[Wayland-bugs] [Bug 756780] wayland: Fix GtkTreeView's search window

[gtk+ (GNOME Bugzilla)]

https://bugzilla.gnome.org/show_bug.cgi?id=756780

--- Comment #19 from Matthias Clasen <mclasen at redhat.com> ---
(In reply to Jonas Ådahl from comment #18)
> (In reply to Matthias Clasen from comment #8)
> > Needed this fixup on top to avoid critical warnings.
> 
> Yes, I have it amended into the commit locally, but forgot to upload again
> after discovering it.
> 
> (In reply to Matthias Clasen from comment #16)
> > Review of attachment 313620 [details] [review] [review]:
> > 
> > did you look if anything else is using this window type, and will be ok with
> > this change ?
> 
> Initially I searched the gtk+ tree, and it seemed safe. Now I did a more
> broad search, and it seems to be assumed to be a regular toplevel window
> sometimes, and the couple of places I saw this it was a
> GDK_WINDOW_TOPLEVEL). I think to limit mapping it as a subsurface only if
> the window type is GDK_WINDOW_POPUP instead. If we want to be extra safe, I
> assume we can use the private set-subsurface API, but I wonder if any client
> using an utility window as a popup expects the grab behaviour that comes
> with a xdg_popup.
> 
> Would could also change the type hint in GtkTreeView to TOOLTIP, as it is a
> no-input window. This is what I did initially, but it seemed odd to make it
> a "tooltip" when it was clearly not one. Is there any expected semantics of
> an utility type hint? 

Unfortunately, the ewmh does not say much about expected semantics of window
types, beyond some general musings:

http://standards.freedesktop.org/wm-spec/wm-spec-latest.html#idm140200472629520

I think the best we can probably do is to make it clear how _we_ are
interpreting the semantics and offer API to override individual aspects of the
semantics (focus-on-map, subsurface, stacking,...)

> (In reply to Matthias Clasen from comment #17)
> > Review of attachment 313621 [details] [review] [review]:
> > 
> > @@ +11168,3 @@
> >    gtk_window_set_modal (GTK_WINDOW (tree_view->priv->search_window), TRUE);
> > +  gtk_window_set_transient_for (GTK_WINDOW (tree_view->priv->search_window),
> > +                                GTK_WINDOW (toplevel));
> > 
> > Is this related to the key event handling changes ? looks unrelated to me.
> 
> In order to map the window as a subsurface we need to make it transient to
> the parent. Without this, we'd fail to map.

I see. Leave it as-is, then.

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.freedesktop.org/archives/wayland-bugs/attachments/20151022/8657986a/attachment.html>