Re: pending stuff in havoc-patches

[Havoc Pennington <hp redhat com>]

Federico Mena Quintero <federico@helixcode.com> writes:
> >  Outstanding questions so I can finish this:
> >  
> >    - Should SHORT_LIVED, PERSISTENT, etc. be GtkWindow flags? 
> >      (These are semantic "types" of window, so we can make
> >      GTK_WIN_POS_MOUSE etc. user-configurable)
> 
> Isn't this policy?  Or is it part of the wm-spec?
>

GTK+ contains plenty of UI policy, so I guess the question would be
"is it too much policy." Owen/Tim could give their view.

I think all gtk_ctree_set_expander_style () crap is broken, and a
gtk_dialog_set_position (WIN_POS_MOUSE) is exactly that (something
programmers should not decide).

So, if it's programmer-settable at all, it should be settable with
semantic flags, not WIN_POS_MOUSE/WIN_POS_CENTER.
 
> I think leaving the hiding/destroying stuff up to the dialog is bad
> API design.  It should be left up to the user.
>

Can you give a reason for this?
 
> This defines UI policy.  This does not belong in GTK+.
>

Lots of things in GTK+ define UI policy. I don't think this
distinction between GTK+ and gnome-libs makes very much sense.

However if you can clarify which policy should be in GTK and which in
gnome-libs, then we could determine where MessageBox belongs.

(I don't see why all GTK apps should reinvent inconsistent UIs, I
think there should be basic policy in GTK itself. And in fact there
_is_ basic policy in GTK itself. So, the "policy vs. mechanism"
dividing line with gnome-libs is not very clear to me.)

Havoc