Bitlbee WHO doesn't report account away status, breaks Quassel away-on-detach

[digitalcircuit]

When sending WHO replies Bitlbee does not check the away settings for the local IRC user. If I set myself as /away, any future replies will wrongly tell the IRC client that I'm not away. See irc_send_who in ​irc_send.c, line 306

This breaks Quassel IRC's "auto-away on detach" functionality. The Quassel core (bouncer) sets /away when the client disconnects, but whenever a WHO reply is received indicating the account nick is no longer away (H instead of G), the core assumes Bitlbee cleared the away status. As a result, the Quassel core won't send the appropriate AWAY\r\n upon reattaching the client. See ​Quassel coresessioneventprocessor.cpp, line 766

Perhaps use a form of set_getstr(&irc->b->set, "away", away) to check what's set by irc_cmd_away in ​irc_commands.c, line 731?

strace output of Bitlbee (strace -e write,read -p $BITLBEE_PID -t -s 400):

16:12:36 read(11, "AWAY :I'm here, just testing Bitlbee\r\n", 512) = 38
16:12:36 write(11, ":localhost 306 digitalcircuit :You're now away: I'm here, just testing Bitlbee\r\n", 76) = 76
16:13:30 read(11, "WHO &bitlbee\r\n", 512) = 33
16:13:30 write(11, ":localhost 352 digitalcircuit &bitlbee quassel localhost localhost digitalcircuit H :0 Digital\r\n:localhost 352 digitalcircuit &bitlbee root localhost localhost root H :0 User manager\r\n:localhost 315 digitalcircuit &bitlbee :End of /WHO list\r\n", 454) = 454

expected reply (G not H):

16:13:30 write(11, ":localhost 352 digitalcircuit &bitlbee quassel localhost localhost digitalcircuit G :0 Digital\r\n:localhost 352 digitalcircuit &bitlbee root localhost localhost root H :0 User manager\r\n:localhost 315 digitalcircuit &bitlbee :End of /WHO list\r\n", 454) = 454

[dx]

Are you using 3.4.1 as you indicated in the version field, or git master? Because the line you highlighted there wasn't in 3.4.1.

[digitalcircuit]

Ack, my bad - I was using 3.4.1, but referencing the source code from git master.

As requested on IRC, I retested with the latest git master nightly Ubuntu package (2015-12-6, for some reason nothing later is available), and the issue persists.

[digitalcircuit]

I think I've found and fixed the problem, so I've put up a ​pull request on Github. If there's anything amiss, I'll try to amend it.

[dx]

That was merged and this is fixed.