**** BEGIN LOGGING AT Tue Jan 25 02:59:57 2011 Jan 25 02:59:58 denkenz: instead of updating all the drivers, i think it's better to make it to work without cnap/cdip, isn't it? Jan 25 03:00:30 demarchi: No, we don't initialize the call structure in some drivers Jan 25 03:00:37 demarchi: So you get garbage Jan 25 03:00:46 demarchi: And D-Bus fails the utf8 check Jan 25 03:02:01 denkenz: what timezone are you on? Jan 25 03:02:11 CST Jan 25 03:03:41 ahn... ok Jan 25 03:05:20 well... here it's GMT-3, Jan 25 03:05:43 i have to go... tomorrow i'll take a look on this Jan 25 08:13:08 When building ofono, is it ok to edit configure file at "remove checking the libudev line" ? Jan 25 08:13:19 otherwise won't build Jan 25 08:25:37 zma: Just install libudev-dev package. Jan 25 08:26:10 You can do ./configure --disable-udev, but that will not get you far when trying to run oFono. It needs this library to auto-detect modems. Jan 25 08:26:35 The only modem that can be used without udev is the phonesim emulator. Jan 25 08:33:55 ok thanks Jan 25 11:04:16 voicecall: __ofono_voicecall_dial() uses : Jan 25 11:05:12 req->ph.type = addr_type; strncpy(req->ph.number, addr, 20); , and I wonder why it does not use string_to_phone_number(). Do you know? Jan 25 14:32:51 zrafa: Because __ofono_voicecall_dial is used by sim toolkit, which receives the number in 'raw' format Jan 25 14:35:29 denkenz: ah, I see. Jan 25 14:53:56 denkenz: did you get my patches for sms storage? Jan 25 15:05:28 demarchi: It seems so Jan 25 15:13:52 denkenz: ok, moving on to next task Jan 25 18:36:00 holtmann: denkenz: what do you think of moving all the enums for cnap,clip,cdip,etc from src/call-settings.c to include/call-settings.h? Jan 25 18:36:17 this way the enum values can be used by drivers too Jan 25 18:38:11 The drivers should really not need these Jan 25 18:39:59 denkenz: no? Jan 25 18:40:03 denkenz: drivers/atmodem/voicecall.c:124: Jan 25 18:40:16 if (clip != 2) { Jan 25 18:40:40 I guess the point is that this is so specific to 27.007 that it's useless outside the AT driver Jan 25 18:41:13 oh jeez, you too Jan 25 18:41:37 I've said this before, I don't believe the enums in drivers are needed Jan 25 18:42:14 If you don't know what clip 2 means or 128/129/145 is, you should not be looking at the driver in the first place Jan 25 18:42:29 well, it makes sense for 27.007 where the spec refers to the numbers Jan 25 18:42:50 denkenz: i'm talking about this: Jan 25 18:42:52 but ISI without enums would be nothing short of code obfuscation Jan 25 18:42:53 20:46 holtmann> demarchi: And we should start using the enum constants here and not the hardcoded values for clip. Jan 25 18:42:57 20:47 holtmann> demarchi: That is a nice cleanup task. Use the proper constants for CLIP validity. At least atmodem needs also fixing. Jan 25 18:43:19 for isi sure Jan 25 18:43:26 for 27.007 I personally could care less Jan 25 18:43:39 yeah Jan 25 18:43:40 I would use the constants since they are defined. Jan 25 18:44:24 holtmann: for the *_validity constants, it's ok, since they are defined in common.h Jan 25 18:44:33 Using constants is fine, I would not go as far as moving them to call-settings.h Jan 25 18:44:54 but i can't do anything with clip, if don't move the enums Jan 25 18:44:56 And using values is fine as well in my book, at least in the driver Jan 25 18:45:12 denkenz: ok, i'll clean that up for the _validity bits then Jan 25 18:47:53 denkenz: about the initialization, i think we should do something else than go through each driver and set the cnap_validity to 2 Jan 25 18:48:30 Feel free to suggest something Jan 25 18:48:51 You might need to ofono_call_init or ofono_call_alloc function. Jan 25 18:50:36 yeah, ofono_call_alloc should fix that. Jan 25 18:55:40 i was thinking about the ofono_call_alloc, but calypsomodem uses a ofono_call struct on stack Jan 25 18:55:58 so... i think the _init is the best one Jan 25 18:56:13 demarchi: Fix calypso to not use the stack. Jan 25 18:56:21 umm, why? Jan 25 18:56:33 using stack is just fine and the way its supposed to be Jan 25 18:56:37 holtmann: i don't know if it's made this way on purpose Jan 25 18:57:24 _init is fine Jan 25 18:57:30 denkenz: For consistency reason with our voice call drivers it might make sense. However I am fine either way. So just use _init. Jan 25 18:57:44 holtmann: It doesn't, trust me Jan 25 18:58:02 all you would do is malloc, write, then free Jan 25 18:58:20 Okay. Jan 25 20:18:57 demarchi: ping Jan 25 20:23:01 denkenz: pong Jan 25 20:23:23 demarchi: So, one thing I don't like about sms pending patch 1: Jan 25 20:23:38 We re-encode the SMS to a PDU even though we already have the pdus available Jan 25 20:23:54 see pending_pdu in tx_queue_entry Jan 25 20:24:34 ok... Jan 25 20:29:51 demarchi: Also in patch 3, don't send out the MessageAdded signal Jan 25 20:30:00 That might cause serious spam Jan 25 20:31:50 Also, you can drop this: Jan 25 20:31:54 + if (backup_entry->msg_list->next != NULL) { Jan 25 20:31:55 + if (sms->ref == 65536) Jan 25 20:31:55 + sms->ref = 1; Jan 25 20:31:55 + else Jan 25 20:31:55 + sms->ref = sms->ref + 1; Jan 25 20:31:55 + } Jan 25 20:32:09 The ref has already been incremented when the message was queued in the first place Jan 25 20:32:17 No real need to do so again Jan 25 20:33:17 Otherwise looks good Jan 25 20:38:20 ok... Jan 26 02:45:45 I wonder if test cases require phonesim being set up? **** ENDING LOGGING AT Wed Jan 26 02:59:57 2011