Mercurial > pidgin
diff src/connection.c @ 13914:3ae8a3935406
[gaim-migrate @ 16414]
First stab at trying to fix the MSN http connect method. It still
doesn't work, and I'm not sure why, but it gets a lot farther in
the signon process now.
For those unfamiliar with the issue, the MSN http connect method
stopped working after all the non-blocking I/O changes. The http
connect method is apparently used by lots of people behind silly
firewalls and stuff, and therefore we really shouldn't release
Gaim 2.0.0 without it working, because people will complain.
The two main problems were
1. The outgoing message queue was removed in favor of buffering all
data to one large buffer. This sounds good in theory... but apparently
each message sent to and from the server has a "SessionID" in the
HTTP header. Every message we send should use the same SessionID as
the last packet we received from the server. So basically you can't
put two messages into the outgoing buffer at the same time because
you don't have the correct SessionID to use for the second message.
You have to wait until you get the reply from the server.
2. There were some strange buffer problems with using the wrong
variable when trying to combine the header+body into one buffer
before sending the message.
I also fixed a small memleak or two, added some comments, and
tried to clean up the code a little.
committer: Tailor Script <tailor@pidgin.im>
author | Mark Doliner <mark@kingant.net> |
---|---|
date | Mon, 03 Jul 2006 20:39:04 +0000 |
parents | 702107dd58f1 |
children | 9c0885611b27 |
line wrap: on
line diff
--- a/src/connection.c Mon Jul 03 18:28:13 2006 +0000 +++ b/src/connection.c Mon Jul 03 20:39:04 2006 +0000 @@ -217,11 +217,8 @@ gaim_account_set_connection(account, NULL); - if (gc->password != NULL) - g_free(gc->password); - - if (gc->display_name != NULL) - g_free(gc->display_name); + g_free(gc->password); + g_free(gc->display_name); if (gc->disconnect_timeout) gaim_timeout_remove(gc->disconnect_timeout);