Re: [gnome-db] Re: dwi and libgda

[Andrew Hill <andru treshna com>]

Rodrigo Moya wrote:

> On Wed, 2002-05-01 at 03:20, Andrew Hill wrote:
>  
>
> > > Several remarks:
> > > 1) I'm trying to create a graphical designer for DWI (spec'd in DWI
> > >  itself!  I'm trying to make DWI that flexible/powerful).  
> > >
> > > 2) I *really* would like to see glade itself split into a bunch of
> > >  widgets/libraries, so that I could embed parts of glade in *my* app.
> > >  (e.g. so that I could embed the glade hierarchical viewer/picker
> > >  in my app).  This may be a tall order but ...
> > >
> > >  ... but it may be philosophically cleaner to add glade support to
> > >  application X, rather than adding support for X into glade ...
> > >
> > >  Glade  should be good at laying oput widgets, and no more.  I'd
> > >  rather add glade to my DB designer, than to force my bloat onto
> > >  glade.
> > >
> > >      
> > agree. I have no need for extra database controls etc for glade and i use it
> > to as my database designer. It would confuse the user. I only use
> > the basic widget set nothing more. The thing i want from glade in the long
> > run is been able to embedded it into java applications some how.  may
> > mean a rewrite compeltely in java :(
> >
> >    
> well, it would confuse the user if those widgets are not available in
> Glade, but if they are (which they are, at least 50%), they won't
> confuse, since it will be just another widget set in the Glade designer,
> nothing more.
I deal with all the database bindings in my code, not the widgets so from
my library more than one version of a GtkEntryBox, which theoritiy does
the same. or so the idea goes anyway....

> Anyway, for the time being, I think we should concentrate on the engine
> (ie, the non-UI stuff), and then we can make tests to see what's the
> best way for the widget stuff.
>
> So, I think we should start integrating the engine into libgda now.
> Linas, how does that sound? do you agree? If so, please tell me when
> you've converted them to GObject's, the files involved, and I'll try to
> make the starting work of the integration, to speed things up.
>
>
>  
>
> > > 3) I've begun converting the DWI internal structures into glib2
> > >  GObjects.  The reason for this is somewhat subtle --- and would
> > >  cloud this email.   But the idea is "wouldn't it be neat to have
> > >  gobjects that know how to manipulate other gobjects (and hook
> > >  them to databases)?"  This way, I can bootstrap up ... use DWI
> > >  to design a DWI designer ...
> > >
> > > 4) The ultimate -- use DWI to power a web interface (using PHP???)
> > >  That way, I can have both a glade and a web interface for the cost 
> > >  of one.  But this idea is still out there; haven't worked on it yet.
> > >
> > >      
> > I've been wokring on this for month and a half.  I've got it so i can 
> > parse a glade
> > XML file and generate all the HTML for it dynamically and lay it out just
> > like it would look in GTK.  Working on web server now for processing. 
> > Difficult
> > bit is usercallbacks. when in gtk you make a shit load of callbacks (cross
> > dependent widgets etc), but callbacks are hugely damaging in the world 
> > of hmtl.
> > But yes it is all possible.  I'd suggest you use CGI instead of PHP 
> > though php is
> > very nice.
> >
> >    
> that sounds really nice, really really nice.

my gtk themes dont work on it though. :(
and no tree widget.

> cheers
>
>
>