Re: Outsanding GLib/GTK+ API bugs

[Owen Taylor <otaylor redhat com>]

<timj redhat com> writes:

> some fixups, partly based on results from the last irc meeting:
> 
> On 29 Aug 2001, Owen Taylor wrote:
> 
> > 59543 Move gbsearcharray to GLib [Owen]
> >   Notes: Owen will move to glib, not install header, and fix GTK+
> >          to not use gbsearcharray.
> >   Puntable: Yes
> >   Breakage: No (binary on some platforms)
> >   Time:  
> 
> you should first just send out a patch for the gtk+ changes and post it
> for review.

Done.
 
> > GObject
> > =======
> > 50877  Rename libgobject to libgruntime??? [X]
> >   Notes:    Most people would like to stay with GObject. Tim
> >             feels strongly that having libgobject and GObject
> >             is confusing.
> >   Puntable: 
> this one is not puntable.

The decision may not be puntable, but the right decision may
well be not to do it at all.
 
> >   Breakage: Yes, lots of fixage.
> 
> the breakage here isn't right, it should involve just changing the
> -lgobject arg to glib's .pc files. if more breakage results, that means
> people are including <gobject/*> in the first place which is wrong and
> needs to be fixed.

And rename the pc files, and fix all uses of them, and of AM_* macros,
and fix up glib-gobject.h, and fix spec files, and reanme bugzila
categories, and so forth....

I've done most of the restructuring so far, and "couple of hours"
is a conservative estimate, even not considering breakages downstream
of GTK+. When you start moving things around and renaming things,
it takes a while to catch everything that got missed.
 
> >   Time:     Couple hours
> 
> probably not, a CVS copy of the directory is required, then cvs remove of the
> gobject/ dir.


> > 55908  Need a function to know if a GBoxed type is reference counte [?] 
> >   Notes:    Consensus was that if you cared for a particular GBoxed type, then
> >             the GBoxed should be a GObject. Some open question about whether 
> >             the is-refcounted parameter to g_boxed_register_static was
> >             necessary.
> >   Puntable: Yes. Worst that happens it that g_boxed_register_static()
> >             is a little more confusing
> >   Breakage: Yes, small amount of fixage.
> >   Time:     0.5 hours
> 
> this one you wanted to find some on to do the is_refcounted argument removal patch.

I wasn't sure here if we had come to a final decision about the
closely related issue of init_func, which if we were changing should
be changed at the same time. I'd have to say it makes me nervous that:

 g_value_set_boxed (&value, NULL);  
 g_assert (g_value_get_boxed (&value) == NULL);

can fail for some boxed types.

If init_func is ever actually used, I think that will could back to
haunt us; and in any case init_func is at best a seldom-useful feature
that I think makes creating your own boxed type a fair bit harder to
figure out.

If we're happy about the status-quo there, then yes, this is just
an easy-to-do mechanical fix at this point.

Regards,
                                        Owen