Re: GSoC07: keep dialogs in libabiword?

Hi Marc,

> However, what if people want the rich text widget, and also a way
> manage/use styles and be able to modify/construct tables? I suspect that
> this will be a _very_ common way consumers will want to use libabiword.
>
> Your solution is: reimplement all style and table dialogs (creating,
> deleting, merging, formatting) again.
>
> Imo, this will result in lots of extra (redundant) work, and lots of
> bugs most likely.

Most of our dialogs are pretty simple, with a few notable exceptions
(styles and lists, to name two).

What I'm concerned about is how the dialogs plug in to AbiWord. In my
mind, we shouldn't have any "dlg" edit methods. The populating of the
dialogs and updating the PieceTable should happen outside of the
AbiWord rich-text widget. Might this mean a possibility of more bugs?
Of course. But it also means that someone could implement dialogs that
we haven't thought about.

Now, if we're talking about sharing glade interface files, I'm fine
with that. They're a nice starting point. And maybe even convenience
code (GInterfaces, perhaps) for populating dialogs that implement
certain interfaces.

What I guess I'm looking for is a good MVC split. The PieceTable is
clearly the Model. The widget is clearly the View. It makes sense both
architecturally and historically to ship the two together. I'm
questioning whether it makes sense to bundle the Controller.

> I really *really* doubt people will be satisfied with just the bare
> libabiword alone. Also I don't see any drawbacks in 'my' approach,
> except the additional library maintenance.

I sincerely and severely doubt this. The places where AbiWord is
gaining real traction are non-traditional devices. Devices with
smaller UIs, different interface guidelines, different toolkits,
different input constraints, and etc. And now we're starting to see
different usage paradigms entirely (cf. Martin's presentation work and
your syntax highlighting/code development environment).

I see one set of hacks for Hildon. And another set for the XO. And
another for presentations. And if we become a code editor, yet another
there. And it really bugs me. There's clearly a need for a versatile
rich-text widget, and we can be that widget. I'm just trying to make
sense of it all and present a modest way to get there.

Best,
Dom
Received on Thu Apr 26 18:01:21 2007