Activity
From 03/05/2014 to 04/03/2014
04/03/2014
- 05:49 AM Feature #2251: improve the call graph generation
- The branch rev 10504 contains the fix for generating the callgraph reports on windows (the filename should have been ...
03/26/2014
- 03:49 PM Feature #2251 (Closed): improve the call graph generation
- 02:23 PM Feature #2251 (Review): improve the call graph generation
- 02:23 PM Feature #2251: improve the call graph generation
- Branch revision 10518 was committed to the 10496/10497, both on trunk master on devsrv01 and filesrv01.
03/25/2014
- 04:15 PM Feature #2251: improve the call graph generation
- > Regarding note 97: I think bzr merge should do the trick, to merge the branch into the trunk
Yes, that was my pl... - 02:12 PM Feature #2251: improve the call graph generation
- The update passed runtime and conversion testing (both server and MAJIC code).
Regarding note 97: I think @bzr mer... - 03:49 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> > BTW, to "release" a branch to the trunk, is there another way beside just copying the branch co...
03/24/2014
- 03:26 PM Feature #2251: improve the call graph generation
- > BTW, to "release" a branch to the trunk, is there another way beside just copying the branch content (without the ....
- 02:21 PM Feature #2251: improve the call graph generation
- See the branch 10517 revision (which was rebased) for the latest changes (some doc changes and a minor fix). This is ...
- 06:44 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> 1. What is your general sense about the performance? Do you feel it is "good", "bad" or somethin...
03/21/2014
- 01:27 PM Feature #2251: improve the call graph generation
- I saw your note about the performance metrics in #2260. The numbers seem pretty good for processing such a large pro...
- 01:24 PM Feature #2251: improve the call graph generation
- Code Review branch rev 10514
I am good with the changes. If you still think it is finished, go ahead with regress... - 12:40 PM Feature #2251: improve the call graph generation
- Branch rev 10514 adds the "-u" argument for CallGraphGenerator, which uses the passed filename list and all ambiguous...
- 10:17 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> One question: it looks to me like each call-graph run is fresh and we don't support incremental c... - 09:20 AM Feature #2251: improve the call graph generation
- Code Review branch rev 10513
Everything looks good.
One question: it looks to me like each call-graph run is fr... - 08:44 AM Feature #2251: improve the call graph generation
- > I think I'm missing some index configuration, as CallGraphWorker.findNodeById doesn't look like it uses an index (i...
- 08:40 AM Feature #2251: improve the call graph generation
- Code Review branch rev 10512
Everything looks good, except that the ProgressPatternWorker change is not needed. @...
03/20/2014
- 05:04 PM Feature #2251: improve the call graph generation
- Branch rev 10512 contains:
- the fix for the FILE-NAME preproc var
- some misc fixes
- some performance improvemen... - 09:30 AM Feature #2251: improve the call graph generation
- OK, thanks for clarifying this.
- 09:02 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> path/to/first/filename.p
> path/to/second/filename.p
> other/path/filename.p
>
> A reverse s... - 08:56 AM Feature #2251: improve the call graph generation
- My concern is that this algorithm can match multiple filenames in the project, making the results ambiguous.
Assum... - 08:25 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> 2. I'm a little worried that the @CallGraphWorker.resolveExternalPrograms()@ approach of using th... - 11:32 AM Feature #1519: report improvements
- A report listing hard coded absolute filenames would be useful.
- 11:32 AM Feature #1519: report improvements
- Dynamic extent and extent parameter reports are already done (they were done before now).
The incompatible program... - 08:53 AM Feature #1512 (Closed): separate match processing into a single pass of all report filters per file
- 08:50 AM Feature #1520: design gap analysis approach
- We want to make the marking as data driven as possible. This will allow rapid enhancement of the facilities without ...
03/19/2014
- 05:11 PM Feature #2251: improve the call graph generation
- Code Review branch rev 10510
1. Please add a comment into the init-rules of post_parse_fixups.xml explaining that ... - 02:36 PM Feature #2251: improve the call graph generation
- Rev 10510 contains:
* fix for the @RUN <internal-proc>@ reported as missing external procedures.
* fix for @RUN por... - 02:16 PM Feature #2251: improve the call graph generation
- There is a way to get rid of the non-indexed suffix search, when looking for external programs: keep, alongside the @...
- 12:26 PM Bug #2262: ON statement preprocessing and parsing issues related to unmatched quotes or tilde
- This testcase (checked in as on_stmt_event_as_single_or_double_quote.p) shows the issues:...
- 11:13 AM Bug #2262 (New): ON statement preprocessing and parsing issues related to unmatched quotes or tilde
03/18/2014
- 04:48 PM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> I think it is best to strip off the .dll from the library names (if present) and let them all cas... - 04:36 PM Feature #2251: improve the call graph generation
- > As a side note, there are some PROCEDURE ... EXTERNAL cases where the library name is defined as i.e. kernel32, wit...
- 04:32 PM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> 1. What is the effort involved to add all unreferenced include files to the dead files list?
Thi... - 04:28 PM Feature #2251: improve the call graph generation
- Code Review branch rev 10507
1. What is the effort involved to add all unreferenced include files to the dead file... - 04:13 PM Feature #2251: improve the call graph generation
- After analysing the report listings for the server sources, I have these cases which are reported wrong:
# @RUN <int... - 01:36 PM Feature #2251: improve the call graph generation
- rev 10507 contains the changes from note 69's review, the PatternEngine changes in note 68 plus:
- licence files for... - 09:38 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> By the way, the rebasing approach that you took looks reasonable. How hard did you find it to us... - 09:24 AM Feature #2251: improve the call graph generation
- > The proposed approach for this plugin is:
Very good. - 09:22 AM Feature #2251: improve the call graph generation
- By the way, the rebasing approach that you took looks reasonable. How hard did you find it to use?
- 09:21 AM Feature #2251: improve the call graph generation
- Code Review branch rev 10504
1. In the native_process function, please add support for the undocumented @kw_ctos@ ... - 08:42 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> > I don't think we need to rely on physical .ast files for the initial list of graph nodes. To d... - 07:30 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> > BTW, do you have a rebase script, so I can rebase?
>
> Not yet. Please go through the proce...
03/17/2014
- 02:51 PM Feature #2251: improve the call graph generation
- In bzr there are these changes:
* Switched to integer token types for call-site-key and node-type.
* Fixed progres... - 11:02 AM Feature #2251: improve the call graph generation
- > If the entire AST structure for an external-procedure is loaded into the DB, the in-DB ASTs can't be used directly ...
- 10:54 AM Feature #2251: improve the call graph generation
- Code Review branch rev 10494
Except for progress.g, I am fine with the other changes.
I have these questions ab... - 10:10 AM Feature #2251: improve the call graph generation
- Please review the progress.g changes in 10494, they are related to parsing the @CREATE automation-object com-handle-v...
- 10:05 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> > Should I have created graph DB nodes for the call-sites' ASTs, too?
>
> I think so, yes. Th... - 08:26 AM Feature #2251: improve the call graph generation
- > Should I have created graph DB nodes for the call-sites' ASTs, too?
I think so, yes. This would be consistent w... - 08:06 AM Feature #2251: improve the call graph generation
- About mapping the @call-site-key@ string values to token types:
* @includes@: this is to used mark edges with includ... - 03:18 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> 1. External targets have a "node-type" but call-sites don't. I prefer the @type@ approach to be ...
03/16/2014
- 11:50 AM Feature #2251: improve the call graph generation
- Please do a bzr check-in for your work each night, unless there are no changes. It doesn't matter what the state of ...
- 11:46 AM Feature #2251: improve the call graph generation
- Code Review branch rev 10492
I have some questions regarding the "database design". There are inconsistencies:
... - 11:00 AM Feature #2251: improve the call graph generation
- > BTW, do you have a rebase script, so I can rebase?
Not yet. Please go through the process manually based on my ...
03/14/2014
- 04:44 PM Feature #2251: improve the call graph generation
- I've added a good version for review in the 2251 branch. @rules/callgraph/callgraph_lib.rules@ contains details about...
03/13/2014
- 10:48 AM Feature #2251: improve the call graph generation
- Yes, I think you are right.
- 10:25 AM Feature #2251: improve the call graph generation
- About the preprocessor parser problem which can be reproduced with an include file like this one:...
03/12/2014
- 01:43 PM Feature #2256: create tools to analyze the .pphints files
- From #2251 note 46:
> On a side note, there are cases in the server project where there are typos in the preprocesso... - 01:40 PM Feature #2256 (New): create tools to analyze the .pphints files
- 01:17 PM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> Anyway, as far as I know, the preprocessor should not be considering the path of the file that co... - 01:08 PM Feature #2251: improve the call graph generation
- > I think this is because FileScope doesn't keep track of the folder in which the 4GL program resides...
As far as... - 01:01 PM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> As far as I know it does work. Have you found a problem?
What is not working for me is this case:... - 12:38 PM Feature #2251: improve the call graph generation
- > I see that the P2J preprocessor does not support relative names via "..", only relative names starting with the loc...
- 12:08 PM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> If the .pphints problems are not hindering your work, then create a new bug task for this and we ... - 11:15 AM Feature #2251: improve the call graph generation
- If the .pphints problems are not hindering your work, then create a new bug task for this and we will work it later. ...
- 10:44 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> The idea: create a common create_widget_or_object rule that matches the left portion of both the ... - 10:20 AM Feature #2251: improve the call graph generation
- > This might need to be fixed in the post-parse fixups phase: just alter the create_widget_stmt to accept the CONNET ...
- 10:00 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> Using VALUE() for CREATE Automation is not noted in the documentation. Are you sure that this wo... - 09:49 AM Feature #2251: improve the call graph generation
- Using VALUE() for CREATE Automation is not noted in the documentation. Are you sure that this works in the 4GL as CO...
- 09:32 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> > we can't parse the CONNECT TO phrase properly.
>
> Please provide more details.
It's a comb... - 09:23 AM Feature #2251: improve the call graph generation
- > we can't parse the CONNECT TO phrase properly.
Please provide more details.
> More, should the CREATE stateme... - 08:57 AM Feature #2251: improve the call graph generation
- About DDE, COM and OCX call-sites:
# a @COM-HANDLE@ var is initialized via @CREATE expr com-handle-var [CONNECT TO e... - 08:55 AM Feature #2251: improve the call graph generation
- We might have created something like that, but I don't recall it.
However, there is some useful code in operators.... - 07:54 AM Feature #2251: improve the call graph generation
- Greg, is there a way to get the concatenated result from a string expression containing only literals, like @"-H loca...
03/11/2014
- 02:33 PM Feature #2251: improve the call graph generation
- I've added to @testcases/uast/callgraph-tests@ the current tests for callgraph generation. What is working:
* RUN ex...
03/10/2014
- 12:51 PM Feature #2251: improve the call graph generation
- Yes, please do this for any "fixes" that are done later but which are needed for call-graph processing.
- 12:45 PM Feature #2251: improve the call graph generation
- The ASTs for a i.e. @INPUT THROUGH@ stmt is broken and fixed by @annotations/input-output.rules:181@, the "fixup KW_T...
- 09:47 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> This means that processing the ASTs at that time *may* generate very different results than on th... - 09:03 AM Feature #2251: improve the call graph generation
- > 1. Is there any reason why we can't generate the callgraph with the final ASTs, after conversion, and should use on...
- 08:41 AM Feature #2251: improve the call graph generation
- > Can you think of any external linkage type that I have not listed above?
I think all forms of the CONNECT method (...
03/07/2014
- 12:26 PM Feature #2251: improve the call graph generation
- For this current work, include any call-site that can cause an external linkage, including the non-4GL ones.
* All... - 11:34 AM Feature #2251: improve the call graph generation
- With a little hack (replace all "long" types with "string" in the @.graphml@ file) I've managed to open the graph in ...
- 07:49 AM Feature #2251: improve the call graph generation
- Blueprints has a built-in feature to serialize the entire DB to a GraphML file. I've tried importing this into yEd, b...
03/06/2014
- 06:12 PM Feature #2251: improve the call graph generation
- I did a quick code review. It looks really good! Very impressive.
In my opinion, the native Java query access is... - 03:33 PM Feature #2251: improve the call graph generation
- I've committed today's work to the task branch on devsrv01. What I've accomplished (the work is still rough, but it w...
- 12:31 PM Feature #2251: improve the call graph generation
- > A first impression is that is just a "shorthand" access the Pipes APIs and the graph's APIs related to nodes/edges....
- 11:53 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> As such I'd like to avoid mixing Groovy and TRPL (Scala). I guess the key questions:
>
> 1. W... - 11:29 AM Feature #2251: improve the call graph generation
- > After some more thought, I think is better to use the Blueprint APIs, which are implemented by both Neo4J and Titan...
- 10:33 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> I like the abstraction of the graph database.
After some more thought, I think is better to use ... - 10:17 AM Feature #2251: improve the call graph generation
- The plan so far is reasonable. I like the abstraction of the graph database.
> I don't like how PatternEngine.cal... - 07:58 AM Feature #2251: improve the call graph generation
- About the PatternEngine's callgraph dependencies: I think I understand why was needed for the PatternEngine to "know"...
- 06:39 AM Feature #2251: improve the call graph generation
- I don't like how @PatternEngine.callGraphMode@ hard-codes a "callgraph-walking" mode in @PatternEngine@, it doesn't f...
- 02:30 AM Feature #2251: improve the call graph generation
- This note will keep a history of the TitanDB required depencies (and their associated licences).
For blueprints, pip...
03/05/2014
- 01:08 PM Feature #2251: improve the call graph generation
- > Titan supports other storage backends, BerkeleyDB and Persistit both being ACID compliant and embeded in the same J...
- 12:43 PM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> > what looks nice to me is that it allows HBase storage and it can be integrated with Hadoop: we... - 12:20 PM Feature #2251: improve the call graph generation
- > At this time, I'm inclined to go with Titan, especially because the storage backend is not restricted to a propriet...
- 10:53 AM Feature #2251: improve the call graph generation
- Greg Shah wrote:
> Instead, I have long been considering using graph database technology. In particular, I have bee...
Also available in: Atom