[PATCH] invoice creation with python-bindings and value of stability patch

[Mike Evans]

On Friday June 18 2010 18:31:53 Mark Jenkins wrote:
> Hi to all again,
> 
> Attached is patch,
> python_bindings_business_invoice_support_and_examples.patch that
> enhances the python bindings to include support for creating, posting,
> and paying invoices.
> 
> Apply with patch -p0.
> 
> The recent discussion re invoice importer inspired this.
> 
> The patch is dependent on three prior patch submissions,
> nice_example_python_scripts.patch,
> python_bindings_gncnumeric_and_accounttype_constants.patch, and
> python_bindings_stability_without_guile.patch
> in that order, which I have re-attached for convenience.
> 
> 
> Because this is a feature enhancement, I don't expect this could be
> considered for application to trunk until 2.4 is branched away -- I know
> that feature freeze was long ago and that I'm late to the party.
> 
> But, I do think this invoice creation patch illustrates the value of
> python_bindings_stability_without_guile.patch as a bug fix that should
> make it into 2.4, even with the flaw that the module linking there comes
> with a portability warning.
> see:
> http://lists.gnucash.org/pipermail/gnucash-devel/2010-June/028760.html
> http://lists.gnucash.org/pipermail/gnucash-devel/2010-June/028790.html
> http://lists.gnucash.org/pipermail/gnucash-devel/2010-June/028815.html
> I don't though if the kind of changes Christian was discussing to avoid
> those linking portability warnings would be appropriate prior to 2.4. On
> balance, I'd rather 2.4 ship with python bindings that are less portable
> than generally crash prone.
> 
> I started working on this invoice creation stuff first and immediately
> started to hit major stability problems (seg faults) that I'm certain
> had nothing to do with those particular enhancements. So I wrote
> python_bindings_stability_without_guile.patch which made those problems
> go away. Then I was able to actually test and finish this invoice
> creation stuff.
> 
> I could re-work things so that
> python_bindings_business_invoice_support_and_examples.patch can be
> applied first (to show stability issues with current init process), and
> python_bindings_stability_without_guile.patch applicable second to help
> illustrate that that stability patch really does make a difference
> between terrible crash and invoice create success. (and by extension,
> the stability patch should make a difference even if applied on its own
> to provide a better environment for the bindings already in trunk)
> 
> I can file a bug report too. Still can't pin down exactly why the
> current python bindings init process (in trunk) is flawed, but I
> speculated before that python and guile can't play nice together. Could
> it be possible that there are symbol conflicts between the two that the
> linker doesn't tell about?
> 
> As always, my branch can be followed here:
> http://svn.parit.ca/gnucash/branches/python-bindings/
> 
> 
> 
> Mark Jenkins
> ParIT Worker Co-op
> 
> cc Mike Evans
> cc Christian Stimming
> cc Christoph Holtermann
> cc fellow ParIT members
I've tried these patches and they seem to work fine.  A few chunks failed so  
some manual editing was required.  Not done any extensive testing as yet 
though, but business data now load and save.

Great stuff Mark.  


-- 
GPG Key: 1024D/050895C2
Keyserver: http://pgp.mit.edu/          
Search String: 0x050895C2