**** BEGIN LOGGING AT Tue Apr 20 03:00:04 2010
Apr 20 17:07:55 <denkenz> akiniemi: 
Apr 20 17:07:56 <denkenz> cc1: warnings being treated as errors
Apr 20 17:07:58 <denkenz> gisi/netlink.c:82: error: redundant redeclaration of ‘if_nametoindex’
Apr 20 17:08:00 <denkenz> /usr/include/net/if.h:194: note: previous declaration of ‘if_nametoindex’ was here
Apr 20 17:16:16 <akiniemi> denkenz: hmph
Apr 20 17:16:56 <denkenz> akiniemi: Not to mention the blatant style violation:
Apr 20 17:16:57 <denkenz> extern unsigned if_nametoindex (char const *name);
Apr 20 17:17:09 <denkenz> no space before parens :)
Apr 20 17:19:55 <akiniemi> denkenz: heh, that style violation comes from a copy-paste ;)
Apr 20 17:20:10 <denkenz> Doesn't matter
Apr 20 17:21:32 <akiniemi> denkenz: wonder why I don't get that warning here
Apr 20 17:22:17 <akiniemi> pessi: ping
Apr 20 17:26:12 <denkenz> I don't really get why they're incompatible
Apr 20 17:26:22 <denkenz> seem to have the same datatypes
Apr 20 17:29:31 <akiniemi> denkenz: I know, we'd need Pekka's brain here. This came in his refactoring patch.
Apr 20 17:30:16 <denkenz> my vote is to remove that for now, as it compiles cleanly then
Apr 20 17:31:28 <akiniemi> denkenz: doesn't compile without it here...
Apr 20 17:32:34 <denkenz> lol
Apr 20 17:36:17 <denkenz> try removing the include and #including net/if.h
Apr 20 17:36:31 <denkenz> Probably there was a conflict between net/if.h and linux/if.h, but not on my system
Apr 20 17:36:56 <denkenz> s/include/extern define/
Apr 20 17:40:15 <akiniemi> denkenz: net/if.h and linux/if.h overlap
Apr 20 17:41:01 <akiniemi> I get redeclaration warnings if I include both
Apr 20 17:41:31 <akiniemi> s/redeclarations/redefinitions
Apr 20 17:41:45 <denkenz> strange, since obviously I don't
Apr 20 17:42:00 <denkenz> sounds like a glibc fuckup
Apr 20 17:43:16 <denkenz> e.g. cat /usr/include/linux/if.h ...
Apr 20 17:43:17 <denkenz> #ifndef _LINUX_IF_H
Apr 20 17:43:19 <denkenz> #define _LINUX_IF_H
Apr 20 17:43:20 <denkenz> #include <net/if.h>
Apr 20 20:24:17 <holtmann> Only net/if.h or linux/if.h are needed. Preferred is net/if.h
Apr 20 22:02:07 <denkenz> balrog-k1n: here?
Apr 20 22:03:22 <balrog-k1n> denkenz: yep
Apr 20 22:03:41 <denkenz> So I was getting that error even with GUI mode
Apr 20 22:04:22 <balrog-k1n> that's strange... can you try and see if you see it after you flip the "Sim inserted" checkbox two times?
Apr 20 22:04:42 <denkenz> sec
Apr 20 22:04:44 <balrog-k1n> wait it might be difficult to flip it quickly enough
Apr 20 22:05:08 <balrog-k1n> i.e. before the first poll
Apr 20 22:05:09 <denkenz> I have to re-bootstrap since the Nokia guys broke the build
Apr 20 22:08:53 <denkenz> So with your patch it now works
Apr 20 22:11:19 <balrog-k1n> i was trying to rebuild current master to see if i can reproduce.. and the compilation hard locked by netbook.. not good :)
Apr 20 22:11:24 <balrog-k1n> with -j 3
Apr 20 22:11:28 <denkenz> heh
Apr 20 22:11:34 <denkenz> Once we remove the sim and oFono detects it
Apr 20 22:11:43 <denkenz> Is there a way to detect insertion?
Apr 20 22:12:10 <balrog-k1n> it should work except the previous poll can't be cancelled in GAtChat, so it'll wait until it times out in the modem
Apr 20 22:12:52 <denkenz> hmm, so the behavior I'm seeing:
Apr 20 22:13:17 <denkenz> default startup of phonesim, sim inserted obviously
Apr 20 22:13:18 <denkenz> AT+CSIM=8,A0F200C0 : Generic SIM access (Assign)
Apr 20 22:13:20 <denkenz> +CSIM: 4,9000
Apr 20 22:13:21 <denkenz> OK
Apr 20 22:13:23 <denkenz> AT+CSIM=8,A0F200C0 : Generic SIM access (Assign)
Apr 20 22:13:24 <denkenz> +CSIM: 4,9000
Apr 20 22:13:26 <denkenz> OK
Apr 20 22:13:28 <denkenz> Then I untick the sim inserted checkbox:
Apr 20 22:13:34 <denkenz> AT+CSIM=8,A0F200C0 : Generic SIM access (Assign)
Apr 20 22:13:38 <denkenz> Queue is now locked...
Apr 20 22:13:47 <denkenz> 1-2 minutes + now
Apr 20 22:13:51 <balrog-k1n> yeah
Apr 20 22:13:57 <balrog-k1n> in other words, it won't detect the re-insertion with phonesim because phonesim never times out
Apr 20 22:14:16 <balrog-k1n> we would need a way to cancel the timed out AT+CSIM in GAtChat
Apr 20 22:14:35 <denkenz> you need to time it out in phonesim
Apr 20 22:14:55 <denkenz> It is impossible to cancel an AT command besides an ATD or a COPS=?
Apr 20 22:15:33 <balrog-k1n> on all the modems i've tried, just typing AT (e.g. starting a new command) would cancel the previous one
Apr 20 22:15:42 <balrog-k1n> i don't know if this is from specification
Apr 20 22:16:18 <denkenz> So the V.250 spec specifies that command cancellation basically happens whenever a character is given after '\r' and before the final response has been received
Apr 20 22:16:38 <denkenz> However, if you read the actual modem specs, none of them allow command cancellation besides ATD and a few others
Apr 20 22:16:54 <balrog-k1n> that's what i expected g_at_chat_cancel to do, but i guess phonesim should have a timeout as well
Apr 20 22:17:22 <denkenz> yeah, g_at_chat can in theory do that, but it won't help against phonesim
Apr 20 22:17:51 <denkenz> And even if the command is cancelled, the final response must still be waited for :)
Apr 20 22:18:20 <balrog-k1n> in this case it would work because the AT+CSIM poll is the only command that sends no response (when sim is extracted in phonesim)
Apr 20 22:18:29 <balrog-k1n> it would also work at least on the Freerunner
Apr 20 22:19:27 <denkenz> Ok, if the Freerunner supports arbitrary command cancellation, then see what you can do in gatchat
Apr 20 22:19:45 <balrog-k1n> when you start a new command the modem basically forgets what it was doing before and just sends a response to the new command
Apr 20 22:20:03 <balrog-k1n> unless the modem is dead as it happens with Freerunner every now and then :)
Apr 20 22:20:13 <denkenz> That's not what happens on e.g. ATD
Apr 20 22:20:45 <denkenz> I doubt anyone actually got cancellation working right
Apr 20 22:21:36 <balrog-k1n> it is probably a potential race condition, but in the case of AT+CSIM poll there should be no problem
Apr 20 22:22:27 <denkenz> yeah but gatchat doesn't know its a CSIM
Apr 20 22:24:21 <denkenz> Also, for your last concern, we can always add g_at_chat_send_full or something that takes a submitted callback
Apr 20 22:24:30 <denkenz> You can do tracking of the timer there
Apr 20 22:25:18 <balrog-k1n> ok, good idea
Apr 20 22:25:20 <denkenz> Just as a general comment: Sit on or disable features which are not bullet-proof or expected to be
Apr 20 22:25:50 <denkenz> Once QA starts playing with this, they will file all sorts of bugs
Apr 20 22:26:07 <denkenz> And since we don't even expect it to work, the whole process will be a time waster
Apr 20 22:26:19 <balrog-k1n> yeah but we also want to be able to expect how ofono works by pointing at parts of the spec
Apr 20 22:26:25 <balrog-k1n> fortunately it's not used in most drivers
Apr 20 22:26:37 <balrog-k1n> s/expect/explain/
Apr 20 22:26:38 <denkenz> I know, but phonesim is used by QA exclusively
Apr 20 22:27:32 <denkenz> I fully agree its useful, don't get me wrong :)
Apr 20 22:27:51 <denkenz> Just maybe make it into an option, e.g. in modem.conf that only we know about
Apr 20 22:28:18 <balrog-k1n> that would work
Apr 20 22:29:31 <balrog-k1n> on the other hand i expect it to kind of work and be useful for proactive command detection, at least with phonesim
Apr 20 22:30:46 <denkenz> phonesim has the vendor proactive command notifications
Apr 20 22:30:52 <denkenz> by default we should use those
Apr 20 22:42:52 <denkenz> balrog-k1n: Your fix to phonesim was pushed btw
Apr 20 22:43:33 <balrog-k1n> thanks
**** ENDING LOGGING AT Wed Apr 21 02:59:56 2010