Mercurial > pidgin.yaz
changeset 15787:b25acae693cd
Change connection.c:gaim_connection_error() to still work even if the
message parameter is NULL. There's no reason to error-out of that
function just because some yuppie didn't pass us an error message.
But we still log a critical message, so those of you at home running
with fatal_criticals will still crash.
I noticed this because Jabber isn't handling <stream:error/> correctly
right now. Anyone else notice this? The code is written to look for
"stream:error", but our xmlnode stuff just sees "error" (even though
the "stream:error" format is sent over the wire). I think it's
because of the libxml change.
To reproduce the problem, sign onto the same resource twice and note
that the error message that Gaim gives you sucks.
author | Mark Doliner <mark@kingant.net> |
---|---|
date | Tue, 13 Mar 2007 07:33:08 +0000 |
parents | e926951e61fe |
children | d94d3d0f1149 |
files | libpurple/connection.c |
diffstat | 1 files changed, 5 insertions(+), 1 deletions(-) [+] |
line wrap: on
line diff
--- a/libpurple/connection.c Tue Mar 13 07:19:06 2007 +0000 +++ b/libpurple/connection.c Tue Mar 13 07:33:08 2007 +0000 @@ -434,7 +434,11 @@ GaimConnectionUiOps *ops; g_return_if_fail(gc != NULL); - g_return_if_fail(text != NULL); + + if (text != NULL) { + g_critical("gaim_connection_error: check `text != NULL' failed"); + text = _("Unknown error"); + } /* If we've already got one error, we don't need any more */ if (gc->disconnect_timeout)