Re: [PATCH] emacs should use GtkFileChooser widget

[Sven Neumann]

Hi,

"Jan D." <address@hidden> writes:

> Some comments on the patch.
>
> If mustmatch_p is true the file is required to exist, but in Emacs
> this does not imply that an open operation is taking place.  It is
> common to open a non-existing file, and the file is saved later.
> Here mustmatch_p is false, so GTK_STOCK_SAVE is inappropriate.  The
> best thing to do is to use GTK_STOCK_OK regardless of mustmatch_p,
> since the GtkFileChooser has no way to implement mustmatch_p the way
> Emacs wants.  The only file dialog that does this correctly is
> Lesstif/Motif.  Mustmatch usually means that Emacs knows the file
> name, but can't find it, so the user is supposed to go to the
> directory where the file is.
> 
> But then there are other problems.  If we are to open a non existing
> file we can't say ACTION_OPEN, because then there is no way for the
> user to enter the name of the file, except to do Ctrl-L.  But Ctrl-L
> is unintuitive so the user might not know about it.  Also, the
> dialog that pops up on Ctrl-L is very annoying (Miles explained that
> very well in another mail).
> 
> So, how about using ACTION_SAVE then?  Well, then there is text in
> the dialog that says "Save", and we are not saving, we are opening a
> non existent file.
> 
> All this is due to the bad design of the dialog.  I now remember why
> I put off adding this new dialog to later.

Yes I noticed this too but I don't follow your argumentation that it's
bad dialog design. IMO the internal emacs API is insufficient here.
There should be a way to distinguish between opening and saving and
mustmatch_p simply is not enough information here.

> Also, I can't get rid of the annoying
> 
> (emacs:21803): Gtk-CRITICAL **: file gtksettings.c: line 447
> (gtk_settings_get_for_screen): assertion `GDK_IS_SCREEN (screen)'
> failed
> 
> (emacs:21803): GLib-GObject-WARNING **: invalid (NULL) pointer instance
> 
> (emacs:21803): GLib-GObject-CRITICAL **: file gsignal.c: line 1726
> (g_signal_handler_disconnect): assertion `G_TYPE_CHECK_INSTANCE
> (instance)' failed
> 
> that GTK writes everytime the dialog is displayed.  I'm sure there
> is a GTK bug in there somewhere, but I have put off finding a
> workaround until the next Emacs release.

I can't reproduce this problem here. What version of GTK+ are you using?


Sven