Bug 529522 - Fails to build against octave-3.2.x
Summary: Fails to build against octave-3.2.x
Keywords:
Status: CLOSED ERRATA
Alias: None
Product: Fedora
Classification: Fedora
Component: mathgl
Version: rawhide
Hardware: All
OS: Linux
low
medium
Target Milestone: ---
Assignee: D Haley
QA Contact: Fedora Extras Quality Assurance
URL:
Whiteboard:
Depends On: 529785
Blocks: 525423
TreeView+ depends on / blocked
 
Reported: 2009-10-18 03:49 UTC by Alex Lancaster
Modified: 2009-12-03 05:40 UTC (History)
3 users (show)

Fixed In Version: 1.9-6.fc11.1
Doc Type: Bug Fix
Doc Text:
Clone Of:
Environment:
Last Closed: 2009-11-04 12:25:39 UTC
Type: ---
Embargoed:


Attachments (Terms of Use)

Description Alex Lancaster 2009-10-18 03:49:48 UTC
I'm updating F-11 to octave 3.2.  (bug #525423)  To do this it will be necessary to rebuild mathgl against the new octave, unfortunately in my scratch build, this failed:


g++ -O0 -g -shared -I/usr/include/octave-3.2.3 -o mathgl.oct mgl_octave.cpp -fPIC -I../include -L../mgl/.libs/ -lmgl
mgl_octave.cpp: In member function 'void octave_swig_type::install_global()':
mgl_octave.cpp:1197: error: 'curr_sym_tab' was not declared in this scope
mgl_octave.cpp:1197: error: 'link_to_global_variable' was not declared in this scope
mgl_octave.cpp:1204: error: 'symbol_record' was not declared in this scope
mgl_octave.cpp:1204: error: 'sr' was not declared in this scope
mgl_octave.cpp:1204: error: 'global_sym_tab' was not declared in this scope
mgl_octave.cpp: In function 'void SWIG_Octave_SetModule(void*, swig_module_info*)':
mgl_octave.cpp:2110: error: 'curr_sym_tab' was not declared in this scope
mgl_octave.cpp:2110: error: 'link_to_global_variable' was not declared in this scope
mgl_octave.cpp: In function 'octave_value_list Fmathgl(const octave_value_list&, int)':
mgl_octave.cpp:111773: error: 'curr_sym_tab' was not declared in this scope
mgl_octave.cpp:111773: error: 'link_to_global_variable' was not declared in this scope

full logs can be accessed from here:

http://koji.fedoraproject.org/koji/taskinfo?taskID=1752592

I also looked for this in rawhide/F-12 and I can't see any rebuild against the new octave, but mathgl does depend on a particular API, so I conclude that this must also be causing broken deps in rawhide.

Could you look find out from upstream what, if any, code changes need to be made to update to latest octave?  Thanks.

Comment 1 Alex Lancaster 2009-10-18 04:24:21 UTC
Hmm, looking into the rawhide report and the database, it appears that mathgl was never built for rawhide.  This should never have happened, packages should always be built for rawhide *first*, then for the released version, otherwise you introduce a broken upgrade path.  The review bug should not be closed until a successful rawhide is done.

Comment 2 D Haley 2009-10-18 04:44:42 UTC
The file that is failing is the autogenerated swig cpp file (swig -octave -c++
-I../include -o mgl_octave.cpp mgl.i). Swig needs to be updated to 1.40 in F-11
in order to support octave-3.2.

Current F-11 shows 1.3.39
F-12 shows 1.3.40

Mathgl F-12 Build (succesful):
http://koji.fedoraproject.org/koji/taskinfo?taskID=1752625

===

From http://sourceforge.net/news/?group_id=1645
---
SWIG-1.3.40 released

SWIG-1.3.40 has been release. A summary of changes is as follows:

- SWIG now supports directors for PHP.
- PHP support improved in general.
- Octave 3.2 support added.
- Various bug fixes/enhancements for Allegrocl, C#, Java, Octave, Perl, Python,
Ruby and Tcl.
- Other generic fixes and minor new features.
---

Adding swig owner to CC

Comment 3 Alex Lancaster 2009-10-18 05:32:15 UTC
(In reply to comment #2)
> The file that is failing is the autogenerated swig cpp file (swig -octave -c++
> -I../include -o mgl_octave.cpp mgl.i). Swig needs to be updated to 1.40 in 

I assume you meant 1.3.40 here?

Ah, that same SWIG problem caused problems with some octave-forge packages and required me to disable them, if F-12 SWIG is now fixed, I can probably re-enable them.

Thanks for doing the F-12 build for mathgl, you'll probably need to file a rel-eng bug: http://fedorahosted.org/rel-eng if you want to get this tagged into f12-final.  It should probably go in, a good justification would be that it's needed in rawhide because without it there will a broken upgrade path for those who were upgrading from F-11 with mathgl installed (you can reference this bug).

So, yes upgrading F-11 to new swig would fix a lot of issues: +1 from me for that, as I don't see one pending updates right now:

http://admin.fedoraproject.org/community/?package=swig#package_maintenance/tools/updates

Comment 4 D Haley 2009-10-21 22:38:51 UTC
There is still problems with the package. I will look at this in about 36 hrs:

package: mathgl-1.9-6.fc11.i586 from fedora-updates-11-i386
  unresolved deps:
     octave(api) = 0:api-v32

Comment 5 Alex Lancaster 2009-10-21 22:52:04 UTC
(In reply to comment #4)
> There is still problems with the package. I will look at this in about 36 hrs:
> 
> package: mathgl-1.9-6.fc11.i586 from fedora-updates-11-i386
>   unresolved deps:
>      octave(api) = 0:api-v32  

Don't worry, the next push will fix this, so need to do anything.  I initially pushed octave 3.2.x to updates-testing without mathgl to get some initial testing from the majority of the Octave community since they are unlikely to have mathgl installed.  Meanwhile SWIG 1.3.40 got pushed to stable and I could rebuild mathgl (note the ".1" after the fc11 disttag which is needed when rebuilding for a branch when you don't need to update newer branches):

http://koji.fedoraproject.org/koji/buildinfo?buildID=137591

I have since added it to the update, so the next push to updates-testing should fix everything, so please don't do any rebuilds:

http://admin.fedoraproject.org/updates/F11/FEDORA-2009-10548

Comment 6 D Haley 2009-10-22 10:17:36 UTC
> Don't worry, the next push will fix this, so need to do anything. 

Great. The bugs that fix themselves are the best ones :)

>Thanks for doing the F-12 build for mathgl, you'll probably need to file a
>rel-eng bug:
This has been done, https://fedorahosted.org/rel-eng/ticket/2520 .

Thanks for the good work.

Comment 7 Fedora Update System 2009-11-04 12:25:05 UTC
mathgl-1.9-6.fc11.1, plplot-5.9.5-1.fc11, pfstools-1.7.0-8.fc11, GMT-4.5.0-4.fc11, octave-forge-20090607-15.fc11, octave-3.2.3-1.fc11 has been pushed to the Fedora 11 stable repository.  If problems still persist, please make note of it in this bug report.

Comment 8 Alex Lancaster 2009-11-04 18:20:14 UTC
Bodhi is not smart about which version it was fixed in: fixing it to be 3.2.3-1.fc11

Comment 9 Anonymous account 2009-12-03 02:06:02 UTC
Hi,

I get a very similar problem in the windows build when installing the database 

> pkg install database-1.0.4.tar.gz
odbc_wrap.cpp: In member function 'void octave_swig_type::install_global()':
odbc_wrap.cpp:1197: error: 'curr_sym_tab' was not declared in this scope
odbc_wrap.cpp:1197: error: 'link_to_global_variable' was not declared in this scope
odbc_wrap.cpp:1204: error: 'symbol_record' was not declared in this scope
odbc_wrap.cpp:1204: error: 'sr' was not declared in this scope
odbc_wrap.cpp:1204: error: 'global_sym_tab' was not declared in this scope
odbc_wrap.cpp: In function 'void SWIG_Octave_SetModule(void*, swig_module_info*)':

etc.

Does this package also need to be fixed?

Cheers,

Comment 10 Alex Lancaster 2009-12-03 05:39:17 UTC
(In reply to comment #9)
> Hi,
> 
> I get a very similar problem in the windows build when installing the database 
> 
> > pkg install database-1.0.4.tar.gz
> odbc_wrap.cpp: In member function 'void octave_swig_type::install_global()':
> odbc_wrap.cpp:1197: error: 'curr_sym_tab' was not declared in this scope
> odbc_wrap.cpp:1197: error: 'link_to_global_variable' was not declared in this
> scope
> odbc_wrap.cpp:1204: error: 'symbol_record' was not declared in this scope
> odbc_wrap.cpp:1204: error: 'sr' was not declared in this scope
> odbc_wrap.cpp:1204: error: 'global_sym_tab' was not declared in this scope
> odbc_wrap.cpp: In function 'void SWIG_Octave_SetModule(void*,
> swig_module_info*)':
> 
> etc.


Is that error from a Windows build?  If so, this isn't the place to report it. :)  If you can reproduce this on Fedora with the latest octave-forge in F-12, then please do open a new bug against the octave-forge component.

> Does this package also need to be fixed?

database was removed from octave-forge because I think it was broken, also the note in the .spec file is that it should be in it's own package, but I don't think it anyone has gotten around to doing that.


Note You need to log in before you can comment on or make changes to this bug.