# HG changeset patch
# User Evan Schoenberg <evan.s@dreskin.net>
# Date 1178219539 0
# Node ID 9f9c486a8aca9951044732a924e6661cc1159b05
# Parent  f0b4efc4a83abb57f2cdd422b1f3ba9119faffbe
flags of 0x00 in oscar_icon_req() can apparently mean 'you should upload your buddy icon' just as flags of 0x41 can.  I'm really unclear as to why Adium with libpurple is getting the 0x00 response while Pidgin with the same libpurple code is getting 0x41 after sending identical data (with identical md5 hashes) for a buddy icon... but that's the way it is. This may fix other conditions in which AIM people claim that their buddy icon isn't being properly set; further insight into what's going on would certainly be appreciated.

diff -r f0b4efc4a83a -r 9f9c486a8aca libpurple/protocols/oscar/oscar.c
--- a/libpurple/protocols/oscar/oscar.c	Thu May 03 17:56:31 2007 +0000
+++ b/libpurple/protocols/oscar/oscar.c	Thu May 03 19:12:19 2007 +0000
@@ -5571,7 +5571,7 @@
 			length = va_arg(ap, int);
 			md5 = va_arg(ap, guchar *);
 
-			if (flags == 0x41) {
+			if ((flags == 0x00) || (flags == 0x41)) {
 				if (!flap_connection_getbytype(od, SNAC_FAMILY_BART) && !od->iconconnecting) {
 					od->iconconnecting = TRUE;
 					od->set_icon = TRUE;