**** BEGIN LOGGING AT Fri Oct 30 02:59:58 2009 Oct 30 16:20:08 denkenz: The EMRDY is only issued once on the f35xx hardware. And querying doesn't work. So please don't rely on it too much. Oct 30 16:20:30 holtmann: Querying works on the f36xx Oct 30 16:20:57 Not on the f35xx :( Oct 30 16:21:04 Strange Oct 30 16:21:11 Does it give you an error? Oct 30 16:21:29 Yes. Oct 30 16:21:52 I think we really have to look into possible firmware updates functionality. Oct 30 16:21:58 I noticed that EMRDY query on the f36xx gives an error Oct 30 16:22:05 But then EMRDY is sent Oct 30 16:22:18 Or EMRDY query suceeds Oct 30 16:22:28 So the current logic seems to work well Oct 30 16:23:33 With f35xx I never get the unsolisted message. Oct 30 16:26:10 strange, is there a way to do bus id detection on whether it is f35xx or f36xx? Oct 30 16:26:47 ATI will tell you. Oct 30 16:27:25 Yes, but can you issue any AT commands before EMRDY? Oct 30 16:28:26 Hard to test. Since it comes when I open a port for the very first time. Every other attempt doesn't give it to me at all. Oct 30 16:28:58 This is the issue, I believe no AT commands actually work until EMRDY is sent Oct 30 16:29:12 And AT*EMRDY without the ? kills that port. Oct 30 16:30:04 Or just some weird unicode char I had in there. Funny. Oct 30 16:30:54 Then just add a timeout of 1 seconds and what for the message. If it arrives fine, if not assume that it is ready. Oct 30 16:31:01 Do you get it every time you open a port? Oct 30 16:31:08 s/what/wait/ Oct 30 16:31:34 Or send ATI, if you get an error then wait for *EMRDY. Oct 30 16:32:02 Yes, that can be done Oct 30 16:32:14 I don't know. My f35xx doesn't support querying it. And it is only sent when we open the port for the first time. So you pick. Oct 30 16:32:36 Yeah we can figure out some heuristics Oct 30 16:33:02 Sending ATI or any other version command and a timeout should be enough Oct 30 16:33:22 ATI5 is nice btw. Oct 30 16:34:25 Yes does look nice Oct 30 16:34:52 Look at the changing channel numbers depending on which ttyACMx you use. Oct 30 16:35:43 btw, cdc-wdm0/1 can also be used Oct 30 16:36:16 With a proper kernel that does non-blocking on these? Oct 30 16:36:48 minicom opens them up fine Oct 30 16:36:54 I presume so will gatchat Oct 30 16:37:15 Remember our discussion about your read() without a poll() in between? Oct 30 16:37:26 vaguely Oct 30 16:39:23 We fixed it upstream kernel, but older kernels are an issue. Oct 30 16:40:07 It doesn't honor the nonblock flag bit Oct 30 16:40:09 I remember now Oct 30 16:41:52 can that be backported into previous kernels? Oct 30 16:42:07 There are maintenance releases Oct 30 18:58:55 denkenz: Can you pull phonesim.git and try to compile it. Oct 30 18:59:35 Maybe a fresh clone is a good idea ;) Oct 30 19:11:50 And after testing, please fix this one: Oct 30 19:11:52 UIC src/ui_controlbase.h Oct 30 19:11:52 Warning: name gbMessage is already used Oct 30 19:16:33 Fixed it by myself. Just renamed them to gbMessage1 and gbMessage2. Oct 30 19:26:03 ok, lemme try Oct 30 19:28:43 /bin/sh: /usr/lib64/qt4/bin/moc: No such file or directory Oct 30 19:29:31 Where is that one on your Ubuntu system. Oct 30 19:29:43 Gentoo installs it into /usr/bin Oct 30 19:31:30 You might wanna simply assume moc/uic are in $path Oct 30 19:31:40 Doesn't work on Fedora. Oct 30 19:32:30 --with-moc=/path/to/moc? Oct 30 19:35:01 FYI from QtCore.pc Oct 30 19:35:09 moc_location=/usr/bin/moc Oct 30 19:35:10 uic_location=/usr/bin/uic Oct 30 19:35:33 It has pkgconfig. Nice. Oct 30 19:36:47 yah, you can even depend on QtCore / QtGui / QtNetwork Oct 30 19:37:24 maybe QtXml too Oct 30 19:54:10 denkenz: Try again. Oct 30 19:54:27 Looks way cleaner now :) Oct 30 19:56:18 Now compiles cleanly Oct 30 19:56:30 And it still works? Oct 30 19:57:02 hang on a sec, in middle of changing some code Oct 30 19:57:05 Will let you know ;) Oct 30 19:58:04 I should already be sleeping. So just fix it if I broke it. Oct 30 19:58:30 AM_CXXFLAGS = -pipe -g -fPIC -Wall -W -D_REENTRANT $(QT_CFLAGS) Oct 30 19:58:42 I am still curious which of these flags are really needed. Oct 30 19:58:54 Yep works Oct 30 19:59:09 Even if you just do AM_CXXFLAGS = $(QT_CFLAGS) Oct 30 19:59:37 pipe Wall / W are harmless Oct 30 19:59:57 I am more curious about -D_REENTRANT Oct 30 20:00:31 It isn't used by phonesim, I'd assume you can safely remove it Oct 30 20:02:41 Removed. Oct 30 20:03:40 For what's it worth, ofono still starts up fine with AM_CXXFLAGS = $(QT_CFLAGS) Oct 30 20:03:46 Now I would even consider making releases for phonesim. Oct 30 20:04:34 Yah, looks decently clean now Oct 30 20:04:56 It wasn't even that hard. No idea what the FSO guys were smoking when they build that tree. Oct 30 20:05:14 It used to be worse Oct 30 20:05:26 pkgconfig in Qt didn't exist until some point Oct 30 20:05:48 But since it does now, time to use it. Oct 30 20:06:03 Sure, you might wanna set release to 0.1 and cut a package Oct 30 20:06:03 But even without pkgconfig, the whole build stuff was just messed up. Oct 30 20:06:35 Yeah I know, I was too lazy to fix it Oct 30 20:06:43 But me and build systems don't mix Oct 30 22:56:01 denkenz: netreg driver on the mailing list. comments are welcome :) Oct 31 00:14:31 denkenz: is there? Oct 31 00:59:10 zhenhua: If we don't have CHLD=1X we can't allow true mpty calls, we can still do something about multicalls Oct 31 00:59:24 zhenhua: And the swap before dial logic is already implemented in the core, see manager_dial Oct 31 01:02:48 padovan: The patch looks good, I'll probably take it in on Monday Oct 31 01:03:08 padovan: I do want to have one small change made though if you can Oct 31 01:03:39 padovan: When we receive service=1, can you wait for 1 second or so to see if roam=1/0 arrives? Oct 31 01:05:50 padovan: Mostly I don't want spurious stuff to happen when service=1, roaming=1 arrives Oct 31 01:07:28 so something like service=1, roam_watch = add_timeout(1, send_registration). Then if roaming=0/1 arrives and roam_watch != 0, g_source_remove(roam_watch), send registration info **** ENDING LOGGING AT Sat Oct 31 02:59:57 2009