Skip to main content
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index] [List Home]
RE: [cdt-dev] Problem with call hierarchy for variables

Somehow this looks familiar, just that I have the problem with the type hierarchy.

 

Are you sure that vic_gridcode.h is included from within your eclipse project, or could it be that the include paths for the indexer contain another source for the header file that is external to the eclipse workspace?

 

Jens.

 

From: cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Schorn, Markus
Sent: Dienstag, 23. März 2010 10:50
To: CDT General developers list.
Subject: RE: [cdt-dev] Problem with call hierarchy for variables

 

Please confirm that you have 'x.cpp' in the editor while message says 'x.h' is not in the index,

Markus. 

 

From: cdt-dev-bounces@xxxxxxxxxxx [mailto:cdt-dev-bounces@xxxxxxxxxxx] On Behalf Of Bursian, Achim
Sent: Monday, March 22, 2010 8:14 PM
To: CDT General developers list.
Subject: [cdt-dev] Problem with call hierarchy for variables
Importance: Low

Hello to all,

we have a strange problem with the function "Open Call Hierarchy" for class member variables after upgrading from Ganymede to Galileo SR1 (yes, I know, SR2 is out, but we are still on SR1).

 

Scenario: A file x.cpp is open in the editor, I select a member variable mval and choose "Open Call Hierarchy". The call hierarchy view shows, and after some milliseconds, I get the message "The file 'x.h' is not included in the index. For headers that are never included, or sources that are not part of the build, consider selecting the preference 'index all files'". No call graph is shown.

 

But that header definitely is part of the index. I can see that it gets added to the index when I turn on the CDT traces. And, using "Open Declaration" does work, it correctly opens x.h and shows the correct line where mval is defined.

Additionally, trying to show the call hierarchy for a C++ method of the same class works as expected. The error is only with member variables.

 

Before I add this to Bugzilla, I need some assistance because: So far I was unable to reproduce this with a small example project. I tried for hours already.

The problem only shows in our huge real-world project (but there it shows very consistently). That project takes more than 50min. to index, so it's not easy to handle - and absolutely impossible to attach to a bug report. See the results in the trace below.

To make things worse, there are parts of the project where the call hierarchy does work fine even for member variables. Up to now I can't see why some areas of the code always fail with the Call Hierarchy, while others always succeed. I'm quite confused.

 

C/C++ Indexer: Project 'XXX' (5963 sources, 8409 headers)
    Options: indexer='PDOMFastIndexer', parseAllFiles=false, unusedHeaders=skip, skipReferences=false, skipImplicitReferences=false, skipTypeReferences=false, skipMacroReferences=false.
    Database: 200904704 bytes
    Timings: 3091514 total, 995236 parser, 122241 resolution, 461911 index update.
    Errors: 0 internal, 960 include, 1920 scanner, 7588 syntax errors.
    Names: 780305 declarations, 4217017 references, 70729(1.40%) unresolved.
    Cache[89mb]: 519133159 hits, 7708(0.00%) misses.
Indexer: completed PDOMRebuildTask[3096407ms]

Yes, there are quite a few errors during parsing, but with Eclipse 3.4/CDT 5.x the call hierarchy does work with the exact same source files.

 

Do you have any ideas how to track this down? Any special traces that I should enable? I can't add a bug report without a reproducible example.

 

Thanks for any help, I'm completely lost and frustrated at the moment!

  -Achim

Back to the top