[Wayland-bugs] [Bug 700166] glade fails to build introspection with Gtk 3.9

[glade (bugzilla.gnome.org)]

https://bugzilla.gnome.org/show_bug.cgi?id=700166
  glade | general | unspecified

--- Comment #3 from Tristan Van Berkom <tristanvb at openismus.com> 2013-05-13 08:11:07 UTC ---
(In reply to comment #2)
> (In reply to comment #1)
> > This is a bug with how introspection works.
> 
> Introspection works the same way gtk-doc has always worked.  Yes, it's not
> ideal, but we have had things building without DISPLAY set (i.e. without
> calling gtk_init()) for years.  I don't see a need to change that now, it's
> often not very difficult to change code run in class_init() to code run inside
> a GOnce in _init() or equivalent.

Colin, sure, we've been doing it forever, and exactly, it's not ideal.

It's really not uncommon for class initializers to call API and set
things up for the duration of a class's life cycle, if that requires
a DISPLAY connection, then it requires a DISPLAY connection.

Any patches that I've seen so far addressing the issue with GTK+ 3.9.x,
have been workarounds to do something different at class_init() time.

There is no reason to claim these classes were doing something wrong
before this bug popped up, so we are obviously patching up the platform
for a problem which is deeper, a problem I outlined in comment 1.

FWIW, in Glade, the only reasonable way of hacking around this bug,
without refactoring hundreds of lines of code, is to do something
completely broken (i.e., an empty ENUM ?) when generating the stock
GEnum when there is no DISPLAY available ... just for the sake of
working around this deeper issue.

-- 
Configure bugmail: https://bugzilla.gnome.org/userprefs.cgi?tab=email
------- You are receiving this mail because: -------
You are on the CC list for the bug.