Re: [Swig-devel] Octave patches: modules, minor fixes, test-suite

[Karl W. <kar...@gm...>]

Confirmed - thanks very much Xavier!

Cheers,
Karl


On 27 February 2012 19:31, Xavier Delacour <xav...@gm...>wrote:

> Yes, you're right, I missed the svn add for created file std_list.i.
> Should be fixed now.
>
> Thanks,
> Xavier
>
> On Mon, Feb 27, 2012 at 11:25 AM, Karl Wette <kar...@gm...> wrote:
> > Hi Xavier,
> >
> > Great, thanks!
> >
> > I noticed however that patch #5 hasn't been committed - this patch adds a
> > new file
> > Lib/octave/std_list.i
> > to fix one of the tests. Perhaps you forgot to add it?
> >
> > Cheers,
> > Karl
> >
> >
> >
> > On 27 February 2012 18:12, Xavier Delacour <xav...@gm...>
> > wrote:
> >>
> >> These look good, and are now committed. Thank you, Karl!
> >>
> >> Xavier
> >>
> >> On Thu, Feb 23, 2012 at 7:05 AM, Karl Wette <kar...@gm...>
> wrote:
> >> > Hi Xavier,
> >> >
> >> > I've posted a few more patches to the Octave module - see below for
> >> > the SourceForge link. They include a further small patch related to
> >> > module loading, and a few patches to fix currently-broken tests in the
> >> > test suite. I'd greatly appreciate it if you could have a look at
> >> > them.
> >> >
> >> > Cheers,
> >> > Karl
> >> >
> >> >
> >> > ---------- Forwarded message ----------
> >> > From: SourceForge.net <no...@so...>
> >> > Date: 23 February 2012 15:00
> >> > Subject: [Swig-devel] [ swig-Patches-3492054 ] Octave patches:
> >> > modules, minor fixes, test-suite
> >> > To: "SourceForge.net" <no...@so...>
> >> >
> >> >
> >> > Patches item #3492054, was opened at 2012-02-23 06:00
> >> > Message generated for change (Settings changed) made by kwwette
> >> > You can respond by visiting:
> >> >
> >> >
> https://sourceforge.net/tracker/?func=detail&atid=301645&aid=3492054&group_id=1645
> >> >
> >> > Please note that this message will contain a full copy of the comment
> >> > thread,
> >> > including the initial issue submission, for this request,
> >> > not just the latest update.
> >> > Category: None
> >> > Group: None
> >> > Status: Open
> >> > Resolution: None
> >> > Priority: 5
> >> > Private: No
> >> > Submitted By: Karl Wette (kwwette)
> >> >>Assigned to: Xavier Delacour (xavier98)
> >> > Summary: Octave patches: modules, minor fixes, test-suite
> >> >
> >> > Initial Comment:
> >> > Patch 1 is another minor tweak to Octave module load behaviour. It
> >> > allows the -globals argument (which gives the namespace global
> >> > variables/constants will be loaded into) to take the special value
> >> > ".", in which case global variables/constants are loaded into the same
> >> > namespace as functions, enums, etc, i.e. there is no "cvar" namespace.
> >> > This allows Octave modules to resemble the behaviour of Python modules
> >> > compiled with the "-builtin" option, which no longer supports "cvar".
> >> >
> >> > Patch 2 comments out the declaration of the function
> >> > swig_register_director(). This function has never been defined, and
> >> > having only the declaration can generate warnings if you're compiling
> >> > with e.g. GCC warning flags.
> >> >
> >> > Patch 3 fixes the Octave path in the examples/test-suite Makefiles:
> >> > OCTAVEPATH should be OCTAVE_PATH.
> >> >
> >> > Patches 4--6 are fixes to tests in the test-suite that are currently
> >> > failing. With these three patches, the Octave test-suite and examples
> >> > all work for built-from-source versions of Octave 3.0.5, 3.2.4, and
> >> > 3.4.3, although binaries shipped with some distributions might still
> >> > have problems (see below).
> >> >
> >> > Patch 4 fixes the director_nspace test. With directors activates, it
> >> > looks like some of the generated code (e.g. tables of class members)
> >> > weren't picking up the namespace the class was declared in. This patch
> >> > fixes the few places where this occurs.
> >> >
> >> > Patch 5 fixes the li_std_containers_int test. It looks like this test
> >> > was failing because the Octave module doesn't supply std_int.i, so
> >> > this patch adds this file from the Python module, changing PyObject to
> >> > octave_value, similarly to the Octave/Python std_vector.i
> >> >
> >> > Patch 6 fixes (mostly) the imports test. This test seems to cause
> >> > particular trouble for Octave 3.2.4, which would always seg-fault on
> >> > exit. As far as I could tell, this was happening for two independent
> >> > reasons:
> >> > * Octave was calling the C++ destructor of the oct_file_unload class
> >> > *twice* for one of the loaded modules (imports_b), instead of calling
> >> > the destructor once each for the two loaded modules (imports_a, and
> >> > imports_b). So instead of using a destructor, the module now registers
> >> > at-exit functions with Octave, which point to functions which do the
> >> > cleanup. The names of the at-exit functions must still be unique,
> >> > however, otherwise Octave still calls the same function twice.
> >> > * The second reason Octave 3.2.4 was seg-faulting seemed to be
> >> > improper cleanup of functions left in the symbol table at exit. So the
> >> > module's at-exit functions now calls clear_functions() to clear all
> >> > functions out of the symbol table.
> >> > With this patch, this test now compiles on a built-from-source version
> >> > of Octave 3.2.4, However the binary of Octave 3.2.4 supplied by Debian
> >> > Squeeze still seg-faults, unless the -f option is supplied to Octave.
> >> >
> >> >
> >> > ----------------------------------------------------------------------
> >> >
> >> > You can respond by visiting:
> >> >
> >> >
> https://sourceforge.net/tracker/?func=detail&atid=301645&aid=3492054&group_id=1645
> >> >
> >> >
> >> >
> ------------------------------------------------------------------------------
> >> > Virtualization & Cloud Management Using Capacity Planning
> >> > Cloud computing makes use of virtualization - but cloud computing
> >> > also focuses on allowing computing to be delivered as a service.
> >> > http://www.accelacomm.com/jaw/sfnl/114/51521223/
> >> > _______________________________________________
> >> > Swig-devel mailing list
> >> > Swi...@li...
> >> > https://lists.sourceforge.net/lists/listinfo/swig-devel
> >
> >
>