Re: Bug in tree selection behavior?

[Mark Jones <mark jones106 verizon net>]

I just wanted to say that I have seen this same behavior in gtkmm.  So I
guess it is the underlying gtk object at "fault".  If "fault"
appropriate.

I had not reported it yet though as I had put that program aside for
something of higher priority and did not have time to examine whether I
was doing something incorrect or not.  I am seeing it in a drum machine
program I started writing, where I have a list of all the samples
available, and when you click on one you can hear a preview.  Well when
you click on one for the first time, it also plays the one at the top of
the list (position 0) as well as the one you have clicked on.

I'm curious to know the solution to this too.  Because doing something
tricky in code (like a static variable that tells your signal handler
whether it has been called before or not to avoid doing anything the
first time through) seems like a work around.

On Sat, 2003-01-25 at 12:01, gtk-list-request gnome org wrote:

> Message: 3
> Subject: Bug in tree selection behavior?
> From: "Andrew S. Townley" <atownley eircom net>
> To: gtk-list gnome org
> Organization: 
> Date: 25 Jan 2003 14:47:32 +0000
> 
> 
> --=-W71rs5jsRUEybKJ0Ca+a
> Content-Type: text/plain
> Content-Transfer-Encoding: quoted-printable
> 
> Hi.  I've just built the following from CVS/HEAD: esound, gnome-xml,
> libxslt, gtk-doc, glib, gob, linc, atk, gnome-common, pango, libIDL,
> ORBit2, intltool, bonobo-activation and gtk+ so that I can see if the
> behavior has been changed.  It hasn't changed from the version that
> comes stock with RedHat 8, so I'm wondering if anyone else things the
> following is a bug.
> 
> PROBLEM:  selection changed signal handler called twice during initial
> tree selection.
> 
> I added the following to testtreeview.c to see what the current behavior
> was:
> 
> diff -r1.33 testtreeview.c
> 4c4
> <=20
> ---
> > #include <stdio.h>
> 599a600,614
> > static void
> > sel_foreach (GtkTreeModel *model, GtkTreePath *path,=20
> >               GtkTreeIter *iter, gpointer data)
> > {
> >       gchar* s =3D gtk_tree_model_get_string_from_iter(model, iter);
> >       printf ("iter =3D '%s'\n", s);
> >       g_free (s);
> > }
> >=20
> > static void
> > selection_changed (GtkTreeSelection *treeselection, gpointer data)
> > {
> >       gtk_tree_selection_selected_foreach(treeselection,
> >               sel_foreach, NULL);
> > }
> 621a637
> >   GtkTreeSelection *sel;
> 760c776,781
> <  =20
> ---
> > =20
> >   sel =3D gtk_tree_view_get_selection(GTK_TREE_VIEW(tv));
> >   g_signal_connect (sel,=20
> >               "changed",=20
> >               G_CALLBACK (selection_changed),
> >               tv);
> 
> Basically, it just prints the selection path each time the "changed"
> signal is fired.  However, when running this program and selecting row 6
> followed by row 4, I get the following output:
> 
> linna$ ./testtreeview
> Running automated tests...
> Passed.
> iter =3D '0'
> iter =3D '6'
> iter =3D '4'
> 
> Since there are no rows visibly selected in the tree view, I would
> expect the changed signal to be fired exactly twice, but it is *always*
> fired for row 0 with the first selection in the tree.  This makes it
> very difficult to write handlers for selection events when you actually
> want some sort of processing to occur with the data in a specific row.
> 
> If other people think this is a bug, I'll enter it in bugzilla.
> 
> Thanks in advance for any comments/information,