It's been quite a while since I've looked at this bug as I no longer hit
it myself due to using a new linux distro with a full GIO
implementation, plus I no longer regularly receive the kind of emails
which trigger this...
Please read back to comments 67 through to 70 (
https://bugzilla.mozilla.or
(In reply to Karl Tomlinson (:karlt) from comment #68)
> Ah, yes. That explains a lot.
> http://git.gnome.org/browse/gconf/log/gconf/gconf-client.c?h=gnome-2-20
>
> I also wonder whether gconf_escape_key() should be used here.
> http://developer.gnome.org/gconf/stable/gconf-gconf.html#gconf-escap
(In reply to era eriksson from comment #59)
> So how about limiting the scope of the fix even further? It seems that the
> majority of the complaints are about patterns where the plus sign follows
> straight after the colon. While there are no registered URL protocols which
> start with a plus si
I've just recompiled using the unpatched TB 16.0b4 with --disable-gio and I get
no errors at all (not via stdout nor popup).. Therefore I'm recompiling now
with --enable-gvfs set explicitly to see if that makes a difference.
My expectation is that the unpatched source will produce the GConf error
I'm afraid that I've tested this patch against 16.0b4 and it does not
resolve the issue on Ubuntu 10.04 which is exactly what I expected,
having previously spent a lot of time tracing this stuff to establish
where the popup dialog comes from.
I honestly don't see how anyone gets a popup with the e
(In reply to jhorak from comment #52)
> Hm, I can't say it's a fault of --enable-gio, this seems to be related to
> g(nome)vfs. There's no libgio in RHEL5 (usually located in glib2 package)
> and g(nome)vfs backend is used during compilation and I see same error
> message.
> Affected versions:
> -
(In reply to Magnus Melin from comment #48)
> FWIW, after bug 794378 is fixed --enable-gio will be the default (bug
> 713802).
Hmm - Thanks for that info... If --enable-gio becomes the default then this bug
may become much more prevalent.
>From my findings, earlier versions of linux (e.g. running
(In reply to sienkiew from comment #45)
> I don't understand why anybody would care about svn+ssh URLs in the email.
Agreed. I can't imagine anyone caring much about svn+ssh URIs and they won't
work in a GNOME environment anyway because of this very issue.
> That is, when using GConf for the loo
Curses! The SVN folks have foiled my plan! :-D
I currently work for a very large global corporation who can be
extremely anal about standards compliance (although usually with the
best intentions). So, I'm approaching the Thunderbird team here with an
open mind that someone may wish to fight to th
I think this is all a bit dependent on the version of GNOME is in use. I
am running Ubuntu 10.04 LTS which still uses GNOME 2.30.2
Looking at the git master for GVFS the latest code no longer appears to have
any reliance on gconf. I guess that the GIO implementation in GNOME 2.30.2 is
somewhat t
Okay - I've confirmed my theory:
I've built a couple of alternative versions of the libgiogconf.so
module.
All my edits/tests were made to gapplookupgconf.c in the function
get_default_for_uri_scheme():
My first test was to call gconf_client_set_error_handling(client,
GCONF_CLIENT_HANDLE_NONE)
I'm pretty certain that my research has shown that this problem is
actually a result of Thunderbird built with support for GIO (disabled by
default and not enabled in the mozilla release builds)
This is kind of the opposite to comment #19 (
https://bugzilla.mozilla.org/show_bug.cgi?id=541130#c19 )
I've been doing a load of additional research on this bug because it's
causing me daily annoyance...
In response to Sandy Knight: This issue will not corrupt your profile.
I've hooked up gdb to thunderbird and in combination with the source
code and a bit of a crash course in reading c++ I feel I
13 matches
Mail list logo
