as in LegacyList, here we need implements Icomparable<ImemptrHolder>

LegacyString(OpenEdge.Core.String)¶

Almost the same issue as in LegacyList, but here there is also the issue with the same class being implemented 2 times with different generics:

class OpenEdge.Core.String
serializable
implements ILongcharHolder
         , ISupportEncoding
         , IComparable<ILongcharHolder>
         , IComparable<OpenEdge.Core.String>
         , IHashable

public class LegacyString
extends BaseObject
implements IlongcharHolder,
           ISupportEncoding,
           Icomparable,
           Icomparable,
           Ihashable
{
   // ..
   public integer compareTo(final object<? extends IlongcharHolder> _pOther)
   {
      // ..
   }
   public integer compareTo_1(final object<? extends com.goldencode.p2j.oo.core.LegacyString> _pOther)
   {
      // ..
   }
   // ..
}

Compilation errors are solved if the implements clause is:

implements IlongcharHolder,
           ISupportEncoding,
           Icomparable<IlongcharHolder>,
           Ihashable

I will start with the iterator issues in LegacyList, not sure what is happening but it is most likely linked to this parsing error:

     [java] WARNING: more than one method def found using fuzzy lookup: Iterator from class OpenEdge.Core.Collections.IList
     [java] 
     [java]    CALLER Iterator [OO_METH_CLASS]:<id_unavailable> @378:32
     [java]       (original-token=["Iterator",<497>,line=378,col=32])
     [java] 
     [java]    PARAMETERS: 0
     [java]    MATCH 0: METRICS 0 exact; 0 cvt; 0 extExact; 0 extCvt; 0 wildcard; [I@1f4cb17b ooLvl; [Ljava.lang.String;@280099a0 overrides; 
     [java]         KW_PUBLIC instance member Iterator (OO_METH_CLASS - openedge.core.collections.iiterator), extent 0, javaname iterator, override false, abstract false in class OpenEdge.Core.Collections.IIterable, tempIdx 0
     [java]    PARAMETERS: 0
     [java]  data
     [java] 
     [java]    MATCH 1: METRICS 0 exact; 0 cvt; 0 extExact; 0 extCvt; 0 wildcard; [I@7aca299e ooLvl; [Ljava.lang.String;@1bf52f10 overrides; 
     [java]         KW_PUBLIC instance member Iterator (OO_METH_CLASS - openedge.core.collections.iiterator), extent 0, javaname iterator, override false, abstract false in class OpenEdge.Core.Collections.IIterable, tempIdx 0
     [java]    PARAMETERS: 0
     [java]  data
     [java] 
     [java]    Lvl01 parse: ./abl/OpenEdge/Core/Collections/ListIterator.cls
     [java]    Lvl02 parse: | ./skeleton/oo4gl/OpenEdge/Core/Collections/Iterator.cls
     [java]    Lvl01 DONE:  ./abl/OpenEdge/Core/Collections/ListIterator.cls
     [java] 
     [java] annotateMethodCall FAILURE: listiterator from class OpenEdge.Core.Collections.ListIterator, access KW_PUBLIC, static false, at new [FUNC_CLASS]:<id_unavailable> @435:16
     [java]    OpenEdge.Core.Collections.ListIterator [CLASS_NAME]:<id_unavailable> @435:20
     [java]    ( [LPARENS]:<id_unavailable> @435:58
     [java]       parameter [PARAMETER]:<id_unavailable> @0:0
     [java]          mBackingList [VAR_CLASS]:<id_unavailable> @435:59
     [java] 
     [java]    PARAMETERS: 1
     [java]        object<? extends progress.collections.ilist>
     [java] 

Why are you converting String.cls? This is already implemented. Please post the only the .cls files which have no implementation in FWD.

This is the support for these in p2j:
- OpenEdge.Core.Collection.List (p2j.oo.core.collections.LegacyList) - STUBS
- OpenEdge.Core.Collection.ListIterator (p2j.oo.core.collections.LegacyListIterator) - STUBS
- OpenEdge.Core.Collection.ObjectStack (p2j.oo.core.collections.ObjectStack) - STUBS
- OpenEdge.Core.Integer (p2j.oo.core.LegacyInteger) - STUBS
- OpenEdge.Core.Memptr (p2j.oo.core.Memptr) - PARTIAL
- OpenEdge.Core.String (p2j.oo.core.LegacyString) - PARTIAL
- OpenEdge.Core.System.InvalidValueSpecifiedError (p2j.oo.core.system.InvalidValueSpecifiedError) - STUBS
- Ccs.Common.Support.IInt64Holder (p2j.oo.common.support.Iint64Holder)(needed in OpenEdge.Core.Integer) - STUBS
- OpenEdge.Core.StringConstant (p2j.oo.core.StringConstant)(needed in OpenEdge.Core.String) - STUBS
If partial support is ok for now, the conversion list is:

abl/OpenEdge/Core/Collections/List.cls
abl/OpenEdge/Core/Collections/ListIterator.cls
abl/OpenEdge/Core/Collections/ObjectStack.cls
abl/OpenEdge/Core/Integer.cls
abl/OpenEdge/Core/System/InvalidValueSpecifiedError.cls
abl/OpenEdge/Core/System/NotFoundError.cls
abl/Ccs/Common/Support/IInt64Holder.cls

After conversion, compilation error are in LegacyInteger and LegacyList, the ones described in #9488-118.

For LegacyInteger, for now make the change by hand.
For LegacyList, these are related to generics (I think) - see the annotateMethodCall FAILURE: errors.

Also, for the CCS and other 'holder' classes - are these fully implemented in FWD?

• IcharacterHolder.java
• IdecimalHolder.java
• Iint64Holder.java
• IlongcharHolder.java
• IprimitiveArrayHolder.java
• IdateTimeTzHolder.java
• IhandleHolder.java
• IlongcharArrayHolder.java
• ImemptrHolder.java
• IprimitiveHolder.java

Some are reported with STUBS runtime support level but as these are interfaces it could be considered FULL.
Iint64Holder converted is the same as the one in p2j, I will remove this one from the ADE conversion list.

• IRowidArrayHolder.cls
• ICharacterArrayHolder.cls
• IMemptrArrayHolder.cls
• IIntegerArrayHolder.cls
• IDateTimeTzArrayHolder.cls
• IDecimalArrayHolder.cls
• IDateHolder.cls
• IDateTimeHolder.cls
• IDateTimeArrayHolder.cls
• IInt64ArrayHolder.cls
• IHandleArrayHolder.cls
• IIntegerHolder.cls
• IRowidHolder.cls
• ILogicalHolder.cls
• ILogicalArrayHolder.cls
• IDateArrayHolder.cls

Ioana-Cristina Prioteasa wrote:

What I could find in /CCcs/Common/Support and is not in FWD:

Please add these to your 'short' list:

The issue in OpenEdge.Core.Collections.List:
annotateMethodCall failed because there was no method match. This happened because in ClassDefinition.checkAccessRights, for the method that should be resolved mdat.access was KW_PK_PRIV, while access was KW_PUBLIC.

   private MemberData checkAccessRights(MemberData mdat, int access)
   {
      if (mdat != null)
      {
         // check for a match to the access mode:
         // 1. searching for private methods match everything
         // 2. a direct match is always OK
         // 3. a public method matches a protected search
         boolean ok = (access == KW_PRIVATE)  ||
                      (access == mdat.access) ||
                      (access == KW_PROTECTD || && mdat.access == KW_PUBLIC);

         // pretend no match if access modes are not right
         if (!ok)
         {
            return null;
         }
      }

      return mdat;
   }

I modified SymbolResolver.calcAccess, which was responsible for the access parameter passed to checkAccessRights. I added a method to check the packages and also account for KW_PK_PRIV. In checkAccessRights:

         boolean ok = (access == KW_PRIVATE)  ||
                      (access == mdat.access) ||
                      ((access == KW_PROTECTD || access == KW_PK_PRIV) && mdat.access == KW_PUBLIC)||
                      (access == KW_PROTECTD  && mdat.access == KW_PK_PRIV);

     [java] annotateMethodCall FAILURE: stack from class OpenEdge.Core.Collections.Stack, access KW_PROTECTD, static false, at super [KW_SUPER]:<id_unavailable> @22:9
     [java]    ( [LPARENS]:<id_unavailable> @22:14
     [java]       parameter [PARAMETER]:<id_unavailable> @0:0
     [java]          expression [EXPRESSION]:<id_unavailable> @0:0
     [java]             poArray [VAR_CLASS]:<id_unavailable> @22:15

METRICS 0 exact; 0 cvt; 0 extExact; 0 extCvt; 0 wildcard; [I@3d42b756 ooLvl; [Ljava.lang.String;@7c93e69 overrides; 
        KW_PROTECTD instance member Stack (OO_METH_VOID), extent 0, javaname stack, override false, abstract false in class OpenEdge.Core.Collections.Stack, tempIdx 2
   PARAMETERS: 1
      KW_INPUT object<? extends progress.lang.object>[]
 data

METRICS 0 exact; 0 cvt; 0 extExact; 0 extCvt; 0 wildcard; [I@4e4c05a0 ooLvl; [Ljava.lang.String;@d763b06 overrides; 
        KW_PROTECTD instance member Stack (OO_METH_VOID), extent 0, javaname stack, override false, abstract false in class OpenEdge.Core.Collections.Stack, tempIdx 4
   PARAMETERS: 1
      KW_INPUT character[]
 data

METRICS 0 exact; 0 cvt; 0 extExact; 0 extCvt; 0 wildcard; [I@2526d5f9 ooLvl; [Ljava.lang.String;@6b529d42 overrides; 
        KW_PUBLIC instance member Stack (OO_METH_VOID), extent 0, javaname stack, override false, abstract false in class OpenEdge.Core.Collections.Stack, tempIdx 6
   PARAMETERS: 1
      KW_INPUT integer
 data

 // no extent in caller
      if (callerExtVal == 0)
      {
         // no extent in the parameter definition
         test = (mem) -> mem.data.getSignature(idx).type.indexOf("[") == -1;

         // no marking needed here
         mark = (m) -> {};
      }

// no extent in caller
      if (callerExtVal == 0)
      {
         test = (mem) ->
         {
            int parmExt = getExtent(mem.data.getSignature(idx).type);
            return parmExt == 0 || parmExt == -1;
         };

         mark = (mem) ->
         {
            int parmExt = getExtent(mem.data.getSignature(idx).type);
            if (parmExt == 0)
            {
               mem.extExact++;
            }
            else if (parmExt == -1)
            {
               mem.extCvt++;  
            }
         };
      }

This would solve the annotateMethodCall issues. I will commit in 9488a, but I am not exactly sure if this is correct.

I was wrong in my last post. This is the original 4GL code for the second annotateMethodCall issue:

block-level on error undo, throw.

using OpenEdge.Core.Collections.ObjectStack.
using OpenEdge.Core.Collections.Stack.
using Progress.Lang.Object.

class OpenEdge.Core.Collections.ObjectStack serializable inherits Stack:

    constructor public ObjectStack (poArray as Object extent):
        super(poArray).
    end constructor.
    // other methods
end class.

constructor protected Stack ( poArray as Progress.Lang.Object extent ):
end.

      if (node.isAnnotation("extent"))
      {
         extent = (int) ((long) node.getAnnotation("extent"));
      }
      else
      {
         Aast child = node.getImmediateChild(KW_EXTENT, null);

         if (child != null)
         {
            // there is an extent, default to indeterminate
            extent = -1;

            Aast num = child.getChildAt(0);

            if (num != null)
            {
               extent = ExpressionConversionWorker.literalToInt(num);
            }
         }
      }

Something like this would solve the issue:

      if (node.isAnnotation("extent"))
      {
         extent = (int) ((long) node.getAnnotation("extent"));
      }
      else if (node.getChildAt(0)!= null && node.getChildAt(0).isAnnotation("extent"))
      {
         extent = (int) ((long) node.getChildAt(0).getAnnotation("extent"));
      }
      else
      {
         Aast child = node.getImmediateChild(KW_EXTENT, null);

         if (child != null)
         {
            // there is an extent, default to indeterminate
            extent = -1;

            Aast num = child.getChildAt(0);

            if (num != null)
            {
               extent = ExpressionConversionWorker.literalToInt(num);
            }
         }
      }

• Added SymbolResolver.getLegacyPackageName method that extracts the legacy package name from the fully qualified class name.
• Corrected SymbolResolver.calcAccessMode to also calculate the package private access mode.
• Modified SymbolResolver.readExtent to resolve the extent in the case the node's type is EXPRESSION and the extent annotation is present in the child.
• Corrected ClassDefinition.checkAccessRights, added the case when a public method matches a package private search.

I am also in the process of consolidating the new list Constantin emailed me, containing needed classes.

I added OpenEdge/Collections/Array.cls to conversion and another annotateMethod failure appears. I am looking into it now.

This is the latest list of what is needed from the OpenEdge, Ccs and OEUnit packages:

Already implemented with reported runtime support FULL¶

• ./skeleton/oo4gl/Ccs/Common/Support/IInt64Holder.cls - existent in p2j at com.goldencode.p2j.oo.common.support.Iint64Holder - rt STUB level reported but is interface, equivalent to full.
• ./skeleton/oo4gl/Ccs/Common/Support/ILongcharHolder.cls - existent in p2j at com.goldencode.p2j.oo.common.support.IlongcharHolder
• ./skeleton/oo4gl/Ccs/Common/Support/IMemptrHolder.cls - existent in p2j at com.goldencode.p2j.oo.common.support.ImemptrHolder
• ./skeleton/oo4gl/Ccs/Common/Support/IPrimitiveHolder.cls - existent in p2j at com.goldencode.p2j.oo.common.support.IprimitiveHolder

• ./skeleton/oo4gl/OEUnit/Assertion/Assert.cls - existent in p2j at com.goldencode.p2j.oo.oeunit.assertion.Assert
• ./skeleton/oo4gl/OEUnit/Assertion/AssertionFailedError.cls - existent in p2j at com.goldencode.p2j.oo.oeunit.assertion.AssertionFailedError
• ./skeleton/oo4gl/OEUnit/Assertion/AssertString.cls - existent in p2j at com.goldencode.p2j.oo.oeunit.assertion.AssertString
• ./skeleton/oo4gl/OEUnit/Data/DataProvider.cls - existent in p2j at com.goldencode.p2j.oo.oeunit.data.DataProvider_
• ./skeleton/oo4gl/OEUnit/Data/Fixture.cls - existent in p2j at com.goldencode.p2j.oo.oeunit.data.Fixture_

• ./skeleton/oo4gl/OpenEdge/Core/DataTypeEnum.cls - existent in p2j at com.goldencode.p2j.oo.core.DataTypeEnum
• ./skeleton/oo4gl/OpenEdge/Core/HashAlgorithmEnum.cls - existent in p2j at com.goldencode.p2j.oo.core.HashAlgorithmEnum
• ./skeleton/oo4gl/OpenEdge/Core/IOModeEnum.cls - existent in p2j at com.goldencode.p2j.oo.core.IomodeEnum
• ./skeleton/oo4gl/OpenEdge/Core/ISupportEncoding.cls - existent in p2j at com.goldencode.p2j.oo.core.ISupportEncoding
• ./skeleton/oo4gl/OpenEdge/Core/ISupportInitialize.cls - existent in p2j at com.goldencode.p2j.oo.core.ISupportInitialize
• ./skeleton/oo4gl/OpenEdge/Core/Collections/AbstractTTCollection.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.AbstractTtcollection
• ./skeleton/oo4gl/OpenEdge/Core/Collections/Collection.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.LegacyCollection
• ./skeleton/oo4gl/OpenEdge/Core/Collections/ICollection.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.Icollection
• ./skeleton/oo4gl/OpenEdge/Core/Collections/IIterable.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.Iiterable
• ./skeleton/oo4gl/OpenEdge/Core/Collections/IIterator.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.Iiterator
• ./skeleton/oo4gl/OpenEdge/Core/Collections/IList.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.Ilist
• ./skeleton/oo4gl/OpenEdge/Core/Collections/IListIterator.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.IlistIterator
• ./skeleton/oo4gl/OpenEdge/Core/Collections/IMap.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.Imap
• ./skeleton/oo4gl/OpenEdge/Core/Collections/ISet.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.Iset
• ./skeleton/oo4gl/OpenEdge/Core/Collections/IStringStringMap.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.IStringStringMap
• ./skeleton/oo4gl/OpenEdge/Core/Collections/Iterator.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.LegacyIterator
• ./skeleton/oo4gl/OpenEdge/Core/Collections/Map.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.LegacyMap
• ./skeleton/oo4gl/OpenEdge/Core/System/ApplicationError.cls - existent in p2j at com.goldencode.p2j.oo.core.system.ApplicationError
• ./skeleton/oo4gl/OpenEdge/Core/System/UnsupportedOperationError.cls - existent in p2j at com.goldencode.p2j.oo.core.system.UnsupportedOperationError
• ./skeleton/oo4gl/OpenEdge/Core/Util/BuilderRegistry.cls - existent in p2j at com.goldencode.p2j.oo.core.util.BuilderRegistry
• ./skeleton/oo4gl/OpenEdge/Core/Util/ConfigBuilder.cls - existent in p2j at com.goldencode.p2j.oo.core.system.ConfigBuilder
• ./skeleton/oo4gl/OpenEdge/Logging/ILogWriter.cls - existent in p2j at com.goldencode.p2j.oo.logging.IlogWriter
• ./skeleton/oo4gl/OpenEdge/Logging/LogLevelEnum.cls - existent in p2j at com.goldencode.p2j.oo.logging.IlogWriter
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/BuilderRegistry.cls - existent in p2j at com.goldencode.p2j.oo.net.http.BuilderRegistry
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/ClientBuilder.cls - existent in p2j at com.goldencode.p2j.oo.net.http.ClientBuilder
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/ClientOptions.cls - existent in p2j at com.goldencode.p2j.oo.net.http.ClientOptions
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/ConfigBuilder.cls - existent in p2j at com.goldencode.p2j.oo.net.http.ConfgiBuilder
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/Cookie.cls - existent in p2j at com.goldencode.p2j.oo.net.http.Cookie
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/Credentials.cls - existent in p2j at com.goldencode.p2j.oo.net.http.Credentials
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/HttpHeader.cls - existent in p2j at com.goldencode.p2j.oo.net.http.HttpHeader
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/ICookieJar.cls - existent in p2j at com.goldencode.p2j.oo.net.http.ICookieJar
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/IHttpClient.cls - existent in p2j at com.goldencode.p2j.oo.net.http.IhttpClient
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/IHttpClientLibrary.cls - existent in p2j at com.goldencode.p2j.oo.net.http.IHttpClientLibrary - rt PARTIAL, but is interface so it should be FULL.
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/IHttpMessage.cls - existent in p2j at com.goldencode.p2j.oo.net.http.IhttpMessage
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/IHttpRequest.cls - existent in p2j at com.goldencode.p2j.oo.net.http.IhttpRequest
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/IHttpResponse.cls - existent in p2j at com.goldencode.p2j.oo.net.http.IhttpResponse
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/Lib/ClientLibraryBuilder.cls - existent in p2j at com.goldencode.p2j.oo.net.http.lib.ClientLibraryBuilder
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/RequestBuilder.cls - existent in p2j at com.goldencode.p2j.oo.net.http.RequestBuilder
• ./skeleton/oo4gl/OpenEdge/Net/UriEncodingTypeEnum.cls - existent in p2j at com.goldencode.p2j.oo.net.UriEncodingTypeEnum

Existent with runtime support PARTIAL¶

• ./skeleton/oo4gl/OpenEdge/Core/Assert.cls - existent in p2j at com.goldencode.p2j.oo.core.Assert
• ./skeleton/oo4gl/OpenEdge/Core/ByteBucket.cls - existent in p2j at com.goldencode.p2j.oo.core.ByteBucker
• ./skeleton/oo4gl/OpenEdge/Core/Memptr.cls - existent in p2j at com.goldencode.p2j.oo.core.Memptr
• ./skeleton/oo4gl/OpenEdge/Core/String.cls - existent in p2j at com.goldencode.p2j.oo.core.LegacyString@
• ./skeleton/oo4gl/OpenEdge/Net/HTTP/TransferEncodingEnum.cls - existent in p2j at com.goldencode.p2j.oo.net.http.TransferEncodingEnum
• ./skeleton/oo4gl/OpenEdge/Net/URI.cls - existent in p2j at com.goldencode.p2j.oo.net.Uri
• ./skeleton/oo4gl/OpenEdge/Core/AbstractAssertionFailedError.cls - existent in p2j at com.goldencode.p2j.oo.core.AbstractAssertionFailedError - rt STUBS but seems at least partially implemented.

Needs to be converted from ADE - runtime support STUB¶

• ./skeleton/oo4gl/OpenEdge/Core/Integer.cls - existent in p2j at com.goldencode.p2j.oo.core.LegacyInteger
• ./skeleton/oo4gl/OpenEdge/Core/TimeStamp.cls - existent in p2j at com.goldencode.p2j.oo.core.TimeStamp
• ./skeleton/oo4gl/OpenEdge/Core/Collections/Array.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.Array
• ./skeleton/oo4gl/OpenEdge/Core/Collections/List.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.LegacyList
• ./skeleton/oo4gl/OpenEdge/Core/Collections/ListIterator.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.LegacyListIterator
• ./skeleton/oo4gl/OpenEdge/Core/Collections/ObjectStack.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.ObjectStack
• ./skeleton/oo4gl/OpenEdge/Core/Collections/Stack.cls - existent in p2j at com.goldencode.p2j.oo.core.collections.LegacyStack
• ./skeleton/oo4gl/OpenEdge/Core/System/InvalidValueSpecifiedError.cls - existent in p2j at com.goldencode.p2j.oo.core.system.InvalidValueSpecifiedError
• ./skeleton/oo4gl/OpenEdge/Core/System/NotFoundError.cls - existent in p2j at com.goldencode.p2j.oo.core.system.NotFoundError
• ./skeleton/oo4gl/OpenEdge/Logging/LogEvent.cls - existent in p2j at com.goldencode.p2j.oo.logging.LogEvent
• ./skeleton/oo4gl/OpenEdge/Logging/LogMessage.cls - existent in p2j at com.goldencode.p2j.oo.logging.LogMessage

There are 2 classes that are from OEUnit that we have with STUB level support, not sure what to do about these:

• ./skeleton/oo4gl/OEUnit/Data/FixtureDataSet.cls - existent in p2j at com.goldencode.p2j.oo.oeunit.data.FixtureDataSet
• ./skeleton/oo4gl/OEUnit/Util/List.cls - existent in p2j at com.goldencode.p2j.oo.oeunit.util.List

Conversion list¶

abl/OpenEdge/Core/Collections/List.cls
abl/OpenEdge/Core/Collections/Array.cls
abl/OpenEdge/Core/Collections/ArrayIterator.cls
abl/OpenEdge/Core/Collections/Stack.cls
abl/OpenEdge/Core/Collections/ListIterator.cls
abl/OpenEdge/Core/Collections/ObjectStack.cls

abl/OpenEdge/Core/Integer.cls
abl/OpenEdge/Core/TimeStamp.cls
abl/OpenEdge/Core/TimeStampConstants.cls

abl/OpenEdge/Core/System/InvalidValueSpecifiedError.cls
abl/OpenEdge/Core/System/NotFoundError.cls

abl/OpenEdge/Logging/LogEvent.cls
abl/OpenEdge/Logging/LogMessage.cls

abl/Ccs/Common/Support/ICharacterArrayHolder.cls
abl/Ccs/Common/Support/IDateTimeHolder.cls
abl/Ccs/Common/Support/IInt64ArrayHolder.cls
abl/Ccs/Common/Support/IIntegerArrayHolder.cls
abl/Ccs/Common/Support/IMemptrArrayHolder.cls
abl/Ccs/Common/Support/IDateArrayHolder.cls
abl/Ccs/Common/Support/IDateTimeTzArrayHolder.cls
abl/Ccs/Common/Support/IIntegerHolder.cls
abl/Ccs/Common/Support/IRowidArrayHolder.cls
abl/Ccs/Common/Support/IDateHolder.cls
abl/Ccs/Common/Support/IDecimalArrayHolder.cls
abl/Ccs/Common/Support/ILogicalArrayHolder.cls
abl/Ccs/Common/Support/IRowidHolder.cls
abl/Ccs/Common/Support/IDateTimeArrayHolder.cls
abl/Ccs/Common/Support/IHandleArrayHolder.cls
abl/Ccs/Common/Support/ILogicalHolder.cls

The cause for the annotateMethodCall failure in OpenEdge.Core.Collections.Array was this line:

return new ArrayIterator(this-object, pIndex).

Please ignore the ./skeleton/oo4gl/OEUnit/Data/FixtureDataSet.cls and ./skeleton/oo4gl/OEUnit/Util/List.cls.

Compilation Errors

• LegacyInteger: we need implements Icomparable<Iint64Holder>
• TimeStamp: we need implements Icomparable<TimeStamp>

2. Issues in Array:

•         // If the new size is 0 then make the array intedeterminate
          if pNewLength eq 0 then
          do:
              assign extent(this-object:Value) = ?.
              return 0.
          end.
  

  
  converts to:
  

           //  If the new size is 0 then make the array intedeterminate
           if (_isEqual(pNewLength, 0))
           {
              Array.this.resizeValue(new unknown());
              storeReturnValue(0);
              return;
           }
  

  
  

     @LegacySignature(type = Type.RESIZE, name = "Value", parameters = 
     {
        @LegacyParameter(name = "size", type = "INT64", mode = "INPUT")
     })
     @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_FULL)
     protected void resizeValue(final int64 _size)
     {
        int64 size = TypeFactory.initInput(_size);
  
        internalProcedure(Array.class, this, "resize-Value", new Block((Body) () -> 
        {
           value = ArrayAssigner.resize(value, size);
        }));
     }
  

  
  The compilation error here is: 'resizeValue(com.goldencode.p2j.util.int64)' in 'com.goldencode.p2j.oo.core.collections.Array' cannot be applied to '(com.goldencode.p2j.util.unknown)'
  I think it would be correct to emit:
  

  Array.this.resizeValue(new int64());
  

• // make room if needed. We are expanding this array potentially
  UpdateLength(max(this-object:Length, mSize + 1, pIndex + 1)).
  

  
  

   method private integer UpdateLength(input pNewLength as integer):
  

  
  converts to:
  

  //  make room if needed. We are expanding the array
  updateLength(integer.maximum(Array.this.getLength(), plus(mSize, cnt), plus(idxForSize, cnt)));
  

  
  but integer.maximum calls int64.maximum that returns int64. The error: 'updateLength(com.goldencode.p2j.util.integer)' in 'com.goldencode.p2j.oo.core.collections.Array' cannot be applied to '(com.goldencode.p2j.util.int64)'.
  Should there be a maximum method in integer?

I think the issue with new unknown() is in /rules/convert/literals.rules:

                  <!-- handle method call and OO method call parms need wrapping -->
                  <rule>
                     (parent.type &gt; prog.begin_meth and parent.type &lt; prog.end_meth) or
                     (parent.type &gt; prog.begin_oo_meth and
                         parent.type &lt; prog.end_oo_meth) or
                     (parent.type == prog.expression and 
                         parent.parent.type &gt; prog.begin_meth and 
                         parent.parent.type &lt; prog.end_meth)                            or
                     (parent.type == prog.expression and 
                         parent.parent.type &gt; prog.begin_oo_meth and 
                         parent.parent.type &lt; prog.end_oo_meth)
                     <!-- lookup override -->
                     <action>classname = getNoteString("classname")</action>

                     <rule>classname == null              and 
                           parent.type == prog.expression and 
                           parent.isAnnotation("classname")
                        <action>classname = getReferenceNoteString(parent.id, "classname")</action>
                     </rule>

I added this and it solves the issue but I am not sure it is ok:

                     <rule>
                        parent.type == prog.oo_meth_void and
                        parent.matchAnnotation("property-access-type", "resize", null)
                        <action>classname = "integer"</action>
                     </rule>

The parent.type check was not needed, it was included in the parent rule. I committed this solution to 9488a rev.15862.
I am looking now at the second compilation error: integer.maximum is returning an int64, inside a function that expects integer.

What I mentioned about MIN/MAX was related to a CASE i * 1, where multiply returns int64 while the expression is really integer... see #9457-27

In your case, I think we need to wrap the minimum argument in the target's expected type.

There is one more issue. In ObjectStack, the call to the Stack constructor with parameter has the wrong name:

   @LegacySignature(type = Type.CONSTRUCTOR, name = "ObjectStack", parameters = 
   {
      @LegacyParameter(name = "piDepth", type = "INTEGER", mode = "INPUT")
   })
   @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_FULL)
   public void __core_collections_objectStack_constructor__(final integer _piDepth)
   {
      integer piDepth = TypeFactory.initInput(_piDepth);

      internalProcedure(ObjectStack.class, this, "ObjectStack", new Block((Body) () -> 
      {
         __core_collections_stack_constructor__(piDepth);
      }));
   }

   @LegacySignature(type = Type.CONSTRUCTOR, name = "ObjectStack")
   @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_FULL)
   public void __core_collections_objectStack_constructor__()
   {
      internalProcedure(ObjectStack.class, this, "ObjectStack", new Block((Body) () -> 
      {
         __core_collections_LegacyStack_constructor__();
      }));
   }

__core_collections_stack_constructor__(piDepth);

__core_collections_LegacyStack_constructor__(piDepth);

I did some tests regarding MIN MAX:

   @Test.
   METHOD PUBLIC VOID Test_Maximum_As_Parameter():
      DEFINE VARIABLE result AS INTEGER NO-UNDO.
      result = GetValue(MAXIMUM(3, 9)).
      Assert:Equals(9, result).
   END METHOD.

   METHOD PRIVATE INTEGER GetValue(INPUT val AS INTEGER):
      RETURN val.
   END METHOD.

result.assign(getValue(new integer(integer.maximum(new integer(3), new integer(9)))));

Ioana-Cristina Prioteasa wrote:

May these be related to 9488 rev. 15860?

See 9488a rev 15863, this solves the constructor problems.

I am trying now to eliminate ArrayIterator from the ADE conversion list, I have the new skeleton version (that contains the needed constructor) and I also added the constructor to the runtime. I eliminated from the class from the conversion list and I deleted the progress class from /abl. I still get the annotateMethod failure. I debugged it and in ClassDefinition.candidates:

Map<String, Map<SignatureKey, MemberData>> map = isStatic ? smethods : imethods;

Committed to 9488a rev. 15864: Relaxed restriction in wrapper detection rule to allow built-in OO method calls to be wrapped appropriately when parameter type requires casting.
The issue was in rules/convert/base_structure.xml, the OO method calls were excluded from wrapping of integer:

             parent.isAnnotation("refid")                                                          and
             ((evalLib("function_call_type", parent.type)                                  and 
               (!parent.isAnnotation("param_modes") or copy.indexPos > 0))                      or 
-             (evalLib("oo_call_type", parent, copy) and !evalLib("is_builtin_cls", parent) and
+             (evalLib("oo_call_type", parent, copy) and
               !evalLib("is_java_cls", parent) and isNote("param_index")))

             <action>

ADE_converted_sources.zip is available at devsrv01:/home/icp/ADE_converted_sources.zip.

I am working now on implementing Progress.Collections.List. We decided to support List<Object> for now and in the test I am inserting in the list OpenEdge.Core.Integer and OpenEdge.Core.String as elements.
For this I tried adding the converted LegacyInteger in 9436b. It comes with a compilation error, not sure why I have not seen it before:

> Compile failed; see the compiler error output for details.[ant:javac] /home/icp/gcd/branches/9436b/src/com/goldencode/p2j/oo/core/LegacyInteger.java:220: error: method hashCode in class BaseDataType cannot be applied to given types;
     [exec] 
     [exec] 
     [exec] * Try:
     [exec] [ant:javac]          storeReturnValue(BaseDataType.hashCode(LegacyInteger.this.getLegacyClass().ref().getTypeName(), LegacyInteger.this.getValue()));> Run with --stacktrace option to get the stack trace.

storeReturnValue(BaseDataType.hashCode(LegacyInteger.this.getLegacyClass().ref().getTypeName(), LegacyInteger.this.getValue()));

storeReturnValue(LegacyInteger.this.hashCode());

Ioana-Cristina Prioteasa wrote:

The problematic line:
[...]
Is this a conversion issue?
For it to compile I modified it to:
[...]
Not sure if that is ok.

What's the 4GL code here?

This is the function:

method public integer HashCode ():
   return hash-code(this-object:GetClass():TypeName. this-object:Value).
end method.

While working on the implementation for ArrayIterator, I discovered a conversion issue that appears twice in OpenEdge.Core.Collections.Array.
Original code:

    /* The actual holding array */
    define protected property Value as Object extent no-undo get. set.

    method public void SetValue(input poValue as Object,
                                input piExtent as integer):
        Assert:IsPositive(piExtent, 'Index').

        if this-object:Length eq ?
        or piExtent gt this-object:Length
        then
            // this method will throw a ResizeError if needed
            UpdateLength(integer(piExtent + round(0.5 * this-object:Length, 0))).

        assign this-object:Value[piExtent] = poValue
               mSize                       = max(mSize, piExtent)
               .
    end method.

• The setter for the Value property:
  

     @LegacySignature(type = Type.SETTER, name = "Value", parameters = 
     {
        @LegacyParameter(name = "var", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT"),
        @LegacyParameter(name = "idx", type = "INT64", mode = "INPUT")
     })
     @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_FULL)
     protected void setValue(final object<? extends _BaseObject_> _var, final int64 _idx)
     {
        object<? extends _BaseObject_> var = TypeFactory.initInput(_var);
        int64 idx = TypeFactory.initInput(_idx);
  
        internalProcedure(Array.class, this, "Value", new Block((Body) () -> 
        {
           assignSingle(value, idx, var);
        }));
     }
  

• The setValue method:
  

     @LegacySignature(type = Type.METHOD, name = "SetValue", parameters = 
     {
        @LegacyParameter(name = "poValue", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT"),
        @LegacyParameter(name = "piExtent", type = "INTEGER", mode = "INPUT")
     })
     @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_FULL)
     /* Set the value for a particular index to an object
  
            @param P.L.Object An object to add to the array
            @param integer  The index at which to add */   
     public void setValue(final object<? extends _BaseObject_> _poValue, final integer _piExtent)
     {
        object<? extends _BaseObject_> poValue = TypeFactory.initInput(_poValue);
        integer piExtent = TypeFactory.initInput(_piExtent);
  
        internalProcedure(Array.class, this, "SetValue", new Block((Body) () -> 
        {
           Assert.isPositive(piExtent, new character("Index"));
  
           if (_or(isUnknown(Array.this.getLength()), () -> isGreaterThan(piExtent, Array.this.getLength())))
           {
              //  this method will throw a ResizeError if needed
              updateLength(InputParameter.cast(integer.class, new integer(new integer(plus(piExtent, round(multiply(0.5, Array.this.getLength()), 0))))));
           }
  
           Array.this.setValue(poValue, piExtent);
           mSize.assign(int64.maximum(mSize, piExtent));
        }));
     }
  

The problem is: assign this-object:Value[piExtent] = poValue converts to Array.this.setValue(poValue, piExtent);
piExtent is integer so it goes on to call the method setValue again, creating a loop resulting in StackOverFlow error. To move on from this and make it call the actual setter I modified to:

Array.this.setValue(poValue, new int64(piExtent));

    method public Object GetValue(input piExtent as integer):
        Assert:IsPositive(piExtent, 'Index').

        if piExtent gt this-object:Size then
            undo, throw new AssertionFailedError(substitute('Requested index &1 is greater than array size &2',
                                                           piExtent, this-object:Size ), 0).

        return this-object:Value[piExtent].
    end method.

Saved to devsrv01 the modified ADE sources, with the def var statement instead of var and generics eliminated: /home/icp/modifiedADESources.zip

Investigating the issue described in #9488-144. This testcase recreates the issue:

class OpenEdge.Core.Collections.Array:
    define protected property Value as Object extent no-undo get. set.

    method public void SetValue(input poValue as Object):
        SetValue(poValue, 1).
    end method.

    method public void SetValue(input poValue as Object, input piExtent as integer):
        assign this-object:Value[piExtent] = poValue.
    end method.

    method public Object GetValue(input piExtent as integer):
        return this-object:Value[piExtent].
    end method.

end class.

Converts into:

   @LegacySignature(type = Type.METHOD, name = "SetValue", parameters = 
   {
      @LegacyParameter(name = "poValue", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT"),
      @LegacyParameter(name = "piExtent", type = "INTEGER", mode = "INPUT")
   })
   @LegacyResourceSupport(supportLvl = CVT_LVL_FULL | RT_LVL_FULL)
   public void setValue(final object<? extends _BaseObject_> _poValue, final integer _piExtent)
   {
      object<? extends _BaseObject_> poValue = TypeFactory.initInput(_poValue);
      integer piExtent = TypeFactory.initInput(_piExtent);

      internalProcedure(Array.class, this, "SetValue", new Block((Body) () -> 
      {
         Array.this.setValue(poValue, piExtent);
      }));
   }

     [java] !!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!
     [java] = [ASSIGN]:12884902253 @0:0
     [java]       (byrule=annotations/method_defs.rules:716)
     [java]    Value [VAR_CLASS]:12884902254 @0:0
     [java]          (property-access-type-byrule=annotations/method_defs.rules:716, direct-property-access=true, byrule=annotations/method_defs.rules:716, refid=12884901981, cls-property-byrule=annotations/method_defs.rules:716, cls-container=./abl/OpenEdge/Core/Collections/Array.cls, property-access-type=setter, extent=-1, static=false, cls-property=true, refid-byrule=annotations/method_defs.rules:716, static-byrule=annotations/method_defs.rules:716, direct-property-access-byrule=annotations/method_defs.rules:716, extent-byrule=annotations/method_defs.rules:354, cls-container-byrule=annotations/method_defs.rules:716)
     [java]    expression [EXPRESSION]:12884902258 @0:0
     [java]          (byrule=annotations/method_defs.rules:716)
     [java]       var [VAR_CLASS]:12884902259 @0:0
     [java]             (extent=-1, byrule=annotations/method_defs.rules:716, extent-byrule=annotations/method_defs.rules:365, refid=12884902241, refid-byrule=annotations/method_defs.rules:366)

OpenEdge.Core.DateArrayHolder implements IDateArrayHolder
OpenEdge.Core.Integer implements IInt64Holder
OpenEdge.Core.LogicalArrayHolder implements ILogicalArrayHolder
OpenEdge.Core.Memptr implements IMemptrHolder
OpenEdge.Core.String implements ILongcharHolder
OpenEdge.Core.DateTimeTzHolder implements IDateTimeTzHolder
OpenEdge.Core.LongcharArrayHolder implements ILongcharArrayHolder
OpenEdge.Core.IntegerArrayHolder implements IInt64ArrayHolder
OpenEdge.Core.DateTimeTzArrayHolder implements IDateTimeTzArrayHolder
OpenEdge.Core.HandleArrayHolder implements IHandleArrayHolder
OpenEdge.Core.WidgetHandle implements IHandleHolder
OpenEdge.Core.ObjectArrayHolder implements IObjectArrayHolder
OpenEdge.Core.DateHolder implements IDateHolder
OpenEdge.Core.ByteBucket implements IMemptrHolder
OpenEdge.Core.DateTimeArrayHolder implements IDateTimeArrayHolder
OpenEdge.Core.DateTimeHolder implements IDateTimeHolder
OpenEdge.Core.DecimalArrayHolder implements IDecimalArrayHolder
OpenEdge.Core.EncodedString implements ILongcharHolder
OpenEdge.Core.RowidArrayHolder implements IRowidArrayHolder
OpenEdge.Core.MemptrArrayHolder implements IMemptrArrayHolder
OpenEdge.Core.Decimal implements IDecimalHolder
OpenEdge.Core.StringArray mplements ILongcharArrayHolder
OpenEdge.Core.LogicalValue implements ILogicalHolder
OpenEdge.Core.LogicalArray implements ILogicalArrayHolder 

• Method naming issue:
  Example: In DateArrayHolder implements IdateArrayHolder method name getValue_1 while in IdateArrayHolder is getValue. This generates this compilation error:Class 'DateArrayHolder' must either be declared abstract or implement abstract method 'getValue(int64)' in 'IdateArrayHolder'
  This issue is seen in DateArrayHolder, DateHolder, DateTimeArrayHolder, DateTimeHolder, DateTimeTzArrayHolder, DateTimeTzHolder, DecimalArrayHolder, HandleArrayHolder, IntegerArrayHolder, LogicalArrayHolder, LongcharArrayHolder, MemptrArrayHolder, ObjectArrayHolder, RowidArrayHolder.
  At parsing I have some: Builtin class com.goldencode.p2j.oo.common.support.IdateTimeArrayHolder is not implemented in FWD!, even tho these are in the conversion list. I am not sure if this message should appear, maybe it's related.

• The compareTo issue we decided to fix by hand
• Another issue in LongcharArrayHolder: LongcharArrayHolder.this.bulkSetValue(new longchar(pcValue[0])); gives this error: Cannot resolve constructor 'longchar(character[])'. The original code:
  

         @param longchar[] The Value array */
      constructor public LongcharArrayHolder (input pcValue as longchar extent):
          this-object().
          assign this-object:Value = pcValue.
      end constructor.
  

Committed in 9488a rev. 15865 a possible solution for the naming issue described in #9488-147. It solves all the cases except in LongcharArrayHolder. The ILongcharHolder is already implemented in FWD and it was not added for conversion, but I am not exactly sure what is going on in that case.

ClassDefinition.loadBuiltinDefinition calls addJavaMethodName(javaname, "", legacySig); with javaname=bulkGetValue and legacySig=get-value(). But when trying to match the method it searches in the map for legacy-sig = bulk-get-value().
I think in ClassDefinition.loadBuiltinDefinition, we need to add some checks to determine if it is a bulk call and add the prefix "bulk-".

               case UNSUBSCRIBE:
                  mname = lsig.type().toString() + "-" + mname;
                  break;

Committed in 9488a rev. 15866: When loading builtin classes, correctly resolve setter and getter names. In ClassDefinition.loadBuiltinDefinition the following setter and getter names were not accounted for: bulk-get, bulk-set, set-all.

Now all method naming issues are solved.

I still have the issue in LongcharArrayHolder with the bulkSetValue call. I tried applying the fix from 9457c rev. 15871, as Constantin advised but this does not solve it unfortunately. Looking into this now.

Ioana, I've created #9923 for the bulkSetValue(new longchar problem - for now, please fix this 'by hand' by changing the generated Java code, to create a character[], copy the pcValue argument there, and pass that as the RVALUE. Thanks.

I did all the necessary modifications and saved the latest converted code at devsrv01:/home/icp/ADE_latest_converted.zip.
This archive includes the ADE converted code, the IHolders and the implementation for the holders.

Ioana, please go through all the changes in 9488a and group them by the common 'feature'; so, file x, y, z fix this problem, these a,b,c fix this problem, etc. I need to see what we want to bring into trunk. Where possible, show any existing redmine task and possible tests for the change.

Classes that implement com.goldencode.p2j.oo.core.collections.Icollection(OpenEdge.Core.Collections.ICollection):

• com.goldencode.p2j.oo.core.collections.AbstractTtCollection - FULL(toTable implemented)
• com.goldencode.p2j.oo.core.collections.Array - STUB now but will be ADE converted and contains ToTable implementation
• com.goldencode.p2j.oo.core.collections.LegacyList - STUB now but will be ADE converted and contains ToTable implementation
• com.goldencode.p2j.oo.core.collections.MapBackedCollection-PARTIAL(toTable also PARTIAL support), tests at /testcases/tests/oo/openedge/core/collections/map_backed_collection/TestMethodToTable.cls - it seem to check the core functionality of the method, but not the values in the returned table.
• com.goldencode.p2j.oo.core.collections.EntrySet extends MapBackedCollection
• com.goldencode.p2j.oo.core.collections.KeySet extends MapBackedCollection
• com.goldencode.p2j.oo.core.collections.ValueCollection extends MapBackedCollection
• com.goldencode.p2j.oo.core.collections.LegacyCollection extends AbstractTtcollection - Reported FULL but toTable is STUB, tests at /testcases/tests/oo/openedge/core/collections/collection/ but none for toTable method
• com.goldencode.p2j.oo.core.collections.LegacySet extends LegacyCollection
• com.goldencode.p2j.oo.net.http.CookieCollection extends LegacyCollection
• com.goldencode.p2j.oo.core.collections.StringCollection extends LegacyCollection
• com.goldencode.p2j.oo.core.collections.Ilist - interface
• com.goldencode.p2j.oo.core.collections.Iset - interface
• com.goldencode.p2j.oo.core.collections.IstringCollection - interface

So for full support for the toTable function, implementation in MapBackedCollection needs to be completed and in LegacyCollection needs to be fully implemented.

Went thru the commits in 9488a. Here is a summary:

Fixes for converting legacy oo¶

• rev. 15840
  - src/com/goldencode/p2j/convert/NameConverter.java - for package name
  - rules/convert/method_definitions.rules - for LegacyResourceSupport
  - rules/convert/base_structure.xml - for LegacyResource
• rev. 15842
  - rules/convert/base_structure.xml - for LegacyResourceSupport at class level
• rev. 15855
  - rules/convert/base_structure.xml - Fix NPE when p2j.cfg.xml config is not present

Supporting Progress v12.8¶

Generics¶

• rev. 15848
  - src/com/goldencode/p2j/uast/progress.g - user_defined_type_name

• rev. 15852
  - possible start for a solution of converting generics in implements clause
  - rules/convert/base_structure.xml - emitting java implements with generic
  - src/com/goldencode/p2j/uast/progress.g - parsing and annotating.
• rev. 15858
  - excluded generics from implements clause, now they are ignored.

Compound assignment operators (*=, +=, -=, /=)¶

• rev. 15849
  - Parser support src/com/goldencode/p2j/uast/progress.g
• rev. 15851
  - Full conversion support for compound operators
  - rules/annotations/assignment_style_stmt_rewriting.rules
  - src/com/goldencode/p2j/uast/progress.g
• rev. 15856
  - Fix regression coming from compound operators changes
  - src/com/goldencode/p2j/uast/progress.g

Support for final property¶

• rev. 15855
  - Conversion support for final property
  - src/com/goldencode/p2j/uast/progress.g
  - rules/convert/variable_definitions.rules
• rev. 15858
  - Added FINAL keyword in define_stmt_pre_scan - src/com/goldencode/p2j/uast/progress.g

Support for ClientPricipal:TENANT-ID, ClientPricipal:TENANT-NAME¶

• rev. 15857
  - Added full conversion support and stub runtime for ClientPricipal:TENANT-ID, ClientPricipal:TENANT-NAME.
  - rules/convert/methods_attributes.rules
  - rules/gaps/expressions.rules
  - src/com/goldencode/p2j/uast/progress.g
  - src/com/goldencode/p2j/util/ClientPrincipal.java
  - src/com/goldencode/p2j/util/ClientPrincipalResource.java
  Testcase:
  

  define variable client as handle no-undo.
  define variable tokenValue as character no-undo.
  define variable domainAccessCode as character no-undo initial "MyAccessCode".
  
  create client-principal client.
  
  client:user-id = "MyUser".
  client:domain-name = "MyDomain".
  client:session-id = "MySessionID".
  
  client:seal(domainAccessCode).
  
  assign tokenValue = string(client:tenant-id).
  message "Tenant ID:" tokenValue view-as alert-box.
  
  assign tokenValue = client:tenant-name.
  message "Tenant Name:" tokenValue view-as alert-box.
  
  delete object client no-error.
  

Support for builtin HASH-CODE function (already in 9436b)¶

• rev. 15850
  - Basic support for HASH-CODE. Fixed and in 9436c.
  - rules/convert/builtin_functions.rules
  - rules/gaps/expressions.rules
  - src/com/goldencode/p2j/convert/SignatureHelper.java
  - src/com/goldencode/p2j/uast/progress.g
  - src/com/goldencode/p2j/util/BaseDataType.java

Bugs fixed¶

• rev. 15861 - Fixes needed for annotateMethodCall failure:
  - Account for KW_PK_PRIV: src/com/goldencode/p2j/uast/ClassDefinition.java, src/com/goldencode/p2j/uast/SymbolResolver.java
  - Correctly resolve extent in EXPRESSION nodes: src/com/goldencode/p2j/uast/SymbolResolver.java
  - details in #9488-125, #9488-126, #9488-127
• rev. 15862 - When an unknown literal is used as a parameter for methods with the annotation property-access-type=resize, assign type integer.
  - rules/convert/literals.rules
  - details in #9488-131, #9488-132
• rev. 15864 - Relaxed restriction in wrapper detection rule to allow built-in OO method calls to be wrapped appropriately when parameter type requires casting.
  - rules/convert/base_structure.xml
  - details in #9488-131, #9488-134, #9488-136
• rev. 15865 - Fixed naming issues for builtin override properties. Corrected ClassDefinition.registerPropertyMethod for the case when the property is builtin but also override.
  - src/com/goldencode/p2j/uast/ClassDefinition.java
  - details in #9488-147, #9488-148
• rev. 15866 - When loading builtin classes, correctly resolve setter and getter names: ClassDefinition.loadBuiltinDefinition
  - src/com/goldencode/p2j/uast/ClassDefinition.java
  - details in #9488-147, #9488-148, #9488-149

Converted STUBS added - these already reaches trunk or will with 9436.¶

• rev. 15843 - Added lang.Stop, collections.Ihashable, collections.Icomparable, collections.Icomparer, collections.IequalityComparer.
• rev. 15844 - Added collections.KeyValuePair.
• rev. 15845 - Added Icollection, Iiterable, Iiterator, Ilist, Imap, Iset, List, SortedSet, HashMap.
• rev. 15846 - Modified some stubs to fix override issues.
• rev. 15847 - Added LockConflict, StopAfter, StopError, UserInterrupt and made Stop not final.
• rev. 15858 - Added converted skeletons for Progress.IO.ByteOrder and Progress.IO.MemoryOutputStream.

MISC¶

• rev. 15841 (Original author CA) - Fix from constantin to bypass some annotation errors by reassigning it to n/a instead of null.
  - src/com/goldencode/ast/AnnotatedAst.java
• rev. 15860(Original author CA) - First pass to allow a skeleton which already exists in p2j.oo to re-use the names from p2j.oo instead of creating new conversion names.
• rev. 15853 - Start for VAR statement: src/com/goldencode/p2j/uast/progress.g
• rev. 15854 - Reverted VAR changes src/com/goldencode/p2j/uast/progress.g
• rev. 15863 (Original author CA)- The Java method name associated with a constructor must be resolved from the ClassDefinition, during conversion. Useful for converting skeleton classes.

----------¶

• all fixed bugs
• from supporting progress v12: Compound assignment operators, Support for final property and Support for ClientPricipal:TENANT-ID, ClientPricipal:TENANT-NAME. The generic change were more so I did not have parsinf errors. For full conversion generics support there is a lot more work to be done.
• maybe the fixes for converting legacy oo, as these don t impact the conversion if the specific parameters are not set.

9488a was rebased from trunk rev 15875 - new rev 15902.

• Support *=, -=, +=, /= operator syntax:
  • rules/annotations/assignment_style_stmt_rewriting.rules
  • src/com/goldencode/p2j/uast/progress.g
• final modifier for class property
  • rules/convert/variable_definitions.rules
  • src/com/goldencode/p2j/uast/progress.g
• ClientPricipal:TENANT-ID and ClientPricipal:TENANT-NAME
  • src/com/goldencode/p2j/uast/progress.g
  • rules/convert/methods_attributes.rules
  • rules/gaps/expressions.rules
  • src/com/goldencode/p2j/util/ClientPrincipal.java
  • src/com/goldencode/p2j/util/ClientPrincipalResource.java
  • FIX: change SignatureHelper for these methods
  • FIX: ClientPrincipal - use UnimplementedFeature.missing to mark that is not implemented
  • FIX: create a task and assign it to Ovidiu for implementing the runtime for this feature (note at the task that the conversion changes are already made)
• support for package-private access modifier
  • src/com/goldencode/p2j/uast/SymbolResolver.java
  • src/com/goldencode/p2j/uast/ClassDefinition.java
• move the changes to 9436b
  • src/com/goldencode/p2j/oo/collections/Icomparable.java
• FWD Bugs:
  • resize needs an integer argument for the index:
    • rules/convert/literals.rules
  • read extent from an EXPRESSION node
    • src/com/goldencode/p2j/uast/SymbolResolver.java
  • fixed the names for the extent property Java methods when loading a builtin .cls from FWD p2j.oo
    • src/com/goldencode/p2j/uast/ClassDefinition.java

• automatically compute member names (ctors, methods, etc) from builtin OO legacy classes in FWD, when converting:
  • rules/annotations/method_defs.rules
  • src/com/goldencode/p2j/uast/ClassDefinition.java
• added is-legacy-oo=true p2j.cfg.xml config parameter, used when converting ADE sources
  • rules/annotations/naming.rules
  • rules/convert/base_structure.xml
  • rules/convert/method_definitions.rules
  • src/com/goldencode/p2j/convert/NameConverter.java
  • src/com/goldencode/p2j/uast/SymbolResolver.java
• generics support
  • src/com/goldencode/p2j/uast/progress.g
• helpers from 9457c - will not reach trunk at this time
  • src/com/goldencode/ast/AnnotatedAst.java
  • src/com/goldencode/p2j/uast/AstGenerator.java
• hash-code changes which are already in 9436b:
  • src/com/goldencode/p2j/uast/progress.g
  • rules/convert/builtin_functions.rules
  • rules/gaps/expressions.rules
  • src/com/goldencode/p2j/convert/SignatureHelper.java
  • src/com/goldencode/p2j/util/BaseDataType.java

For details on how we properly document the licensing of the ADE code, see Inclusion of Apache 2 Licensed ADE Code.

I have tried to convert the ADE source code with the latest revision of 9488a, however conversion of multiple files fails with:

     [java] Caused by: java.lang.StringIndexOutOfBoundsException: begin 41, end 28, length 32
     [java]     at java.base/java.lang.String.checkBoundsBeginEnd(String.java:4606)
     [java]     at java.base/java.lang.String.substring(String.java:2709)
     [java]     at com.goldencode.p2j.uast.SymbolResolver.createClassMappings(SymbolResolver.java:2531)
     [java]     at com.goldencode.p2j.uast.SymbolResolver.initPossibleClasses(SymbolResolver.java:9160)
     [java]     at com.goldencode.p2j.uast.SymbolResolver.<init>(SymbolResolver.java:1250)
     [java]     at com.goldencode.p2j.uast.AstGenerator.parse(AstGenerator.java:1615)
     [java]     at com.goldencode.p2j.uast.AstGenerator.processFile(AstGenerator.java:1055)

I have done some investigation and the root cause seems to be related to #7255. In that task, SymbolResolver.getClassFileList has been modified to call getCanonicalPath instead of getAbsolutePath. The return value of the getCanonicalPath call is used as a key in a map. The problematic difference between these methods here is when it is called for .. getCanonicalPath will return the actual absolute path, however getAbsolutePath returned [the absolute path]/.. So, now, no matter if we initially have the path as an absolute path or ., it will use the same values from the map. This is wrong as the values from the map are relative to the initial path.

I will try to understand the minimal requirements to reproduce this issue, then if it also reproduces with trunk, will open a new task and fix that.

I found a fix for the annotateMethodCall error from above. In ClassDefinition.isInheritedFrom we have logic for checking both the implemented interfaces and the extended classes. The problem occurs for the recursive calls for the interfaces. Only the implemented interfaces from the inheritance tree would be checked. So, if we have this relation:
(class A (implements)-> interface I1 (extends)-> interface I2), a call to check whether class A is inherited from interface I2 would have returned false. As a fix, I have made it call isInheritedFrom in ClassDefinition.isInheritedFrom for all the inherited interfaces. Up until now, it was calling isImplemented. However, this method only checks the interfaces that are implemented by this one, however it doesn't check the classes that are extended.

1. 15903 - Contains some patches from trunk regarding propath processing
2. 15903 - Solution for the annotateMethodCall problem

Next, I will focus on the other problem exposed in #9488-164.

I think I understand the root cause for this problem: WARNING: more than one method def found using fuzzy lookup: GetIterator from class Progress.Collections.IList. I get this error when processing ./abl/OpenEdge/Core/Collections/List.cls for this file Progress.Collections.IList. It's important to note that the package name starts with Progress, as there is also an OpenEdge version. In the skeleton version for this class, it inherits from Progress.Collections.ICollection and Progress.Collections.IIterable. Progress.Collections.ICollection also inherits from Progress.Collections.IIterable. In Progress.Collections.IIterable, method GetIterator is defined. Therefore, in ClassDefinition.candidates it gets to have the same method twice(once coming directly from Progress.Collections.IIterable and another one from Progress.Collections.ICollection), even though it actually represents the same class method. I will try to refactor that code to use a set instead of the list in order to avoid this duplication.

I fixed the problem from #9488-166 in 9488a/ rev. 15905. I have made ClassDefinition.candidates return a Set of MethodData instances. Up until now, it was returning a list of MatchMetrics, however, I don't think any other data was used up until ClassDefinition.candidates. I am creating back the List of MatchMetrics in FuzzyMethodLookup.fuzzyMethodLookup based on the unique method candidates.

I have committed 9488/ rev. 15907 for adding the overloaded methods indexOf in TextOps.
After that, I investigated class Memptr, where there are 2 methods with the same signature, one for this explicit getter:

    method public int64 GetPointerValue ():
        return this-object:PointerValue.
    end method.

        get():
            return get-pointer-value(mmValue).
        end get.

However, in a different class only with these constructs, the logic for avoiding the naming collision works, the problem is that GetPointerValue is a method from an implemented interface, so the name wasn't checked for collisions. Currently, I have this quick fix, which works for the ADE source, however, I would want to improve it before committing:

=== modified file 'src/com/goldencode/p2j/uast/ClassDefinition.java'
--- old/src/com/goldencode/p2j/uast/ClassDefinition.java        2025-09-30 07:55:55 +0000
+++ new/src/com/goldencode/p2j/uast/ClassDefinition.java        2025-10-02 14:32:31 +0000
@@ -879,6 +879,10 @@
             }
             else if (override && (javaname = lookupOverrideName(legacySig, override, true)) != null)
             {
+               if (javaMethodCollisions.contains(javaname + "(" + jparms + ")"))
+               {
+                  javaname = calculateUniqueJavaMethodName(accessor, jparms, legacySig);
+               }
                // first time in (for an override)
                addJavaMethodName(javaname, jparms, legacySig);

Other compiling issues¶

1.     [javac] /home/ap/gcd/ade/icp/icp/ADE_project/ADE/src/com/goldencode/p2j/oo/core/EventArgs.java:69: error: cannot find symbol
       [javac]          if (_not(EventArgs.empty.isValid()))
       [javac]                                  ^
       [javac]   symbol:   method isValid()
       [javac]   location: variable empty of type ContextLocal<object<? extends EventArgs>>
   

2.     [javac] /home/ap/gcd/ade/icp/icp/ADE_project/ADE/src/com/goldencode/p2j/oo/core/EventArgs.java:74: error: no suitable method found for storeReturnValue(ContextLocal<object<? extends EventArgs>>)
       [javac]          storeReturnValue(EventArgs.empty);
       [javac]          ^
       [javac]     method BlockManager.storeReturnValue(boolean) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to boolean)
       [javac]     method BlockManager.storeReturnValue(double) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to double)
       [javac]     method BlockManager.storeReturnValue(long) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to long)
       [javac]     method BlockManager.storeReturnValue(int) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to int)
       [javac]     method BlockManager.storeReturnValue(String) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to String)
       [javac]     method BlockManager.storeReturnValue(BaseDataType) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to BaseDataType)
       [javac]     method BlockManager.storeReturnValue(Stream) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to Stream)
       [javac]     method BlockManager.storeReturnValue(WrappedResource) is not applicable
       [javac]       (argument mismatch; ContextLocal<object<? extends EventArgs>> cannot be converted to WrappedResource)
   

3.     [javac] /home/ap/gcd/ade/icp/icp/ADE_project/ADE/src/com/goldencode/p2j/oo/core/LongcharArrayHolder.java:219: error: no suitable constructor found for longchar(character[])
       [javac]          LongcharArrayHolder.this.bulkSetValue(new longchar(pcValue[0]));
       [javac]                                                ^
       [javac]     constructor longchar.longchar(BaseDataType) is not applicable
       [javac]       (argument mismatch; character[] cannot be converted to BaseDataType)
       [javac]     constructor longchar.longchar(String) is not applicable
       [javac]       (argument mismatch; character[] cannot be converted to String)
       [javac]     constructor longchar.longchar(char) is not applicable
       [javac]       (argument mismatch; character[] cannot be converted to char)
       [javac]     constructor longchar.longchar(Text) is not applicable
       [javac]       (argument mismatch; character[] cannot be converted to Text)
   
   

4.     [javac] /home/ap/gcd/ade/icp/icp/ADE_project/ADE/src/com/goldencode/p2j/oo/core/util/TokenResolver.java:729: error: incompatible types: int64 cannot be converted to integer
       [javac]             silent(() -> tokenVal.assign(pError.ref().getMessage(integer.maximum(new integer(1), idx))));
   

The reason for the 1st and 2nd problem is that the converted code looks like this:

         if (_not(EventArgs.empty.isValid()))
         {
            EventArgs.setEmpty(new object(ObjectOps.newInstance(EventArgs.class)));
         }

         storeReturnValue(EventArgs.empty);

Calling get() on EventArgs.empty fixes both issues.
The 3rd problem has been discussed previously, according to #9488-151, the solution was to just manually modify the converted code. The 4th problem is that, in that case, integer.maximum returns an int64, and the result is passed as the argument to LegacyError.getMessage, which only accepts integer.

Right now, I don't have a concrete solution for these last problems, however, the good part is that, after manually modifying the converted code in order to fix these issues, it built successfully.

1. rev. 15908 - fixes the duplicate getter. I have used a similar approach as I previously posted in #9488-170. Basically, I have introduced the logic for checking name collisions for overridden methods. However, initially, I am only checking the methods in the current class, instead of the entire class hierarchy. The reason is that, after the 1st method is processed, in an upper class, the method is already marked for collisions. So, checking the entire class hierarchy, would have changed the method name even for the 1st method in the current class.
2. rev. 15909 - fixes the missing get for EventArgs.empty. The reason the getter was not added, is this the condition from oo_calls.rules was evaluated as false:
   

               <!-- static variable access must dereference a context local -->
               <rule>
                  ref2.isAnnotation("static")                         and
                  #(java.lang.Boolean) ref2.getAnnotation("static")   and
                  !ref2.isAnnotation("cls-property")                  and
                  !evalLib("is_java_cls", ref2)
                  <action>
                     lastid = createJavaAst(java.method_call, "get", closestPeerId)
                  </action>
                  <action on="false">lastid = closestPeerId</action>
               </rule>
   

   
   The reason is !ref2.isAnnotation("cls-property") was false. To be honest, I don't quite understand why this check is needed. However, I didn't either find out what cls-property really means. I would assume it means it is a data member in a class. And, if my assumption is correct, I don't understand, why it needs NOT to be a class member in order to generate that getter. So, for now, I have deleted that condition from the rule and it is working fine.
3. rev. 15910 - fixes the 4th problem from the compiling issues in #9488-170. The solution I have committed is to just have an implementation of the maximum method in integer class which returns an integer instance. IMHO, this is a correct solution, as up until now, it was just going up to the base implementation from int64, so it would have always returned an int64 instance. However, Ioana also encountered a similar issue, and she proposed this solution in #9488-131, however it wasn't implemnted. As far as I understand, it was solved by wrapping the return values into an integer when it was called. However, I don't really see the advantage for this conversion solution compared to just adding the method in integer class. However, I am open to make the conversion change if it is really needed.

With the last revision, the code converts and compiles successfully except for LongcharArrayHolder.this.bulkSetValue(new longchar(pcValue[0])); issue. However, as discussed previously(#9488-151), the solution is just to modify the converted code. The conversion list used is in #9488-168.

Andrei, in this pass, I'd like to have only unimplemented files in FWD (RT_LVL_STUB) in a branch.

From #10522, please also add OpenEdge.Core.Collections.StringCollection (and dependencies) plus check the changes from OpenEdge.Core.Collections.AbstractTTCollection in #10522 if they make sense.

• files which have also a history list from our the hand-written .cls file - should we keep that history?
• some interfaces are pretty simple (like in IdecimalHolder.java) - should we keep our version or the converted one?
• AbstractTtcollection - this has a significant hand-written implementation, but it still may be buggy (see the #10522 issues); I think we can use the converted one, but we need to test the projects using it.
• Array - this again has a significant hand-written implementation, without using temp-tables. I think we need to use our implementation and just double-check if our impl matches the logic from the 4GL Array.cls code.

Artur, please create a separate branch and place there all p2j/oo .java files, without any DMO files or other artifacts (name_map.xml, etc).

• files which have also a history list from our the hand-written .cls file - should we keep that history?

We keep the history if we have manual changes in the file. If the file is just converted, then our old history entries don't apply.

some interfaces are pretty simple (like in IdecimalHolder.java) - should we keep our version or the converted one?

Unless there is some value in the existing implementation, I would expect to use the converted one.

AbstractTtcollection - this has a significant hand-written implementation, but it still may be buggy (see the #10522 issues); I think we can use the converted one, but we need to test the projects using it.

Is this a case where our existing implementation eliminated the temp-table usage? If so we might want to keep it. If there is no improvement, then use the converted version.

• Array - this again has a significant hand-written implementation, without using temp-tables. I think we need to use our implementation and just double-check if our impl matches the logic from the 4GL Array.cls code.

Agreed

Greg Shah wrote:

• files which have also a history list from our the hand-written .cls file - should we keep that history?

We keep the history if we have manual changes in the file. If the file is just converted, then our old history entries don't apply.

Well, there were some changes which are brought automatically now via conversion while in the past were added manually.

some interfaces are pretty simple (like in IdecimalHolder.java) - should we keep our version or the converted one?

Unless there is some value in the existing implementation, I would expect to use the converted one.

Understood.

AbstractTtcollection - this has a significant hand-written implementation, but it still may be buggy (see the #10522 issues); I think we can use the converted one, but we need to test the projects using it.

Is this a case where our existing implementation eliminated the temp-table usage? If so we might want to keep it. If there is no improvement, then use the converted version.

Thanks, we'll move to the converted variant. There is no temp-table usage in this.

• Array - this again has a significant hand-written implementation, without using temp-tables. I think we need to use our implementation and just double-check if our impl matches the logic from the 4GL Array.cls code.

Agreed

Understood. As a general idea - I think anything which has temp-tables for these cases, we will rewrite to not use them.

As a general idea - I think anything which has temp-tables for these cases, we will rewrite to not use them.

Yes, exactly right. Temp-tables should not be the basis for non-database data structures like collections or arrays.

Andrei, please create a separate branch and place there all p2j/oo .java files, without any DMO files or other artifacts (name_map.xml, etc).

Constantin Asofiei wrote:

Andrei, please create a separate branch and place there all p2j/oo .java files, without any DMO files or other artifacts (name_map.xml, etc).

Ok, I am on it.

1. Apart from the converted classes, I have also ported the change in integer.java from 9488a. The need for this change is explained in #9488-171
2. I have slightly modified the initial converted class TokenResolver because of the changed name of some getter/setter methods in ClientPrincipalResource. tenantId/tenantName have changed to getTenantId/getTenantName since i made the initial conversion.
3. I have added the converted Array as Array_conv and AbstractTTCollection as AbstractTTCollection_conv as these files need further inspection.

Please add the history number to integer.java

• use the converted AbstractTtCollection.java
• manually check the existing Array.java implementation with the 4GL Array.cls
• remove the two _conv files

1. replaced AbstractTtCollection with the converted version
2. Added the history number to integer

Regarding Array.java, it seems there are multiple missing methods in our current implementation. The protected methods, I think can be ignored as these methods can't be called from any converted code. However, there are also some public methods that are not present in our current implementation(setDefaultArraySize, addNull, add(final integer _pIndex, final object _pElement), ...). I think these should be added. There are also methods that have slightly different implementations, I will compare each of them and see if they can behave differently.

• In #10522 there are changes in StringCollection.java, AbstractTtcollection.java, LegacyList.java, LegacyListIterator.java
• All of them are also in 9488a. I would assume that all the functionality in the converted classes is also in 10522a, even though implementation per-se can be slightly different.

While checking the converted Array.java, I have across an issue that was also spotted by Ioana back when she was working on this task, but it seems it was never properly fixed(#9488-144). The problem is in setValue(final object _poValue, final integer _piExtent) method which has a recursive call:

   public void setValue(final object<? extends _BaseObject_> _poValue, final integer _piExtent)
   {
      object<? extends _BaseObject_> poValue = TypeFactory.initInput(_poValue);
      integer piExtent = TypeFactory.initInput(_piExtent);

      internalProcedure(Array.class, this, "SetValue", new Block((Body) () -> 
      {
         Assert.isPositive(piExtent, new character("Index"));

         if (_or(isUnknown(Array.this.getLength()), () -> isGreaterThan(piExtent, Array.this.getLength())))
         {
            //  this method will throw a ResizeError if needed
            updateLength(InputParameter.cast(integer.class, new integer(new integer(plus(piExtent, round(multiply(0.5, Array.this.getLength()), 0))))));
         }

         Array.this.setValue(poValue, piExtent);
         mSize.assign(int64.maximum(mSize, piExtent));
      }));
   }

This can be solved by wrapping piExtent to an int64. This way the call will resolve to another correct method. The same problem is also in getValue(final integer _piExtent), which has also a recursive call.

I am not sure if it is worth finding here the conversion solution, or we can just manually modify the converted file and defer this problem to another task?

Andrei, so you have a 4GL method with the same name (and signature) as a converted 4GL getter or setter. Please create a standalone test for this.

This testcase from #9488-146 reproduces the problem:

class OpenEdge.Core.Collections.Array:
    define protected property Value as Object extent no-undo get. set.

    method public void SetValue(input poValue as Object):
        SetValue(poValue, 1).
    end method.

    method public void SetValue(input poValue as Object, input piExtent as integer):
        assign this-object:Value[piExtent] = poValue.
    end method.

    method public Object GetValue(input piExtent as integer):
        return this-object:Value[piExtent].
    end method.

end class.

And it gets converted to ShowHide

public class ArrayRecursive
extends BaseObject
{
   @LegacySignature(type = Type.PROPERTY, extent = -1, name = "Value", dataType = "OBJECT", qualified = "progress.lang.object")
   private object<? extends _BaseObject_>[] value = TypeFactory.objectExtent(_BaseObject_.class);

   @LegacySignature(type = Type.CONSTRUCTOR, name = "Array_recursive")
   public void __arrayRecursive_constructor__()
   {
      internalProcedure(ArrayRecursive.class, this, "Array_recursive", new Block((Body) () -> 
      {
         __lang_BaseObject_constructor__();
      }));
   }

   @LegacySignature(type = Type.GETTER, name = "Value", returns = "OBJECT", qualified = "progress.lang.object", parameters = 
   {
      @LegacyParameter(name = "idx", type = "INT64", mode = "INPUT")
   })
   protected object<? extends _BaseObject_> getValue(final int64 _idx)
   {
      int64 idx = TypeFactory.initInput(_idx);

      return function(ArrayRecursive.class, this, "Value", object.class, new Block((Body) () -> 
      {
         storeReturnValue(subscript(value, idx));
      }));
   }

   @LegacySignature(type = Type.SETTER, name = "Value", parameters = 
   {
      @LegacyParameter(name = "var", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT"),
      @LegacyParameter(name = "idx", type = "INT64", mode = "INPUT")
   })
   protected void setValue(final object<? extends _BaseObject_> _var, final int64 _idx)
   {
      object<? extends _BaseObject_> var = TypeFactory.initInput(_var);
      int64 idx = TypeFactory.initInput(_idx);

      internalProcedure(ArrayRecursive.class, this, "Value", new Block((Body) () -> 
      {
         assignSingle(value, idx, var);
      }));
   }

   @LegacySignature(type = Type.GETTER, name = "Value", extent = -1, returns = "OBJECT", qualified = "progress.lang.object")
   protected object[] bulkGetValue()
   {
      return extentFunction(ArrayRecursive.class, this, "Value", object.class, new Block((Body) () -> 
      {
         storeExtentReturnValue(value);
      }));
   }

   @LegacySignature(type = Type.SETTER, name = "Value", parameters = 
   {
      @LegacyParameter(name = "var", type = "OBJECT", extent = -1, qualified = "progress.lang.object", mode = "INPUT")
   })
   protected void bulkSetValue(final object<? extends _BaseObject_>[] _var)
   {
      object<? extends _BaseObject_>[] var[] = new object[][]
      {
         TypeFactory.initInput(_var)
      };

      internalProcedure(ArrayRecursive.class, this, "Value", new Block((Body) () -> 
      {
         value = (object[]) assignMulti(value, var[0]);
      }));
   }

   @LegacySignature(type = Type.SETTER, name = "Value", parameters = 
   {
      @LegacyParameter(name = "var", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT")
   })
   protected void setAllValue(final object<? extends _BaseObject_> _var)
   {
      object<? extends _BaseObject_> var = TypeFactory.initInput(_var);

      internalProcedure(ArrayRecursive.class, this, "Value", new Block((Body) () -> 
      {
         assignMulti(value, var);
      }));
   }

   @LegacySignature(type = Type.LENGTH, name = "Value", returns = "INTEGER")
   protected integer lengthOfValue()
   {
      return function(ArrayRecursive.class, this, "length-of-Value", integer.class, new Block((Body) () -> 
      {
         storeReturnValue(ArrayAssigner.lengthOf(value));
      }));
   }

   @LegacySignature(type = Type.RESIZE, name = "Value", parameters = 
   {
      @LegacyParameter(name = "size", type = "INT64", mode = "INPUT")
   })
   protected void resizeValue(final int64 _size)
   {
      int64 size = TypeFactory.initInput(_size);

      internalProcedure(ArrayRecursive.class, this, "resize-Value", new Block((Body) () -> 
      {
         value = ArrayAssigner.resize(value, size);
      }));
   }

   @LegacySignature(type = Type.METHOD, name = "SetValue", parameters = 
   {
      @LegacyParameter(name = "poValue", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT")
   })
   public void setValue(final object<? extends _BaseObject_> _poValue)
   {
      object<? extends _BaseObject_> poValue = TypeFactory.initInput(_poValue);

      internalProcedure(ArrayRecursive.class, this, "SetValue", new Block((Body) () -> 
      {
         setValue(poValue, new integer(1));
      }));
   }

   @LegacySignature(type = Type.METHOD, name = "SetValue", parameters = 
   {
      @LegacyParameter(name = "poValue", type = "OBJECT", qualified = "progress.lang.object", mode = "INPUT"),
      @LegacyParameter(name = "piExtent", type = "INTEGER", mode = "INPUT")
   })
   public void setValue(final object<? extends _BaseObject_> _poValue, final integer _piExtent)
   {
      object<? extends _BaseObject_> poValue = TypeFactory.initInput(_poValue);
      integer piExtent = TypeFactory.initInput(_piExtent);

      internalProcedure(ArrayRecursive.class, this, "SetValue", new Block((Body) () -> 
      {
         ArrayRecursive.this.setValue(poValue, piExtent);
      }));
   }

   @LegacySignature(type = Type.METHOD, name = "GetValue", returns = "OBJECT", qualified = "progress.lang.object", parameters = 
   {
      @LegacyParameter(name = "piExtent", type = "INTEGER", mode = "INPUT")
   })
   public object<? extends com.goldencode.p2j.oo.lang._BaseObject_> getValue(final integer _piExtent)
   {
      integer piExtent = TypeFactory.initInput(_piExtent);

      return function(ArrayRecursive.class, this, "GetValue", object.class, new Block((Body) () -> 
      {
         storeReturnValue(ArrayRecursive.this.getValue(piExtent));
      }));
   }
}

I have a question, are the methods for setting a value at a position in an extent always having int64 as the type for the index parameter? If so, it seems that something like this solves the setter problem by wrapping the parameter into int64:

=== modified file 'rules/annotations/oo_references.rules'
--- old/rules/annotations/oo_references.rules   2025-10-01 13:25:42 +0000
+++ new/rules/annotations/oo_references.rules   2026-01-16 10:43:56 +0000
@@ -818,8 +818,11 @@
                   <action>idxref = subref.getChildAt(0)</action>
                   <action>subref.remove()</action>
                   <action>eref = createProgressAst(prog.expression, copy, 1)</action>
-                  <action>eref.putAnnotation("classname", "integer")</action>
+                  <action>eref.putAnnotation("classname", "int64")</action>
                   <action>eref.putAnnotation("wrap", true)</action>
+                  <action>eref.putAnnotation("wrap_parameter", true)</action>
+
+                  
                   <rule>idxref.type == prog.expression
                      <action>idxref = idxref.getChildAt(0)</action>
                   </rule>

Andrei, we don't need to wrap it. This needs to be solved in ClassDefinition.hasCollision - that currently uses exact signature match, with no concept of overloading.

Constantin Asofiei wrote:

Andrei, we don't need to wrap it. This needs to be solved in ClassDefinition.hasCollision - that currently uses exact signature match, with no concept of overloading.

Actually, you are right - the fix needs to be where we emit the getter/setter call, and only if the index parameter doesn't match int64.

This patch seems to solve the issue both for getters and setters: ShowHide

=== modified file 'rules/annotations/oo_references.rules'
--- old/rules/annotations/oo_references.rules   2025-10-01 13:25:42 +0000
+++ new/rules/annotations/oo_references.rules   2026-01-16 11:59:29 +0000
@@ -818,11 +818,13 @@
                   <action>idxref = subref.getChildAt(0)</action>
                   <action>subref.remove()</action>
                   <action>eref = createProgressAst(prog.expression, copy, 1)</action>
-                  <action>eref.putAnnotation("classname", "integer")</action>
-                  <action>eref.putAnnotation("wrap", true)</action>
                   <rule>idxref.type == prog.expression
                      <action>idxref = idxref.getChildAt(0)</action>
                   </rule>
+                  <rule>idxref.type != prog.var_int64
+                     <action>eref.putAnnotation("classname", "int64")</action>
+                     <action>eref.putAnnotation("wrap_parameter", true)</action>
+                  </rule>
                   <action>idxref.move(eref, 1)</action>
                </rule>
                <action>execLib("rewrite_prop_to_method", copy)</action>
@@ -941,6 +943,10 @@
                   <action>idxref = subref.getChildAt(0)</action>
                   <action>subref.remove()</action>
                   <action>idxref.move(copy, 1)</action>
+                 <rule>idxref.type != prog.var_int64
+                    <action>idxref.putAnnotation("classname", "int64")</action>
+                    <action>idxref.putAnnotation("wrap_parameter", true)</action>
+                 </rule>
                </rule>
                <action>execLib("rewrite_prop_to_method", copy)</action>
                <action>execLib("wrap_argument", copy)</action>

However, I am a little confused about the wrap annotation that was already being set. I had to remove it and only set wrap_parameter. If I kept both, literals would have been wrapped 2 times: assign this-object:Value[2] = poValue. -> ArrayRecursive.this.setValue(poValue, new int64(new int64(2)));

Andrei, lets get this in another branch, as I want the p2j.oo in trunk soon.

Otherwise, how is Array.java looking when compared with Array.cls? Do we need to change anything?

Constantin Asofiei wrote:

Andrei, lets get this in another branch, as I want the p2j.oo in trunk soon.

Ok, got it.

Otherwise, how is Array.java looking when compared with Array.cls? Do we need to change anything?

• indexOf_
• lastIndexOf_
• public object remove(final integer _pIndex)
• public object set(final integer _pIndex, final object _pElement)
• subList
• addArray(final integer _pIndex, final object[] _pArray)

Also, some methods like setSize seem to have a slightly different behaviour regarding handling error cases. I will come up with the whole list of differences soon and will try to implement those.

The missing methods are not part of the skeleton Array.cls - so if you add them, you need to add them to the skeleton, too. And also check if the methods are overrides from some interface.

Constantin Asofiei wrote:

And also check if the methods are overrides from some interface.

Hmm, it seems that these new methods are defined in IList interface, which is implemented by the new Array class. However, the old one implements Icollection.
Apart from adding the new methods in our Array class, should I also change it to implement Ilist?

PS: Ilist interface actually extends Icollection. So, the inheritance tree won't change by much.

I think we have the wrong hierarchy in FWD. For 11.7, Array implements IList, which in turn implements ICollection.

Please check github for both 11.7 and 12.8 at https://github.com/progress/ADE/ and lets fix the hierarchy and the interfaces (including in skeleton .cls),

Andrei Plugaru wrote:

Currently, one thing that still needs to be done is regarding Arra.toTable. Right now we only have a stub for this method, however I think it is a good moment to have the implementation.

Yes, implement that.

1. get the converted DMO used in this method(Ttarray_1) and add it to the branch
2. create the temp table dynamically, similar to how it is done in List.ToTable:
   

           create temp-table tt.
           tt:add-new-field('sequence', 'integer').
           tt:add-new-field('objectref', 'Progress.Lang.Object').
           tt:add-new-index('objidx').
           tt:add-index-field('objidx', 'objectref').
           tt:add-new-index('seq', yes, yes).
           tt:add-index-field('seq', 'sequence').
           tt:temp-table-prepare('ttList').
   

I would prefer the 2nd option as we don't have to add another DMO interface.

The temp-table will be dynamic, via create temp-table. Just make sure its definition is right (fields/types and indexes, temp-table name).

Constantin, I have committed 9488b/ rev. 16344 with the proper implementation for toTable in Array.
Please review!

Andrei, please fix the javadoc formatting at the new methods you've added - I know most methods are missing javadoc, but if you add it, make sure is following the standard.

Also, please update the copyright year on the file (and any other files which require it).

Otherwise, did you test toTable and the other methods in FWD?

Constantin Asofiei wrote:

Andrei, please fix the javadoc formatting at the new methods you've added - I know most methods are missing javadoc, but if you add it, make sure is following the standard.

Also, please update the copyright year on the file (and any other files which require it).

Ok, will do.

Otherwise, did you test toTable and the other methods in FWD?

No, I haven't. Is there a suite of tests specifically for these classes? Or how exactly should they be tested?

Andrei Plugaru wrote:

No, I haven't. Is there a suite of tests specifically for these classes? Or how exactly should they be tested?

See the new testcases project, the tests/oo/openedge/core/collections/array/ folder.

1. test testMethodToTableArray fails. At a first glance, it seems it is regarding the error management.
2. testPropertyValueArray fails because of a conversion error:
   

       /* Default constructor */
       constructor public Array():
           this-object(Array:DefaultArraySize).
           // we don't have any elements yet
           assign mSize = 0.
       end constructor.
   

   
   Got converted to
   

      public void __core_collections___core_collections_Array_constructor__static___constructor__()
      {
   ...
      }
   

So, my question is wouldn't be better to move forward to the converted version or should I try to completely align our implementation to the converted one.
To be noted that the converted version doesn't use a temp table as the backing structure, it is an array.

Andrei Plugaru wrote:

I have tested the current Array class using the tests from testcases and unfortunately, there are a few dozen failures. There are quite a few regarding missing errors/warnings.

Is this from your implementation or from trunk?

To be noted that the converted version doesn't use a temp table as the backing structure, it is an array.

What are you converting? See https://github.com/progress/ADE/blob/release-12.8.x/corelib/OpenEdge/Core/Collections/Array.cls - it has a ttArray temp-table.

Constantin Asofiei wrote:

Andrei Plugaru wrote:

I have tested the current Array class using the tests from testcases and unfortunately, there are a few dozen failures. There are quite a few regarding missing errors/warnings.

Is this from your implementation or from trunk?

Here I mean the one from 9488b, which is based on the one from trunk + the new methods + toTable method.

To be noted that the converted version doesn't use a temp table as the backing structure, it is an array.

What are you converting? See https://github.com/progress/ADE/blob/release-12.8.x/corelib/OpenEdge/Core/Collections/Array.cls - it has a ttArray temp-table.

In the original OE code ttArray is only used for toTable method. However, I dropped it as we used the dynamic temp table for that. ttArray is not used as the backing structure. The backing structure is an extent.

Andrei Plugaru wrote:

In the original OE code ttArray is only used for toTable method. However, I dropped it as we used the dynamic temp table for that. ttArray is not used as the backing structure. The backing structure is an extent.

Hmm... lets do this: get the converted Array.java and change it after in FWD, to get rid of the DMO and toTable to use a dynamic temp-table.

Constantin Asofiei wrote:

Hmm... lets do this: get the converted Array.java and change it after in FWD, to get rid of the DMO and toTable to use a dynamic temp-table.

Well, I already done this :)) This is what I was referring to in:

However, I have also tested the converted version of the Array(with the custom toTable) and it passes most of the tests. There are 2 issues:

In 9488b/16345 I have committed the converted Array class with the changes about the DMO and toTable method + the new header for the file.
The only thing left to do from my POV is to do the cosmetic changes about the Java doc and copyright year.

Constantin, about the Javadocs for the methods: it seems that in the original OE code most of the methods have some documentation, however they are either lost or put in completely wrong places. I think if we really want to have the right and complete Javadocs, I should go over the original OE code, copy the documentation from there and then adapt it to match our standard.

Do you think it is worth doing that?

Andrei Plugaru wrote:

Do you think it is worth doing that?

Andrei, when I mentioned the javadocs, I meant what you manually wrote - not to add/fix 4GL javadocs; sorry for the confusion.

The same for the custom Array.toTable impl: please use our standard formatting for the hand-written code. Also, enhance the history entry to specify that toTable was re-written to use a dynamic temp-table.

Constantin Asofiei wrote:

Andrei, when I mentioned the javadocs, I meant what you manually wrote - not to add/fix 4GL javadocs; sorry for the confusion.

The same for the custom Array.toTable impl: please use our standard formatting for the hand-written code. Also, enhance the history entry to specify that toTable was re-written to use a dynamic temp-table.

Ooh, got it.

Constantin, I have addressed the copyright year, history entry and the javadocs in 9488b/rev. 16346

Andrei Plugaru wrote:

Constantin, I have addressed the copyright year, history entry and the javadocs in 9488b/rev. 16346

Thanks. Please put this in both conversion and runtime regression.

We need also the skeletons updated with the Array.cls change for the IList interface (this will be committed when 9488b is merged). Please post the skeleton patch here.

Just a heads up: the readme in the root project file needs to be edited in the branch to contain the ADE files which have been included in FWD.

I have committed 9488b/16425 with the updated readme. For now, I have just added the directories with the ADE files as I saw there was a TODO in this regard. However, I can also add the exact files if needed.

Regarding the testing, currently there is only one more application that needs to be tested. For the others, both the conversion and runtime succeeded.

Andrei, something else: if we patch skeleton I assume is mandatory for 9488b to be in trunk, right?

Constantin Asofiei wrote:

Andrei, something else: if we patch skeleton I assume is mandatory for 9488b to be in trunk, right?

Hmm, I am not really aware of all the situations when using out of sync skeleton and p2j can cause problems. The changes in skeleton are mostly additive(new methods, constructors, properties), however also some class hierarchies have been changed.

I have spotted an error in skeleton's oo4gl/OpenEdge/Core/Collections/Array.cls, IListIterator is missing the qualifier, here is the patch:

=== modified file 'oo4gl/OpenEdge/Core/Collections/Array.cls'
--- old/oo4gl/OpenEdge/Core/Collections/Array.cls    2026-02-23 15:21:00 +0000
+++ new/oo4gl/OpenEdge/Core/Collections/Array.cls    2026-03-04 09:02:12 +0000
@@ -116,10 +116,10 @@
     method public integer IndexOf( input pElement as Progress.Lang.Object ):
     end.

-    method public IListIterator ListIterator():
+    method public OpenEdge.Core.Collections.IListIterator ListIterator():
     end.

-    method public IListIterator ListIterator(input pIndex as integer):
+    method public OpenEdge.Core.Collections.IListIterator ListIterator(input pIndex as integer):
     end.

     method public integer LastIndexOf(input pElement as Progress.Lang.Object):
@@ -128,7 +128,7 @@
     method public Progress.Lang.Object Remove(input pIndex as integer):
     end.

-    method public IList SubList(input pFromIndex as integer,
+    method public OpenEdge.Core.Collections.IList SubList(input pFromIndex as integer,
                                 input pToIndex as integer):
     end.

Stefanel Pezamosca wrote:

I have spotted an error in skeleton's oo4gl/OpenEdge/Core/Collections/Array.cls, IListIterator is missing the qualifier, here is the patch:
[...] Should I commit it to skeleton repo ?

Please do.

From Progress packages - the interfaces I think can be marked as FULL; please group them together in the .

From other packages from ADE, which have FWD counterparts: check if the .java file has all methods marked as CVT_LVL_FULL and RT_LVL_FULL - in that case I think we can mark the class a FULL, also (assuming after a check of the ADE .cls file the implementation looks right). I'd like to leave these 'as is' for now.

For p2j.oo.web. package, this I don't think can be converted from ADE (at this time) without some other changes from #6506, WEB-CONTEXT and maybe also WEB stream.

• classes which are in FWD, but are full STUB - these can be converted from ADE
• for packages which also have other classes in p2j.oo, what .cls files are not in FWD
• for remaining packages, a list of them and what they do

Constantin Asofiei wrote:

From Progress packages - the interfaces I think can be marked as FULL; please group them together in the .

Modified Icollection, Icomparable, Icomparer, IequalityComparer, Ihasable, Iiterable, Iiterator, Ilist, Imap, Iset

From other packages from ADE, which have FWD counterparts: check if the .java file has all methods marked as CVT_LVL_FULL and RT_LVL_FULL - in that case I think we can mark the class a FULL, also (assuming after a check of the ADE .cls file the implementation looks right). I'd like to leave these 'as is' for now.

• AssertObject, all methods are CVT_LVL_FULL | RT_LVL_FULL but compared to 12.8.4.0 it seems there are a lot of methods missing
• AbstractAssert
• AbstractAssertionFailedError
• Assert (I need to double check the methods, there are a lot)
• ByteBucket (doesn't use MemoryOutputStream)
• JsonEntityWriter (write_1 that uses Cookie is incomplete but marked as FULL, found one so probably there are more)
• MultipartEntityWriter (CRLF is combined, but those are separated in the original code, uses Memptr from OpenEdge.Core.Memptr but the converted code seems to use a different interface)
• DefaultMessageWriterBuilder (NewMessageWriter method is missing)
• ContentDispositionHeaderBuilder (looks ok)
• HttpRequestDecorator (looks ok)
• TransferEncodingEnum (looks ok)
• RemoteWebRequest (looks a bit off, but I've seen var being used without def)

For p2j.oo.web. package, this I don't think can be converted from ADE (at this time) without some other changes from #6506, WEB-CONTEXT and maybe also WEB stream.

Otherwise, from ADE repository, we need to check:

• classes which are in FWD, but are full STUB - these can be converted from ADE

I think only those with RT_LVL_STUB can be converted now.

• for packages which also have other classes in p2j.oo, what .cls files are not in FWD
• for remaining packages, a list of them and what they do

I will check for what other classes are being used, but do not exists in FWD. For the rest, I will try to create a wiki (here for the moment).

I created a zfile to convert all the classes mentioned in #9488-227 and there are a lot of failures because of the var defined variables (without defined), this was fixed previously by replacing this definition with one that uses the define variable statement.

I converted OpenEdge/Core/Assertion/AssertError.cls, but only the methods converted (without any of the logic). Is the goal to convert those classes fully or there should also be changes done by hand?

Dănuț Filimon wrote:

I created a zfile to convert all the classes mentioned in #9488-227 and there are a lot of failures because of the var defined variables (without defined), this was fixed previously by replacing this definition with one that uses the define variable statement.

I converted OpenEdge/Core/Assertion/AssertError.cls, but only the methods converted (without any of the logic). Is the goal to convert those classes fully or there should also be changes done by hand?

•    @LegacySignature(type = Type.EXECUTE)
     public void __src_corelib_openedge_core_assertion_AssertError_execute__()
     {
        onBlockLevel(Condition.ERROR, Action.THROW);
        {
        }
     }
  

  is not needed
• there are two constructors:
  

  
     @LegacySignature(type = Type.CONSTRUCTOR, name = "AssertError")
     public void __src_corelib_openedge_core_assertion_assertError_constructor__()
     {
        internalProcedure(AssertError.class, this, "AssertError", new Block((Body) () -> 
        {
           __lang_BaseObject_constructor__();
        }));
     }
  
     @LegacySignature(type = Type.CONSTRUCTOR, name = "AssertError")
     public static void __src_corelib_openedge_core_assertion_AssertError_constructor__static__()
     {
        externalProcedure(AssertError.class, "AssertError", new Block((Body) () -> 
        {
           onBlockLevel(Condition.ERROR, Action.THROW);
        }));
     }
  
  

  while the one from trunk only has one.

Dănuț Filimon wrote:

Dănuț Filimon wrote:

I created a zfile to convert all the classes mentioned in #9488-227 and there are a lot of failures because of the var defined variables (without defined), this was fixed previously by replacing this definition with one that uses the define variable statement.

Yes, FWD doesn't support the short-hand var statement.

• [...] is not needed
• there are two constructors:
  [...] while the one from trunk only has one.

What do you mean exactly? Did you try to convert with trunk and 9488a? Please upload to devsrv01 the converted files (don't post them here at this time).

Constantin Asofiei wrote:

Dănuț Filimon wrote:

Dănuț Filimon wrote:

I created a zfile to convert all the classes mentioned in #9488-227 and there are a lot of failures because of the var defined variables (without defined), this was fixed previously by replacing this definition with one that uses the define variable statement.

Yes, FWD doesn't support the short-hand var statement.

• [...] is not needed
• there are two constructors:
  [...] while the one from trunk only has one.

What do you mean exactly? Did you try to convert with trunk and 9488a? Please upload to devsrv01 the converted files (don't post them here at this time).

I am converted them one by one now and checking for dependencies, but there are also a few conversion issues. I am running the conversion with trunk + a patch from 9488a (the branch is old and some changes are already in trunk so a rebase is problematic).

I currently have a conversion issue. SortedSet is being used, but it is not part of the ADE code while it can be found in the skeleton. This is used by LogMessage.cls (dependency).

Committed 9488c/16551 containing conversion changes from 9488a + additional change to fix the LegacyString conversion. The Join(Cast()) call was troublesome and it would throw an error while loading the class because it could not annotate the method call.

I've uploaded ade.src.9488c.7z to devsrv01:/tmp/ which contains the conversion for:

abl/src/corelib/OpenEdge/Core/Assertion/AssertError.cls
abl/src/corelib/OpenEdge/Core/Assertion/AssertObject.cls
abl/src/corelib/OpenEdge/Core/Collections/LinkedList.cls
abl/src/corelib/OpenEdge/Core/Collections/ListNode.cls
abl/src/corelib/OpenEdge/Core/Assert.cls
abl/src/corelib/OpenEdge/Core/String.cls
abl/src/corelib/OpenEdge/Core/Memptr.cls
abl/src/OpenEdge/DataAdmin/Error/DataAdminError.cls
abl/src/OpenEdge/DataAdmin/Error/OperationError.cls

I had to use List.cls from skeleton because of a few issues, other classes had to be removed from skeleton because the conversion failed.

There are a few issues which I found that still give me issues during conversion, mostly related to generics.

#9488-74 has been reported before, but not fixed. I also found it easier to just comment out the methods involved and go on with the conversion and uncover other issues.

• Other chained calls which I removed related to LogMessage
• GetByte from ByteBucket.cls is a method call, which is also a reserved keyword (I adjusted progress.g)
• var statement => def var
• extent variables which assign 128 characters at once, but are not formatted properly
• cast calls using generics
• method calls which can't be found (an extent variable without a size is passed as parameter and the ParameterKey shows character[] instead of KW_INPUT character⁵)

I've reached a stable conversion with a few more sources and I'm trying to patch p2j.oo.* with what I have currently.

var statement => def var

We should just add a new rule to match this language statement. I don't think it is correct to change def var to support var.

cast calls using generics

Please document the syntax and capabilities of the "built-in" generics approach. We will then discuss how we make the correct changes to support it.

Greg Shah wrote:

var statement => def var

We should just add a new rule to match this language statement. I don't think it is correct to change def var to support var.

cast calls using generics

Please document the syntax and capabilities of the "built-in" generics approach. We will then discuss how we make the correct changes to support it.

I am currently working on getting the var statement implemented. I've also discovered that method calls can use the name of reserved_functions, which was not specified in progress.g (I've tested all reserved_functions keywords and made methods with each one, the code built and ran fine).

I will also document the generics.

Greg, about the VAR statement. Wouldn't the current structure of DEFINE VARIABLE be reasonable? During conversion, we can fix the VAR statement and turn it into a DEFINE VARIABLE statement.

Dănuț Filimon wrote:

Greg, about the VAR statement. Wouldn't the current structure of DEFINE VARIABLE be reasonable? During conversion, we can fix the VAR statement and turn it into a DEFINE VARIABLE statement.

No, it is not a good fit. The core structure is different in important ways, the worst of which is that it can implement more than one variable for the same statement.

We can reuse many of the sub-rules but the main var_stmt needs to be its own thing. And we will want to deal with it downstream in a different way.

Dănuț Filimon wrote:

Committed 9488c/16551 containing conversion changes from 9488a + additional change to fix the LegacyString conversion. The Join(Cast()) call was troublesome and it would throw an error while loading the class because it could not annotate the method call.

I do not plan to commit the 9488a changes to trunk at this time. Any changes made to p2j.oo existing (or new) classes, please add them in a distinct branch.

I've uploaded ade.src.9488c.7z to devsrv01:/tmp/ which contains the conversion for:

• OpenEdge.Core.Assertion classes in p2j.oo are implemented in a peculiar way. At this time, existing classes should not be changed. The assert effort needs to be on its own branch.
• String, Memptr and others: lets discuss before moving to converting existing classes, which are not marked as 'stub' support
• I recall in 9488a was an option to emit the LegacyResource and LegacyResourceSupport annotations - these do not exist here

• lets make a list of the next candidate ADE .cls files to convert and discuss them
• use a separate branch to prepare p2j.oo code which gets converted and will be committed (with the usual license notices in the .java file and to p2j/readme file)
• leave 9488c only with FWD changes (not p2j.oo) and we will discuss them
• further work for unsupported 4GL features (like the var statement) need to be worked in separate branches

Actually, for generics there is #9868 and for the var statement there is #9767 - lets work them in those tasks, please.

Constantin Asofiei wrote:

I do not plan to commit the 9488a changes to trunk at this time. Any changes made to p2j.oo existing (or new) classes, please add them in a distinct branch.

I've uploaded ade.src.9488c.7z to devsrv01:/tmp/ which contains the conversion for:

• OpenEdge.Core.Assertion classes in p2j.oo are implemented in a peculiar way. At this time, existing classes should not be changed. The assert effort needs to be on its own branch.
• String, Memptr and others: lets discuss before moving to converting existing classes, which are not marked as 'stub' support
• I recall in 9488a was an option to emit the LegacyResource and LegacyResourceSupport annotations - these do not exist here

is-legacy-oo and legacy-res-support-type are actually broken, I got lots of errors while using them.

My conclusion:

• lets make a list of the next candidate ADE .cls files to convert and discuss them

I've posted a list of classes in #9488-227, we can take anything that has PARTIAL support.

• use a separate branch to prepare p2j.oo code which gets converted and will be committed (with the usual license notices in the .java file and to p2j/readme file)
• leave 9488c only with FWD changes (not p2j.oo) and we will discuss them

Got it.

Dănuț Filimon wrote:

is-legacy-oo and legacy-res-support-type are actually broken, I got lots of errors while using them.

What errors do you get?

My conclusion:

• lets make a list of the next candidate ADE .cls files to convert and discuss them

I've posted a list of classes in #9488-227, we can take anything that has PARTIAL support.

why PARTIAL? Are all STUB classes converted?

Please go through the list and strike-through all that were fixed via 9488b.

Constantin Asofiei wrote:

Dănuț Filimon wrote:

is-legacy-oo and legacy-res-support-type are actually broken, I got lots of errors while using them.

What errors do you get?

With is-legacy-oo set to true, I get a lot of:

     [java] WARNING: Can not copy annotations for member in OpenEdge.Core.Util.TokenResolverEventArgs: parent is null
     [java] WARNING: Can not copy annotations for member in OpenEdge.Core.Util.TokenResolverEventArgs: parent is null
     [java] WARNING: Can not copy annotations for member in OpenEdge.Core.Util.TokenResolverEventArgs: parent is null

     [java] ------------------------------------------------------------------------------
     [java] Code Conversion Annotations
     [java] ------------------------------------------------------------------------------
     [java] 
     [java] Optional rule set [customer_specific_annotations_prep] not found.
     [java] ./abl/src/corelib/OpenEdge/Core/Assert.cls
     [java] MISSING_REFID for property Size [VAR_INT]:103079236434 @2733:22
     [java]       (tempidx-file=./abl/src/corelib/OpenEdge/Core/Collections/ICollection.cls, tempidx=0, cls-container=./abl/src/corelib/OpenEdge/Core/Collections/ICollection.cls, found-in-source-file=./abl/src/corelib/OpenEdge/Core/Collections/ICollection.cls, found-in-cls=OpenEdge.Core.Collections.ICollection, cls-property=true, access-mode=2133)
     [java] 
     [java] 
     [java] EXPRESSION EXECUTION ERROR:
     [java] ---------------------------
     [java] apropref = propref.getImmediateChild(prog.kw_get, null)
     [java]                    ^  { Cannot invoke "com.goldencode.ast.Aast.getImmediateChild(int, com.goldencode.ast.Aast)" because the return value of "com.goldencode.expr.CompiledExpression.getVar(int)" is null }
     [java] ---------------------------
     [java] ERROR!  Active Rule:
     [java] -----------------------
     [java]       RULE REPORT      
     [java] -----------------------
     [java] Rule Type :   WALK
     [java] Source AST:  [ Size ] BLOCK/CLASS_DEF/BLOCK/METHOD_DEF/BLOCK/STATEMENT/KW_IF/EXPRESSION/EQUALS/OBJECT_INVOCATION/VAR_INT/ @2733:22 {103079236434}
     [java] Copy AST  :  [ Size ] BLOCK/CLASS_DEF/BLOCK/METHOD_DEF/BLOCK/STATEMENT/KW_IF/EXPRESSION/EQUALS/OBJECT_INVOCATION/VAR_INT/ @2733:22 {103079236434}
     [java] Condition :  apropref = propref.getImmediateChild(prog.kw_get, null)
     [java] Loop      :  false
     [java] --- END RULE REPORT ---

My conclusion:

• lets make a list of the next candidate ADE .cls files to convert and discuss them

I've posted a list of classes in #9488-227, we can take anything that has PARTIAL support.

why PARTIAL? Are all STUB classes converted?

I based it on

String, Memptr and others: lets discuss before moving to converting existing classes, which are not marked as 'stub' support

Please go through the list and strike-through all that were fixed via 9488b.

The following were added in trunk/16440.

  src/com/goldencode/p2j/oo/common/support/IdateTimeTzHolder.java
  src/com/goldencode/p2j/oo/common/support/IdecimalHolder.java
  src/com/goldencode/p2j/oo/common/support/Iint64Holder.java
  src/com/goldencode/p2j/oo/common/support/IlongcharArrayHolder.java
  src/com/goldencode/p2j/oo/common/support/IprimitiveArrayHolder.java
  src/com/goldencode/p2j/oo/core/DataTypeHelper.java
  src/com/goldencode/p2j/oo/core/EventArgs.java
  src/com/goldencode/p2j/oo/core/ImemoryPointer.java
  src/com/goldencode/p2j/oo/core/LegacyDecimal.java
  src/com/goldencode/p2j/oo/core/LegacyInteger.java
  src/com/goldencode/p2j/oo/core/LongcharArrayHolder.java
  src/com/goldencode/p2j/oo/core/Session.java
  src/com/goldencode/p2j/oo/core/StringConstant.java
  src/com/goldencode/p2j/oo/core/TimeStamp.java
  src/com/goldencode/p2j/oo/core/TimeStampConstants.java
  src/com/goldencode/p2j/oo/core/collections/AbstractTtcollection.java
  src/com/goldencode/p2j/oo/core/collections/Array.java
  src/com/goldencode/p2j/oo/core/collections/LegacyList.java*
  src/com/goldencode/p2j/oo/core/collections/LegacyListIterator.java*
  src/com/goldencode/p2j/oo/core/collections/LegacyStack.java*
  src/com/goldencode/p2j/oo/core/collections/ObjectStack.java*
  src/com/goldencode/p2j/oo/core/collections/StringCollection.java*
  src/com/goldencode/p2j/oo/core/system/InvalidValueSpecifiedError.java
  src/com/goldencode/p2j/oo/core/system/NotFoundError.java
  src/com/goldencode/p2j/oo/core/util/Token.java
  src/com/goldencode/p2j/oo/core/util/TokenResolver.java
  src/com/goldencode/p2j/oo/core/util/TokenResolverEventArgs.java
  src/com/goldencode/p2j/oo/logging/LogEvent.java
  src/com/goldencode/p2j/oo/logging/LogLevelEnum.java
  src/com/goldencode/p2j/oo/logging/LogMessage.java
  src/com/goldencode/p2j/oo/security/Principal.java