**** BEGIN LOGGING AT Fri Feb 18 02:59:57 2011
Feb 18 13:53:26 <Gzajac> denkenz: Hi Denis, the crash I had on emulator with gsmdial came from the fact I didn't test the User ID (e.g. if we are in root).
Feb 18 13:53:26 <Gzajac> As I have modified the ofono.conf to launch ofonod without root rights, I also did a ATD*99 without having the root rights. As we can't preform a PPP connect without being root we crashed. 
Feb 18 13:53:26 <Gzajac> We have to test the user ID before performing the PPP connect to avoid crash.
Feb 18 13:54:05 <Gzajac> Like it is done in test server.
Feb 18 15:30:40 <denkenz> Gzajac: That feels like a hack inside test-server, we should not crash if we lack privileges
Feb 18 15:31:41 <denkenz> are we not handling ppp_net_new errors properly somehow?
Feb 18 15:47:12 <Gzajac> denkenz: it seems so
Feb 18 15:47:33 <Gzajac> we fail on ioctl() call
Feb 18 15:50:21 <denkenz> that shouldn't crash us though
Feb 18 15:50:30 <denkenz> failing to create ppp_net is totally fine
Feb 18 15:51:46 <Gzajac> yes
Feb 18 15:53:26 <Gzajac> btw, when I launch in root ofonod and I try a gsmdial I also fail to create ppp_net()
Feb 18 15:53:31 <Gzajac> but there is no crash
Feb 18 15:56:46 <denkenz> so lets figure out why and fix it
Feb 18 15:57:14 <crevetor> Hi guys, I have a problem. When I try to activate a ConnectionContext it seems like the modem "reboots"
Feb 18 15:57:21 <crevetor> it's an N900 with ofono 0.42
Feb 18 15:57:37 <crevetor> It might be kernel related...
Feb 18 15:58:05 <Gzajac> denkenz: ok, I will compare the 2 behaviors with test server and emulator deeply.
Feb 18 15:59:01 <denkenz> great, thanks
Feb 18 15:59:43 <denkenz> crevetor: only the Nokia guys can help you here I suspect
Feb 18 16:00:29 <crevetor> denkenz: are there some here ?
Feb 18 16:01:46 <denkenz> in the channel yes, whether they are really here I have no idea ;)
Feb 18 16:07:22 <crevetor> :)
Feb 18 16:11:38 <crevetor> Here's a log of what happens : http://dpaste.com/430245/
Feb 18 16:21:17 <crevetor> denkenz: who's from nokia here ?
Feb 18 16:22:30 <denkenz> try bothering akiniemi, or simply post to the mailing list
Feb 18 16:24:35 <crevetor> akiniemi: Are you here ?
Feb 18 16:42:27 <denkenz> fdanis: For CIND stuff I'd rather do it the other way around, have netreg implement COPS queries and notify the emulator
Feb 18 16:49:46 <fdanis> denkenz: I do not understand why COPS should be better than ofono_netreg_get_status()
Feb 18 16:50:23 <denkenz> for HFP we only have a read, so that is fine
Feb 18 16:50:38 <denkenz> But long term we can also have COPS manual / auto register as well
Feb 18 16:50:59 <denkenz> And I do not want to introduce things like signal strength watch, that is pointless
Feb 18 16:51:38 <denkenz> Just have netreg atom do something like __ofono_atom_foreach
Feb 18 16:52:02 <denkenz> I also want CIND indicators to be expandable and not static
Feb 18 16:52:58 <fdanis> denkenz: you want to introduce an API in emulator to be notified of signal strength modification ?
Feb 18 16:53:31 <denkenz> I want something like:  ofono_emulator_set_indicator(em, "signal", 4)
Feb 18 16:54:29 <denkenz> and ofono_emulator_add_indicator(em, "signal", min, max)
Feb 18 16:54:43 <denkenz> or whatever makes sense there
Feb 18 16:54:54 <denkenz> And have the emulator atom bootstrap the default set
Feb 18 16:58:37 <fdanis> I do not understand how the emulator will bootstrap the indicators
Feb 18 16:59:21 <denkenz> just have it call ofono_emulator_add_indicator(em, "signal", min, max); ofono_emulator_add_indicator(em, "roam", 0, 1), etc
Feb 18 16:59:43 <denkenz> For HFP we know the required ones
Feb 18 16:59:49 <denkenz> In the future we might have other indicators
Feb 18 17:00:34 <fdanis> which atom will perform this call ? netreg atom or emulator ?
Feb 18 17:00:48 <denkenz> most likely emulator_register
Feb 18 17:00:58 <denkenz> maybe we also need a 'default' value in the add_indicator call
Feb 18 17:01:17 <denkenz> so a CIND? will return something sensible always
Feb 18 17:02:12 <denkenz> This makes the signal, roam, service indicator implementation really simple
Feb 18 17:02:20 <fdanis> ok, so what should I do in netreg atom ?
Feb 18 17:02:49 <denkenz> when netreg receives updates to these, it will call __ofono_atom_foreach(modem, EMULATOR_HFP, some_func)
Feb 18 17:02:57 <denkenz> with some_func calling set_indicator()
Feb 18 17:03:12 <fdanis> ok
Feb 18 17:03:17 <denkenz> When netreg gets destroyed, it calls __ofono_atom_foreach again, with defaults
Feb 18 17:03:30 <denkenz> no watches, etc required
Feb 18 17:03:34 <fdanis> ok
Feb 18 17:04:35 <fdanis> when HFP HF connects to ofono HFP AG and modem is already online, how the indicators will be set correctly ?
Feb 18 17:05:32 <denkenz> good point, so we'll need to run an EMULATOR_HFP atom watch in netreg
Feb 18 17:05:43 <denkenz> and bootstrap the indicator there
Feb 18 17:05:52 <fdanis> ok
Feb 18 17:07:05 <fdanis> I suppose that I will need to use HAL for battchg indicator, right ?
Feb 18 17:07:19 <fdanis> like in BlueZ
Feb 18 17:07:26 <denkenz> Let plugins figure this out
Feb 18 17:07:44 <denkenz> basically I want the plugins to provide indicator values or even custom indicators
Feb 18 17:07:51 <denkenz> so HAL could indeed be one source
Feb 18 17:07:56 <denkenz> Or it could be something else entirely
Feb 18 17:08:55 <denkenz> Setting battchg to 100% for now seems fine to me
Feb 18 17:09:30 <fdanis> so, for now, I should create a new plugin which will manage the battchg indicator : setting battery indicator to 100%
Feb 18 17:09:52 <denkenz> Or just simply use the maximum value as the default in add_indicator
Feb 18 17:10:03 <denkenz> and leave it at that
Feb 18 17:19:13 <fdanis> watch on emulator should be set in  ofono_netreg_register or ofono_netreg_create ?
Feb 18 17:24:51 <denkenz> probably register is better
Feb 18 17:25:06 <denkenz> Basic rule of thumb is:
Feb 18 17:25:32 <denkenz> _create is called for the driver, driver is assumed to be initializing at that point
Feb 18 17:26:03 <denkenz> so only the basic structures that are absolutely required
Feb 18 17:26:19 <denkenz> _register is called when the driver has initialized, so the rest can be initialized here
Feb 18 17:27:22 <fdanis> ok, thanks
**** ENDING LOGGING AT Sat Feb 19 02:59:57 2011