python GnuCash interface to SQL backend

[Sébastien de Menten]

On Saturday, November 15, 2014, Christian Stimming <christian at cstimming.de>
wrote:

> Dear Sébastien,
>
> I really try not to be rude, but a little bit it seems to me as if you
> don't
> accept "no" as an answer here. You asked whether the gnucash developers
> support an alternative SQL access layer written in python from scratch, and
> John's and other answers clearly said "no". What else are you looking for?
>
>
Hello Christian,

I don't think I have asked the question you point here above but I admit I
have asked a lot of other questions (to say the least ;-) ) and I got the
answers of John and Derek right : they do not think it can be successful to
write an external application/library that works directly on the SQL
database due mainly to the complexity of the GnuCash engine (and how it
handles/keeps the data in sync)


> John as already outlined many important aspects about our object model. In
> case you haven't see this so far, some current documentation is also here
> http://wiki.gnucash.org/wiki/C_API and the linked "Entity-Relationship
> Diagram" there, http://wiki.gnucash.org/wiki/images/8/86/Gnucash_erd.png .
>
> Thank you for the links (I had read these pages before starting working on
the project and there were very useful).


> But let's just make this clear: You asked whether your idea would be
> endorsed
> and supported by us, and the answer was clearly a "no". If you like to
> continue your idea, feel free to do so. But just don't repeatedly discuss
> here
> whether we want to change our answer (at this point in time). Thanks!
>
>
As written here above, I have not asked this question.

To recap, I could/can not find a satisfying solution to my needs (work on
GnuCash books from python):
- going the XML way (instead of SQL) has been suggested but suffers from
all the drawbacks from the SQL way (data corruption)
- using the official python binding is painful for me (swig/C code, install
on windows)

I have done in the summer a pure python package that worked for my needs.
As this was very useful, I planned to make it more robust and release it.
Before doing that I wanted to notify you/the core devs and get some advice,
which I got thanks to the time John and Derek spent on this thread (and I
can't be more grateful to them).

I continue to work on piecash and it is going smoothly but I have been
warned of the difficulties lying ahead.

Kind regards

Sebastien