Project

General

Profile

Activity

From 12/08/2022 to 01/06/2023

01/06/2023

05:53 PM Feature #6821: java collection performance
Greg Shah wrote:
> Eric Faulhaber wrote:
> > Greg Shah wrote:
> > > > Greg, are you suggesting to maintain for tem... Eric Faulhaber
03:41 PM Feature #6821: java collection performance
> Please backport this 3821c/14472 to 6129b and commit it there.
Committed as 6129b rev 14343. Stanislav Lomany
02:19 PM Feature #6821: java collection performance
Eric Faulhaber wrote:
> Greg Shah wrote:
> > > Greg, are you suggesting to maintain for temp-tables a map of Record... Greg Shah
02:18 PM Feature #6821: java collection performance
Do we assign a new @rowid@ when we copy temp-table records from one temp-table to another, like in table parameter pr... Greg Shah
02:07 PM Feature #6821: java collection performance
I made an assumption that temp-table record IDs are not unique across multiplex IDs, which is wrong, as you're saying... Stanislav Lomany
02:05 PM Feature #6821: java collection performance
Stanislav Lomany wrote:
> > Do I understand correctly that only the most recently added entry will be kept and any o... Eric Faulhaber
02:03 PM Feature #6821: java collection performance
Greg Shah wrote:
> One thing that is not obvious to me is how the cache is used. Since we are using temp-tables wit... Eric Faulhaber
02:01 PM Feature #6821: java collection performance
> Do I understand correctly that only the most recently added entry will be kept and any older entry is overwritten? ... Stanislav Lomany
01:56 PM Feature #6821: java collection performance
Greg Shah wrote:
> > Greg, are you suggesting to maintain for temp-tables a map of RecordBuffer.dmoClass -> Set of r... Eric Faulhaber
01:29 PM Feature #6821: java collection performance
One thing that is not obvious to me is how the cache is used. Since we are using temp-tables with this mechanism, it... Greg Shah
01:25 PM Feature #6821: java collection performance
> Greg, are you suggesting to maintain for temp-tables a map of RecordBuffer.dmoClass -> Set of records?
Not exac... Greg Shah
12:52 PM Feature #6821: java collection performance
@invalidateRecords@ method is called only for temp-tables. And called only by @ACCEPT-CHANGES@ method (I'm interested... Stanislav Lomany
12:25 PM Feature #6821: java collection performance
All usage of @Session.invalidateRecords()@ comes from @RecordBuffer.pruneSessionCache()@ which is only called in 2 pl... Greg Shah
12:05 PM Feature #6821: java collection performance
Stanislav Lomany wrote:
> > TableMapper$LegacyFieldInfo.getMutableInfo 1.8 million
>
> This one looks like cachin... Constantin Asofiei
11:39 AM Feature #6821: java collection performance
> TableMapper$LegacyFieldInfo.getMutableInfo 1.8 million
This one looks like caching may be slightly improved, but... Stanislav Lomany
02:14 AM Feature #6821: java collection performance
Stanislav Lomany wrote:
> > You also need checks if this is the single resource in the chain being deleted, in this ... Constantin Asofiei
03:05 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
I understand there is no performance difference in lambdas themselves. I'm wondering about the call to @LambdaFactor... Greg Shah
02:51 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Greg Shah wrote:
> The dynamic generation of the lambdas for an external procedure is only done the first time the e... Constantin Asofiei
10:44 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
The dynamic generation of the lambdas for an external procedure is only done the first time the external proc is load... Greg Shah
10:22 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Eric: Are you OK with the proxy and persistence changes? Greg Shah
10:22 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Code Review 6819_20221223_6129b_14341.patch
I have no objections. Greg Shah
02:44 PM Support #4032: block/transaction processing context-local and other performance improvements
Constantin Asofiei wrote:
> * I think I'll do some experimenting if keeping the reference of @ProcedureManager$Proce... Constantin Asofiei

01/05/2023

02:37 PM Feature #6821: java collection performance
> You also need checks if this is the single resource in the chain being deleted, in this firstResource/lastResource ... Stanislav Lomany
06:20 AM Feature #6821: java collection performance
Constantin Asofiei wrote:
> Attached is a space-delimited file with all FOR EACH loops which have a List, ArrayList ... Constantin Asofiei
04:26 AM Feature #6821: java collection performance
Forgot to add the csv file. Constantin Asofiei
04:25 AM Feature #6821: java collection performance
Keep in mind that this:... Constantin Asofiei
04:25 AM Feature #6821: java collection performance
Stanislav Lomany wrote:
> > what branch and revision did you use for this patch?
>
> 3821c rev 14467.
The patch ... Constantin Asofiei
01:54 PM Support #4032: block/transaction processing context-local and other performance improvements
Attached is a trace file with @ContextLocal.get@ usage in a customer application. This is expanded on 3 levels, leve... Constantin Asofiei
08:10 AM Feature #3853: implement LOG-MANAGER runtime
> > Is this different than @MESSAGE VIEW-AS ALERT-BOX@? Or is this just the case where the 4GL handles the really lo... Greg Shah
08:10 AM Feature #3853: implement LOG-MANAGER runtime
> It is @ContextLocal@, which creates an instance of the log manager wrapper through @class Fallback extends ThreadLo... Greg Shah
07:57 AM Feature #3853: implement LOG-MANAGER runtime
Greg Shah wrote:
> We must solve this properly. The behavior needs to match the 4GL implementation while implementi... Galya B
07:51 AM Feature #3853: implement LOG-MANAGER runtime
Greg Shah wrote:
> Is this different than @MESSAGE VIEW-AS ALERT-BOX@? Or is this just the case where the 4GL handl... Galya B
07:49 AM Feature #3853: implement LOG-MANAGER runtime
Greg Shah wrote:
> > Our implementation of @LOG-MANAGER@ runs purely on the server.
>
> It is certainly preferab... Galya B
07:48 AM Feature #3853: implement LOG-MANAGER runtime
> Do we want also the 32767 chars limit for @LOG-MANAGER:WRITE-MESSAGE@?
Yes. Also, please implement the limit of... Greg Shah
06:47 AM Feature #3853: implement LOG-MANAGER runtime
A few observations in relationship to max log message length (tested):
* By default statement length is limited to... Galya B
04:47 AM Feature #3853: implement LOG-MANAGER runtime
Greg Shah wrote:
> Does the 4GL have an implicit set of settings for @LOG-MANAGER@ in an appserver process? If so, ... Galya B
03:50 AM Feature #3853: implement LOG-MANAGER runtime
Constantin Asofiei wrote:
> Galya: please post what logmanager related config needs to added.
4GL LOG-MANAGER con... Galya B
03:06 AM Feature #3853: implement LOG-MANAGER runtime
Greg Shah wrote:
> For @logthreshold@ and @numlogfiles@ and the other items from #3853-19, we should be able to conf... Constantin Asofiei

01/04/2023

02:44 PM Feature #6817 (WIP): replace INSTANTIATING-PROCEDURE and other handle<ExternalProgram> with the a...
Hynek Cihlar
11:23 AM Feature #6817: replace INSTANTIATING-PROCEDURE and other handle<ExternalProgram> with the actual ...
In Eclipse, there is a way to look for a field or local variable of a certain type. Other IDEs should have something... Constantin Asofiei
12:46 PM Feature #3853: implement LOG-MANAGER runtime
Greg Shah wrote:
> From Galya via email:
>
> > I've tested gui, batch, web gui clients and they are working well ... Constantin Asofiei
12:33 PM Feature #3853: implement LOG-MANAGER runtime
From Galya via email:
> I've tested gui, batch, web gui clients and they are working well with @LOG-MANAGER@. But ... Greg Shah
12:29 PM Feature #3853: implement LOG-MANAGER runtime
For @logthreshold@ and @numlogfiles@ and the other items from #3853-19, we should be able to configure these via 2 me... Greg Shah
08:04 AM Feature #3853: implement LOG-MANAGER runtime
Galya Bogdanova wrote:
> open a @-debugalert@ separate task (maybe #4065 or this one #3853) to migrate the logic to ... Galya B
07:55 AM Feature #3853: implement LOG-MANAGER runtime
Galya Bogdanova wrote:
> P.S.S. The easiest quickest patch to complete #5753 would be to just switch the dependency ... Galya B
04:58 AM Feature #3853: implement LOG-MANAGER runtime
*Current clientlog implementation and -debugalert*
Currently the code uses @-clientlog@ startup param to enable lo... Galya B
03:10 AM Feature #3853: implement LOG-MANAGER runtime
*Rotation and testing*
A substantial part of LOG-MANAGER's logic is related to handling rotation of log files and is... Galya B
08:54 AM Feature #6821: java collection performance
> what branch and revision did you use for this patch?
3821c rev 14467. Stanislav Lomany
05:17 AM Feature #6821: java collection performance
Stanislav Lomany wrote:
> > Related to map access (hashCode calls):
> > ** @HandleChain.delete@ 880k
>
> Maybe w... Constantin Asofiei
05:13 AM Feature #6821: java collection performance
Stanislav Lomany wrote:
> Please review the diff which replaces "for-each" loops with indexed "for" loops (I'll add ... Constantin Asofiei
08:32 AM Feature #6824 (WIP): lazy initialization of collection class members
Constantin Asofiei

01/03/2023

02:54 PM Feature #6821: java collection performance
> Related to map access (hashCode calls):
> ** @HandleChain.delete@ 880k
Maybe we don't need to perform @first/la... Stanislav Lomany
08:33 AM Feature #3853: implement LOG-MANAGER runtime
> Do we show the same error when LOG-MANAGER:WRITE-MESSAGE tries to write more than 15000 chars? If yes, then we'll h... Greg Shah

01/02/2023

05:21 PM Feature #6821: java collection performance
Please review the diff which replaces "for-each" loops with indexed "for" loops (I'll add headers later), specificall... Stanislav Lomany

12/28/2022

07:27 AM Feature #6410: implement additional built-in OO classes/interfaces
Greg Shah wrote:
> Yes, please use @PROVERSION@ and preprocessing to conditionally implement logic by OE version.
... Marian Edu

12/23/2022

11:55 AM Feature #6824: lazy initialization of collection class members
Some notes :
* @BufferImpl.callbacks@ can be lazily created, not all buffers are used in FILL operations.
* @Record... Constantin Asofiei
08:30 AM Feature #3853: implement LOG-MANAGER runtime
@LOG-MANAGER:WRITE-MESSAGE@ supports by default text up to 15000 characters (actually it should be less, since that i... Galya B
05:45 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Constantin Asofiei wrote:
> Constantin Asofiei wrote:
> > Attached patch is built on top of #6939 patch. Eric, ple... Constantin Asofiei

12/21/2022

02:29 PM Feature #6827: reduce the memory footprint of WrappedResource implementations: replace BDTs with ...
I think the @object@ fields need to remain as-is, for the reference counting to work (which affects when these get de... Constantin Asofiei
02:28 PM Feature #6827: reduce the memory footprint of WrappedResource implementations: replace BDTs with ...
This code:... Constantin Asofiei
09:38 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Constantin Asofiei wrote:
> Attached patch is built on top of #6939 patch. Eric, please review and use this one for... Constantin Asofiei

12/20/2022

02:19 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Attached patch is built on top of #6939 patch. Eric, please review and use this one for your tests. Constantin Asofiei
04:40 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
I've tested with a 'static bytecode' (i.e. 10k .java files each with 300 lambdas, checked the disassembled .class and... Constantin Asofiei
07:47 AM Feature #7000: Replace old-style synchronization primitives with java.util.concurrent classes
Greg Shah wrote:
> Do you have specific areas seen in profiling where these old-style primitives are heavily involve... Igor Skornyakov
06:24 AM Feature #7000: Replace old-style synchronization primitives with java.util.concurrent classes
Do you have specific areas seen in profiling where these old-style primitives are heavily involved? Greg Shah
03:22 AM Feature #7000: Replace old-style synchronization primitives with java.util.concurrent classes
FWD code contains many places where old-style synchronization primitives (@synchronized@ blocks, @wait@/@notify@) are... Igor Skornyakov
02:58 AM Feature #7000 (New): Replace old-style synchronization primitives with java.util.concurrent classes
Igor Skornyakov

12/19/2022

06:08 PM Feature #3853: implement LOG-MANAGER runtime
> > The base implementation is already handled at conversion and it is stubbed out in the runtime.
>
> Does it mea... Greg Shah
06:35 AM Feature #3853: implement LOG-MANAGER runtime
Greg Shah wrote:
> The base implementation is already handled at conversion and it is stubbed out in the runtime.
... Galya B
11:28 AM Feature #6821: java collection performance
For @ArrayList.iterator()@, the top callers are:
* @PreselectQuery.components()@ - 430k
* @AnnotatedAst.notifyListe... Constantin Asofiei
10:26 AM Feature #6821: java collection performance
The ones that are most important are listed in #6821-4. We aren't suggesting changing things everywhere. Greg Shah
10:22 AM Feature #6821: java collection performance
That may be true. There're about 2k foreach loops, and about 10% of them can be updated. If we want, we can can limit... Stanislav Lomany
09:57 AM Feature #6821: java collection performance
Stanislav Lomany wrote:
> > Actually, I doubt that such optimization will result in any significant performance in a... Igor Skornyakov
09:56 AM Feature #6821: java collection performance
Small improvements over millions of calls add up. The fact that the profiling tools don't show obvious hotspots mean... Greg Shah
09:45 AM Feature #6821: java collection performance
Greg Shah wrote:
> > I would perform benchmarking/profiling first.
>
> This is what was done. See #6821-4.
Th... Igor Skornyakov
09:31 AM Feature #6821: java collection performance
> The point with all the Map and Set iterators is that they allocate lots of short-lived objects.
That's true. I m... Stanislav Lomany
09:25 AM Feature #6821: java collection performance
> Actually, I doubt that such optimization will result in any significant performance in all situations.
I did som... Stanislav Lomany
09:18 AM Feature #6821: java collection performance
> I would perform benchmarking/profiling first.
This is what was done. See #6821-4. Greg Shah
05:12 AM Feature #6821: java collection performance
Actually, I doubt that such optimization will result in any significant performance in all situations. See e.g. a dis... Igor Skornyakov
04:54 AM Feature #6821: java collection performance
Stanislav Lomany wrote:
> I assume we're taking about replacing
Correct.
> I found that there're not so many plac... Constantin Asofiei
04:45 AM Feature #6821 (WIP): java collection performance
Constantin,
> For example, using Java for-each loop for an an ArrayList instead of a @for@ and getting the element... Stanislav Lomany
09:08 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
> the performance decreased a lot (there may be some internal storage which affects the INVOKEDYNAMIC instruction at ... Greg Shah
09:08 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
From Constantin:
> More with the the lambda for ControlFlowOps, I think the lambda performance is directly proport... Greg Shah

12/16/2022

03:31 AM Bug #6999: OO methods/properties used in direct Java calls
Consider this type:... Constantin Asofiei
03:27 AM Bug #6999 (New): OO methods/properties used in direct Java calls
Constantin Asofiei

12/15/2022

09:28 AM Feature #6827 (WIP): reduce the memory footprint of WrappedResource implementations: replace BDTs...
Sorry, guys, I think I've reviewed all FWD classes implemening @WrappedResource@ but there're no usages of BDT. I've ... Stanislav Lomany

12/14/2022

01:26 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
The lambdas will created like this:... Constantin Asofiei
09:15 AM Feature #6410: implement additional built-in OO classes/interfaces
Yes, please use @PROVERSION@ and preprocessing to conditionally implement logic by OE version. Greg Shah
09:09 AM Feature #6410: implement additional built-in OO classes/interfaces
Greg Shah wrote:
> OE 11.7.x is in use right now. They can't move to OE 12 without facing serious issues which have... Marian Edu
07:42 AM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Greg Shah wrote:
> Can you do something low effort to work around this issue until we have 6129b changes ready?
I... Vladimir Tsichevski
07:32 AM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Can you do something low effort to work around this issue until we have 6129b changes ready? Greg Shah

12/13/2022

04:54 PM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Greg Shah wrote:
> Vladimir: Please check your failing scenario in 6129b and let us know what you find.
Yes, in... Vladimir Tsichevski
04:53 PM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Constantin Asofiei wrote:
> Check your classpath. ConvertedClassName no longer exists in 6129b.
Yes, I forgot to ... Vladimir Tsichevski
04:29 PM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Check your classpath. ConvertedClassName no longer exists in 6129b. Constantin Asofiei
03:43 PM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Greg Shah wrote:
> Vladimir: Please check your failing scenario in 6129b and let us know what you find.
I canno... Vladimir Tsichevski
02:04 PM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Our OO implementation is much more advanced in vranch 6129b than in 3821c. In 6129b, at this time we only use @Legac... Greg Shah
07:30 AM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Vladimir Tsichevski wrote:
> No, it is the Java sources which will not compile.
> Note: the name if the field ref... Marian Edu
07:24 AM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Marian Edu wrote:
> All classes in @OpenEdge@ and @Progress@ packages were created *manually*, this is the expecte... Vladimir Tsichevski
07:13 AM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Vladimir Tsichevski wrote:
> The fields in @OpenEdge.Core.DataTypeEnum@ are named after the corresponding Progress b... Marian Edu
06:59 AM Bug #6993 (Review): OO: incorrect field names in DataTypeEnum.java
Fixed in 3821c rev. 14438. Vladimir Tsichevski
06:45 AM Bug #6993 (WIP): OO: incorrect field names in DataTypeEnum.java
Vladimir Tsichevski
06:42 AM Bug #6993: OO: incorrect field names in DataTypeEnum.java
Please do fix this. Greg Shah
06:31 AM Bug #6993: OO: incorrect field names in DataTypeEnum.java
The fields in @OpenEdge.Core.DataTypeEnum@ are named after the corresponding Progress base data types, like @characte... Vladimir Tsichevski
06:24 AM Bug #6993 (Closed): OO: incorrect field names in DataTypeEnum.java
Vladimir Tsichevski

12/08/2022

03:13 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Yes. I assume the proxy case is being worked on in #6939. Greg Shah
02:42 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Greg, the 8 million iterations chosen for the MthdTest was because this is the number of Utils.invoke for a large app... Constantin Asofiei
01:52 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
> How will it pass the parameters?
We can handle this with a bit of "standardization" in the registration or via a... Greg Shah
01:41 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
I've tested something like this (for my original MthdTest):... Constantin Asofiei
12:52 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Constantin Asofiei wrote:
> Greg Shah wrote:
> > We can generate Java code to register lambdas with CFO during an i... Constantin Asofiei
12:41 PM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Greg Shah wrote:
> We can generate Java code to register lambdas with CFO during an initializer. It could be an arr... Constantin Asofiei
11:24 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
We can generate Java code to register lambdas with CFO during an initializer. It could be an array where each elemen... Greg Shah
11:20 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Greg, for ControlFlowOps, this lambda method approach can't be easily used - because we will have to generate 'on the... Constantin Asofiei
10:13 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
> For this, we need to change the FWD proxy impl to send the lambda beside the Method to the handler.
Eric's idea ... Greg Shah
10:03 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
Greg Shah wrote:
> I'm especially excited to see how this improves the @ControlFlowOps@ usage since we could elimina... Constantin Asofiei
08:02 AM Feature #6819: refactor FWD proxy implementation to use ReflectASM instead of Java Method reflection
I'm especially excited to see how this improves the @ControlFlowOps@ usage since we could eliminate the dispatching c... Greg Shah
 

Also available in: Atom