Re: [gnome-db] What about the new realice of gnome-db?

[Vivien Malerba <vmalerba gmail com>]

2006/1/5, Rodrigo Moya <rodrigo gnome-db org>:
> On Thu, 2006-01-05 at 09:24 -0600, Daniel Espinosa wrote:
> > One question...
> >
> > If the chages made for Vivien, ADD new API to libgda; is it necesary
> > to jump to 3.0 series? Is it not too fast?
> >
> > I think a good product need to be stable enough to usable in
> > production; even in API/ABI.
> >
> that's why I proposed putting Vivien's changes before 2.0, and after
> that, try to not break the API in a while.
>
> Vivien, what do you think? Big API changes have proved to be a problem
> (bindings, porting apps, etc), so if we could reduce it, it would be
> great.

The changes I've made do break the current API/ABI.

The reason is that it's a reorganization of the objects between libgda
and libgnomedb. I could not simply only ADD to the existing API even
though I understand the need for stability.

Basically the reorganization bas the following benefits:
- less lines of code
- more features
- easier to understand and more coherent API
- makes room for future improvements
- better testing framework

This reorganization is a necessary step as I discovered while I tried
to add more features to libgnomedb, and because of historical reasons
(as I first merged the mergeant functionalities into libgnomedb).

I don't expect anymore such big changes afterwards (understand: we can
have long stabilization periods afterwards).

The current stage is that libgda works well, and that libgnomedb
compiles and I expect it to work quite soon.

If you really want to postpone the 2.0 release by a few months, then
we can make those modifications into it; that would be worth.

Vivien