[PATCH] python bindings stability by going without guile

[Geert Janssens]

On Sunday 27 June 2010, Christoph Holtermann wrote:
> Hello !
> 
> Am 16.06.2010 17:20, schrieb Mark Jenkins:
> > Attached is a patch that appears to improve stability for the python
> > bindings.
> >
> > Instead of initializing guile and using the gnucash module loading
> > mechanisms, I directly link to all needed modules and just use the
> > direct initialization functions in pretty much the same way that
> > Cutecash does.
> >
> > I've been doing some work with the bindings for business objects
> > (invoice creation), for which I'll be submitting additional patches
> > later today. Applying this stability patch has made the difference
> > between seg faulting and things actually working.
> >
> > Christoph, I wonder if this helps with the stack blow out problems
> > you've been having?
> 
> Just now I tried applying the patch. I am not familiar with doing that.
> Shortly : It seems to work. I didn´t get a crash until now ( without
> excessive
> testing).
> 
> I´d like to know if I did things halfway correctly :
> I tried
> patch -i python_bindings_stability_without_guile.patch
> 
> in the gnucash directory of Rev 19255.
> 
If you update your svn working directory to at least 19294, you will find this 
patch applied already. You could continue your testing from there if you like.

> It did not find the file "src/optional/python-bindings/gnucash_core.i".
>  When I told it  to use "./src/optional/python-bindings/gnucash_core.i" it
>  did though.
> 
> I couldn´t make it to accept Makefile.am in the same way. I applied the
> changes by vim then.
> 
> In the directory of the python-bindings I did
> 
> make clean
> make
> sudo make install
> 
> It seemed to work. It´s surely not completely the right way, is it ?
> 
Seems about right to me.

> I didn´t understand Rev 533 in the patchfile for I only know of Rev 19255.
> 
> Could you explain ?
> 
I don't know about this part.

Geert