Re: GTK 3 support status

[Tristan Van Berkom <tvb gnome org>]

On Tue, Dec 6, 2011 at 1:15 AM, John Lindgren <john lindgren aol com> wrote:
> Hi Tristan,
>
> This makes a bit more sense now.
>
>
> On 12/05/2011 09:27 AM, Tristan Van Berkom wrote:
> > On Mon, Dec 5, 2011 at 11:18 PM, John Lindgren<john lindgren aol com>
> >  wrote:
> > > It looks to me as though there are 3 separate problems contributing here:
> > >
> > > 1. GtkLabel does not take into account gtk_widget_set_size_request() when
> > > reporting its "natural" size:
> > > https://bugzilla.gnome.org/show_bug.cgi?id=662043
> >
> > Not a bug.
> >
> > No widgets should ever peek at the explicit user set size request to
> > implement the size negotiation vfuncs.
> >
> > Those are handled in GtkWidgetClass.adjust_size_request() base class
> > implementation.
>
>
> So using gtk_widget_set_size_request() as a way to set the width of a label
> is no longer valid in GTK+ 3.  Fair enough.

That is true, GtkLabel in GTK+2 did do some sneaky things like peeking at
the user set size request, it no longer does this.

gtk_widget_set_size_request() should still succeed with the expected behaviour
of setting the minimum size of a widget, however it does not limit the size
of any widget.

> > > 2. Nor does it take into account gtk_label_set_width_chars():
> > > https://bugzilla.gnome.org/show_bug.cgi?id=665560
> >
> > It does, width-chars, as always, only effects the minimum size, not
> > the natural size.
>
>
> I get it now.  You set the minimum width of a label with
> gtk_label_set_width_chars() and the natural/default width with
> gtk_label_set_max_width_chars().
>
> The documentation is very misleading; actually, the entire documentation for
> gtk_label_set_width_chars() consists of the sentence fragment, "Sets the
> desired width in characters of label to n_chars."  I do not think that most
> people would make the connection between "desired width" and "minimum
> width".
>
> Likewise, the documentation for gtk_label_set_max_width_chars() is, "Sets
> the desired maximum width in characters of label to n_chars."  I do not
> think that most people would make the connection between "desired maximum
> width" and "natural/default width".

I couldn't agree more, and I'm sorry for lack of time and patience for
documenting
GtkLabel properties better.

Would you like to re-phrase that documentation yourself and send it back
to the list ?

I'll take that and correct anything that might be inaccurate and send
it back to this list... (perhaps we should CC gtk-devel-list in this case).

Others will have a (short) chance to interject... and if there are no objections
to the new documentation within a short time I'll go ahead and commit
that to GTK+ so that next time around the online docs will be better.

Reasonable plan ?

> > > 3. GtkWindow ignores gtk_window_set_default_size() and instead goes to
> > > its
> > > "natural" size if the window is not resizable:
> > > https://bugzilla.gnome.org/show_bug.cgi?id=665596
> >
> > As I mentioned in the previous mail, I believe this has *always* been the
> > case,
> > GtkWindow always shrinks down to minimum size (or now in GTK+3 'natural
> > size')
> >
> > This is how you get the shrinking dialogs when closing the expander in
> > them
> > (for non-resizable windows).
> >
> > However, it could be considered a bug, one would expect GtkWindow to make
> > an attempt to honor a user set default size for a non-resizable window.
> >
> > Perhaps it's even a regression from GTK+2, not so sure.
>
>
> I think it should be considered a bug; however, I do not care so much about
> gtk_window_set_default_size() now that I can set the size of the label
> directly.  Using gtk_window_set_default_size(), you might recall, was your
> suggestion from an earlier mail:

Yes, whether it is a regression or not I think it's a valid bug, apis in GTK+
should definitely make an attempt at succeeding ;-)

However, GtkWindow itself is a very tricky object to not break (on X11
platform it has to deal with this 'window constraint' api to cooperate
with the X server), it's possible that it just doesn't make sense to implement
this properly.

In which case the fix might be adding documentation to
gtk_window_set_resizable()
to indicate that the window will always be forced to the natural size
requests of its
content, regardless of any previously set default size.

> > So I'm attaching my version of your test case, actually except for
> > the label being somehow severely broken here, your code works
> > nicely... I added a line that says gtk_window_set_default_size()
> > and I think that will fix all your problems.
>
>
> Thanks for your help.  I will close #662043 and #665560 but add a comment
> that the documentation is wrong and needs to be updated.
>
> -- John