More KMyMoney peeking: is our file format clearly documented?

[Josh Sled]

On Thu, 2004-06-24 at 11:13, Derek Atkins wrote:

> While I agree that's an issue, the reason I said that I understand why
> they don't is that about 60-70% of the code in gnucash is UI code, and
> much of that has reliance on gnome/gtk.  The lines between UI and
> non-UI have been blurred significantly (certainly outside the engine).

Yes; as a contributer to this problem, I'm looking forward to finding
ways to fix it.  The G2 port needs to be completed, first, however. :/


> I also believe that there are TOO MANY shared libs in gnucash.  Why
> are gnc-modules, the engine, and app-utils all in separate libraries?
> Nobody else is using gnc-modules (and why should they!?)..  And
> app-utils doesn't work without the engine..  I see no point in that
> separation.

Yes.  While we're on the arch. topic, what else should be removed?

* engine subsumes app-utils
* delete gnc-modules
* g-wrap moves to swig
* reports move from guile-generated to embedded-guile.
  * embedded generic script?
[* remove Scheme entirely]



> > In particular, the goal of the gnucash "engine" has (always) been to 
> > provide a GUI-neutral, indeed, OS-neutral way of accessing financial 
> > data through a well documented, supported API.   This GUI neutrality
> > is what allowed the Motif and GTK versions of GnuCash to co-exist,
> > and even helped start a KDE port, back when, until it withered.

Yes, but as mentioned there's too much useful code in the higher-level
and directly in the GUI code.  We can refactor it down into an
abstracted engine and people have been actively doing that.  But until
then it's hard to say that it's practically possible to write a QT/KDE
front-end.


> Sure, but the engine is a fairly small portion of the code..  The only
> potential reason I can think of for someone NOT to use it is that it's
> HARD to just get the engine code.  Playing devil's advocate here...
> It depends on glib (which is a gnome library)..  

Well, glib is pretty unencumbered beneath it; if depending on glib is a
problem ... well ... you probably have bigger fish to fry.


> Back to reality, and having dealt with the code myself, I can
> certainly understand the frustration with trying to re-build the qt
> frontend.  Gnucash just has too much crap in it.  :)

Yes.  We need to get the Gnome2 port done ASAP, and then start the
distillation of GnuCash.  Smaller, Faster, Leaner.  Simplicity.

...jsled

-- 
http://www.asynchronous.org/ - `a=jsled; b=asynchronous.org; echo ${a}@${b}`