GNOME Bugzilla – Bug 654852
doesn't build against master glib/gobject-introspection (cannot create 'Variant' instances)
Last modified: 2011-09-15 12:32:51 UTC
Trying to build caribou against master glib and gobject-introspection gives this error: GEN org.gnome.Caribou.Antler.service ../tools/make_schema.py caribou.settings.CaribouSettings Traceback (most recent call last):
+ Trace 227802
maker.create_schemas()
self._create_schema(self.settings, doc, schema)
self._create_schema(s, doc, schemalist)
getattr(setting.gvariant, "print")(False)),
return GLib.Variant(self.variant_type, self.value)
(v, rest_format, _) = creator._create(format_string, [value])
v = constructor(args[0])
return info.invoke(cls, *args)
make[2]: *** [org.gnome.caribou.gschema.xml.in] Error 1 make[2]: Leaving directory `/home/danw/gshell/caribou/data' I believe the handling of GVariant changed recently in glib/g-i (bug 646635), so the code here probably needs to be updated for that. Or maybe an update is needed to pygobject? (Building caribou against master pygobject as well does not fix the error.) To reproduce: install the gnome-shell jhbuild config (http://live.gnome.org/GnomeShell#building), then do "jhbuild buildone glib" "jhbuild buildone gobject-introspection" "jhbuild buildone caribou". Workaround: when the build fails, type "3" to "Give up on module", then run "make" in caribou/data *from outside the jhbuild environment* (so it uses the system glib/g-i), then "jhbuild buildone -n caribou" to finish up the build.
Can't figure out what alternative API is provided now that they took away g_variant_new, opened a bug for it, bug 654859.
I'm getting the following error in Mageia Cauldron: Traceback (most recent call last):
+ Trace 228412
from main import CaribouDaemon
from gi.repository import Gio
from ._gi import _API, Repository
this is due to the usage of pyatspi. It should use gobject introspection instead. API is quite different, and cannot port it myself.
(In reply to comment #2) > I'm getting the following error in Mageia Cauldron: > > Traceback (most recent call last): > This is not the same bug (different component, different problem). Maybe it would be more helpful if this was filed against pyatspi2?
this is fixed now