Re: gtk_entry and "changed" signal in gtk+-2.0

[Owen Taylor <otaylor redhat com>]

Oliver Rauch <oliver rauch rauch-domain de> writes:

> Owen Taylor wrote:
>
> > It's still an "official" signal, on the GtkEditable interface:
> >
> >  http://developer.gnome.org/doc/API/2.0/gtk/gtk-gtkeditable.html
>
> Is it also still an official signal of the GtkEntry interface?

Since, GtkEntry implements the GtkEditable interface, yes, it has
this signal.

> >  http://bugzilla.gnome.org/show_bug.cgi?id=64998
>
> Ok. I see.
> > [ What if the user deleted the text and typed something else in? I don't think
> >   it should be a huge problem for most apps if the contents of an entry are
> >   temporarily in an empty state. ]
>
> This is a problem when you do the following:
>
> static gchar *text;
>
> entry_changed_callback(GtkEntry *entry, void *data)
> {
>   if (text)
>     free(text)
>
>   text = strdup(gtk_entry_get_text(entry));
> }
>
>
> main()
> {
> ...
> text=strdup("Hello"); /* this is the global defined *text */
> gtk_entry_set_text(entry, text);
> ...
> }
>
>
> The problem is that the first callback frees the memory in which
> the text "Hello" is stored, in the second callback the text
> is not available any more.

Certainly, the idea is to fix the double ::changed signal on the
principle of least surprise;  the general rule, however, is that you must
make sure that any parameters passed to a function live until the end
of the function.

                                        Owen