**** BEGIN LOGGING AT Wed Mar 04 11:44:39 2009 Mar 04 12:10:38 raster: ok Mar 04 12:10:49 private calls Mar 04 12:10:52 internal to elementary Mar 04 12:10:58 using any calles not in Elementary.h Mar 04 12:11:00 with EAPI Mar 04 12:11:11 WILL create problems Mar 04 12:11:20 ok Mar 04 12:11:28 i can and will break that api at will Mar 04 12:11:32 without you even knowing it Mar 04 12:11:36 until things just fall apart Mar 04 12:11:37 :) Mar 04 12:12:15 but another question, is it possible to change the theme of only one widget? Mar 04 12:12:23 no Mar 04 12:12:33 if the widget supports it - u can change styles Mar 04 12:12:49 ok Mar 04 12:15:14 mrmoku: To be honest, looks like there's no definitive policy about the "stable" branch (no release policy, no guidelines for distro maintainers, nothing like that). And some really important bugfixes are not there for no apparent (to me) reason. So, i'd go like that: coordinated with OE kernel maintainers (Mickey) and tried to convince them to test andy-tracking a bit later (when he finally fixes HDQ overflow issues (i have no clear idea ... Mar 04 12:15:19 ... about that) and reverts 6410 cpufreq patches. If no regressions are found, i'd go with that one (future) andy-tracking rev. But that's just me, probably i'm completely wrong here... Mar 04 12:20:15 mrmoku: Another idea is to provide both stable (without any backporting/cherry-picking) and andy-tracking for the users to choose. That way kernel devs will get much more feedback and at the same time those who want something "stable" will have it at the price of some known bugs. Mar 04 12:25:04 PaulFert`: hmm... thanks for your thoughts Mar 04 12:25:27 PaulFert`: offering two kernels is something I do not want to touch... Mar 04 12:25:39 don't even know if that what be feasible with OE/opkg... Mar 04 12:26:13 PaulFert`: I was mainly asking for the kernel bug with current Qi (I know andy reverted it)... Mar 04 12:26:25 PaulFert`: wanted to supply a Qi-friendly kernel :-) Mar 04 12:26:46 (and not get regressions by dropping some needed patch) Mar 04 12:26:47 mrmoku: OE can prepare as many kernels as you need it. And you don't really need opkg to install any kernel. So that's probably a non-issue. Mar 04 12:27:13 mrmoku: current stable should work even with non-broken Qi. Mar 04 12:27:14 PaulFert`: well... not completely true, because of the modules, no? Mar 04 12:27:28 :-) Mar 04 12:27:38 * mrmoku likes the term 'even with non-broken Qi' Mar 04 12:27:59 yeah... I was thinking more about the other patches that are applied Mar 04 12:28:08 mrmoku: hm, to install kernel modules are not needed ;) But yes, provide an opk for modules for every kernel binary you give, that shouldn't be hard. Mar 04 12:29:34 PaulFert`: I can imagine to provide an shr-unstable with an andy-tracking kernel though Mar 04 12:29:56 mrmoku: What i can say is that i don't like the current stable. Looks like it has no clear goal and maintaince. Mar 04 12:30:02 (that would give it more testing too and not confuse everybody when bug reports come in...) Mar 04 12:30:43 PaulFert`: SHR is using andy-tracking with a fixed rev right now Mar 04 12:31:17 I might as well just add the patch for gsm initialization and be happy for testing... Mar 04 12:31:29 mrmoku: I hope SHR will move to OE really soon and will use OE kernel, looks to me like the most sane way. Mar 04 12:31:32 (shr-testing I meant) Mar 04 12:31:56 PaulFert`: well... what do you mean by OE kernel? we use the kernel from OE... Mar 04 12:32:10 We have the same kernel as FSO Mar 04 12:32:15 mrmoku: i mean the same kernel that is used by FSO, yes. :) Mar 04 12:32:38 hi, i am trying to compile some gtkmm application for Om2008 and there are not available a some libraries (gtkmm, glibmm, libsigc++) for opkg-target. Is gtkmm supported? How can i install the libraries (trying to compile by my self, but it is difficult)? Mar 04 12:32:51 PaulFert`: just that I asked mickeyl if ms5.1 branch will update to a kernel with the gsm fix Mar 04 12:33:03 and that wont happen in the ms5.1 branch (what I can understand) Mar 04 12:33:28 because shr-testing is based on that branch we either have to live with that kernel or differ Mar 04 12:33:56 now that Qi is fixed-broken... we might as well live with it? Mar 04 12:34:04 mrmoku: yes :( Mar 04 12:34:24 (would be to the contrary of what I said on the ML though... wanted to provide a gsm-fixed kernel) Mar 04 12:35:05 PaulFert`: and then there was one bug report about garbled screen on resume in combination with Qi... Mar 04 12:35:07 mrmoku: But can you? Do you have powers to add kernel patches to OE? And what's the current policy about OM kernels in OE? Mar 04 12:35:19 mrmoku: Saw that, yes Mar 04 12:35:40 PaulFert`: well I have the power to add a patch to OE in our overlay that adds patches to the kernel ;) Mar 04 12:36:26 anyway... that garbled screen report has nothing to do with the gsm fix... so I will keep that kernel for shr-testing Mar 04 12:36:47 Thanks again for your input :-) Mar 04 12:37:14 mrmoku: If you like it from a maintenance POV, well, backport two GSM-fixes and the latest LCM fix. Probably BT-broken-after-suspend worth picking too. Mar 04 12:38:21 PaulFert`: I don't want to introduce instability in shr-testing... though that would be only acceptable if the prooved to be gooood Mar 04 12:38:35 for the upcoming shr-unstable that would be interesting though Mar 04 12:38:55 or use andy-tracking directly if it is fixed to work somehow Mar 04 12:45:51 hey, my new moko has a resistor between some sd card pins Mar 04 12:46:03 isn't that the hardware fix for interference with GPS? Mar 04 12:46:05 yay! Mar 04 12:46:11 spaetz: japp Mar 04 12:46:18 cool Mar 04 12:47:26 Steve wrote that A7 manufacturing was put on hold because of alsa states issues. Isn't that ridiculous? Mar 04 12:57:07 mrmoku: just flashing shr-testing to NAD "rootfs" should work fine? Mar 04 12:57:19 Or do I have to flash the kernel separately to the kernel partition? Mar 04 13:03:28 * spaetz discovers uImage-om-gta02 an flashes that to "kernel" Mar 04 13:04:45 spaetz: if the kernel differs you have to flash it... does not change that often though Mar 04 13:06:30 k, thanks Mar 04 13:07:04 I don't really intent to change shr-testing really, anyway. It's my stable fallback and used for flashing the sd card with a new version Mar 04 13:38:51 PaulFert`> WTF sushama removed Joerg from CC in bug ticket? What for? :: Well *I* didn't ask for that. Maybe you go ask her... :-/ Mar 04 13:45:04 mmh, shr busybox doesn't do "tar xz" Mar 04 13:45:17 no ungzipping build into tar Mar 04 13:46:20 spaetz: yeah... gzip/bzip2 was removed from busybox with the last busybox update some time ago me thinks Mar 04 13:46:38 mmh, no catastrophy but a pity Mar 04 13:47:44 especially as busybox has gzip built in anyway Mar 04 13:48:33 /bin/gzip -> busybox Mar 04 13:53:24 stomp busybox deep in the bin Mar 04 13:53:37 weird: CONFIG_FEATURE_TAR_GZIP=y Mar 04 13:53:43 in http://cgit.openembedded.net/cgit.cgi?url=openembedded/tree/packages/busybox/busybox-1.13.2/defconfig&h=fso/milestone5 Mar 04 13:54:04 unless SHR overrides that somehow Mar 04 13:57:16 spaetz: Try swapping the parameters around. I think busybox's tar had some trouble with that when I used that... Mar 04 14:01:01 mmh, that might be. weird busybox Mar 04 14:01:17 mrmoku: didn't you patch illume power settings to be disabled? Mar 04 14:01:43 Just flashed new shr-unstable and illume says "30 secs" to blank and "1 sec" to suspend after blank Mar 04 14:06:31 qte is gpl now Mar 04 14:13:08 blarg, I seem to have used a very shr-unstable image. now that explains a lot Mar 04 14:13:14 very OLD Mar 04 14:24:11 spaetz: well... without knowing what I'm doing :P Mar 04 14:25:04 spaetz: http://git.shr-project.org/git/?p=shr-overlay.git;a=commit;h=d37ab40f1a22dcea18f6af829e80ec893474771a Mar 04 14:25:13 would be the commit which was supposed to do it... Mar 04 14:25:56 the values in e config are there... though it does not blank the screen for me Mar 04 14:26:08 what it does is the swallow first tap behaviour though... Mar 04 14:27:26 do you know if it's allready planned the image creation date of the new shr-unstable? with new englightment inside? Mar 04 14:27:46 s/englightment/enlightment/ Mar 04 14:27:46 a21_work meant: do you know if it's allready planned the image creation date of the new shr-unstable? with new enlightment inside? Mar 04 14:38:01 a21_work: no planned date... have to do some infrastructure work for that Mar 04 14:38:26 is shr taking some parts of qtextended? Mar 04 14:39:02 this text-Ywrite support for example Mar 04 14:44:08 Serdar: no... don't think so. qtextended is framebuffer based and SHR is X based... nothing to simply take Mar 04 14:48:02 hmm framebuffer, that's a nice idea. Mar 04 14:49:23 shr guys: is the framework in shr testing updated with the fixes for agps? Mar 04 14:49:47 http://git.freesmartphone.org/?p=framework.git;a=commitdiff;h=5271e445b327c2132eee6a1f43fcf58c37c67e00 , http://git.freesmartphone.org/?p=framework.git;a=commitdiff;h=0838645958d50df38e64429a27672570bf58f8b4 and http://git.freesmartphone.org/?p=framework.git;a=commitdiff;h=ad2ec48cdb5ab9bdddc15adfc05cbd2e5b8a2cee Mar 04 14:51:48 Zorkman: no Mar 04 14:52:11 there is logs of stuff in between what shr uses and those agps fixes... Mar 04 14:52:29 s/logs/lots/ Mar 04 14:52:29 mrmoku meant: there is lots of stuff in between what shr uses and those agps fixes... Mar 04 14:53:39 k, np Mar 04 14:53:58 is it something we can expect for near future or better not hope for it too soon Mar 04 14:53:59 ? Mar 04 14:54:48 well... depends on what fso guys advise... if the say the stuff in between is safe to take... then ok Mar 04 14:55:14 otherwise i don't know... would have to cherry-pick those fixes maybe Mar 04 14:55:28 (which I would prefer not to have to :P) Mar 04 14:56:07 check, i'll bribe mickey ;) Mar 04 15:20:00 DocScrutinizer: Do you by any chance know, is it true that A7 was put on hold because of the mythical 'alsa states issues'? Mar 04 15:24:14 when i try to change the location of the map dir of tangogps to my card it crashes. any else has this problem? Mar 04 15:26:35 mokel: yep have that one too :( Mar 04 15:26:39 damn Mar 04 15:26:49 i solved it by softlinking Mar 04 15:26:58 but thats not the best ^^ Mar 04 15:27:09 PaulFert`: it was delayed. I don't know recent status though Mar 04 15:27:19 mrmoku: did you try the current qi version? Mar 04 15:28:39 DocScrutinizer: So, what do you think about the echo issue in general? Where the feedback really comes from, can i help to find it, is it worth trying? Mar 04 15:29:07 PaulFert`: mostly due to fab test-sw had the wrong alsa-setup so devices didn't pass aaudio-test Mar 04 15:29:48 (echo) no use in trying, as we can't change anything wrt case and/or mechanics Mar 04 15:30:06 AT%Nxxxx should fix echo issue for good Mar 04 15:30:51 and I know for sure other devices like Nokia etc have same physical echo issue Mar 04 15:31:06 they cure it by AEC as well Mar 04 15:32:02 maybe our design is more sensible to feedback from earpiece to mic than some others Mar 04 15:32:15 anyway, as I said, we won't change anything Mar 04 15:33:51 DocScrutinizer: looks like it doesn't fix it in a good way everybody likes. Other phones don't sound so "harsh" while lowering echo considerably. And if it's really the casing then some easy case modding might help. Mar 04 15:34:24 there won't be any more casemodding for GTA02 Mar 04 15:34:41 DocScrutinizer: even DIY? ;) Mar 04 15:34:54 you can tweak AT%xxxx to meet your taste Mar 04 15:35:10 And get loud echo again :) Mar 04 15:35:59 duh, why that? Is it confirmed the setting mickey introduced is the *only one* to cancel echo? Mar 04 15:37:20 But i think that you really can't cure _any_ level of feedback with AEC? Sometimes it's just too much, and either it'll echo or it'll sound "harsh". Mar 04 15:38:16 nah, you usually won't talk while listening to farend. So it's pretty easy to stop feedback in even the worst situations Mar 04 15:39:10 IIRC the setting was "long AEC" though. Maybe "short AEC" might better please your ears Mar 04 15:39:42 I wonder if anyone using the headset ever experiened echo, i don't remember any clear reports about it. Mar 04 15:40:51 And as echo and "harsh" sound is on the remote end, no setting can please my ears, but i care about those who i talk with :) Mar 04 15:41:00 probably noone ever used the headset (except me ~1year ago, when I noticed evil buzz and first got aware we have a serious EMI issue) Mar 04 15:41:25 Who can saw few words about xf86-video-glamo status? is it a works? at least sometimes?) Mar 04 15:43:24 PaulFert`: you know EMI is caught up by pin4-mic-input. Buzzfix won't help here. HS-cord makes things much worse, so no good perspective for using headset anyway Mar 04 15:44:22 though I remember some reports stopping hs-buzz by applying a ferrite bead to hs-cord close to jack. Makes sense Mar 04 15:46:07 PaulFert`: could you have another short look into BIP? ;-D Mar 04 15:46:42 PaulFert`: nevermind Mar 04 15:46:58 it's really just been me killing it. Works again ;-) Mar 04 15:49:05 :) Mar 04 15:52:00 mokel: yeah...flashed current qi Mar 04 15:52:09 works great... suspend/resume also Mar 04 15:52:46 mokel: you get your screen garbled on every resume? you do something special? rotation? Mar 04 15:52:47 :D Mar 04 15:54:00 nothing special Mar 04 15:54:36 hmm... don't know then... you could try to boot via u-boot from nor and see if it happens there? Mar 04 15:55:17 with uboot everythis is fine Mar 04 15:55:27 everything Mar 04 16:00:50 python-elementary Toolbar! is it supposed to work by now? Mar 04 16:01:10 I always get "error, no callback for clicked available or something like this" Mar 04 16:01:45 "ERROR: no callback available for the item" Mar 04 16:02:01 in shr-testing? ... no Mar 04 16:02:08 shr-unstable Mar 04 16:02:14 seems to be the same Mar 04 16:02:15 neither ;) Mar 04 16:02:17 yep Mar 04 16:02:22 k :-( Mar 04 16:02:38 will work around for now Mar 04 16:02:45 problem is I have to change the Makefile to support a different OE tree for testing and unstable Mar 04 16:02:45 but lists might work, right? Mar 04 16:02:54 lists should work Mar 04 16:03:05 you can check with the elementary-test app what is there Mar 04 16:17:17 mmh, lists seem implemented in C but not python Mar 04 16:17:29 at least not in the version we have in unstable Mar 04 16:23:34 spaetz: thats possible :( Mar 04 16:24:16 spaetz: I will try to get around and do some Makefile work tonight Mar 04 16:24:31 and we have to think about branches... Mar 04 16:25:59 maybe branch what we have now as 'testing' branch and build shr-testing from that branch Mar 04 16:26:46 mmh, yep. but List is not working in current python-elementary, so it's of no use to update that... :-( Mar 04 16:27:02 from c_elementary import Object, Window, Background, Icon, \ Box, Button, Scroller, Label, Toggle, Frame, Table, \ Clock, Layout, Hover, Entry, AnchorView, AnchorBlock, Bubble, \ Photo, Hoversel, Toolbar, init, shutdown, run, exit Mar 04 16:27:10 (no List) Mar 04 16:31:40 Sarguns Mar 04 16:34:51 need to pester morphis Mar 04 16:36:27 spaetz: successful pestering :-) me is off for lunch now :P Mar 04 16:42:19 * Failed to download http://build.shr-project.org/shr-testing/ipk//armv4/Packages.gz, error 404 Mar 04 16:42:33 it should be armv4t Mar 04 16:42:38 I suppose Mar 04 16:42:41 ;) Mar 04 16:42:48 current shr-testing Mar 04 16:45:50 a21_work: just delete armv4.conf in /etc/opkg Mar 04 16:46:30 ...or maybe one day we will have that repo also? Mar 04 16:46:38 s/repo/feed/ Mar 04 16:46:39 a21_work meant: ...or maybe one day we will have that feed also? Mar 04 16:46:45 not that I'd know of, no Mar 04 16:47:26 would it be possible to compile a kernel without all the console messages? it should be faster on suspend/resume because the glamo does not have to display all the lines Mar 04 16:51:44 mokel: there were some mails about this in smartphones-userland Mar 04 16:52:26 what is smartphones-userland? Mar 04 16:52:33 mailing list Mar 04 16:52:38 ah, ok Mar 04 16:54:04 look at Mar 04 16:54:08 http://projects.linuxtogo.org/pipermail/smartphones-userland/2009-March/001223.html Mar 04 16:54:43 i didn't test it myself Mar 04 16:56:40 mmh, does shr provide the source of their kernel? Mar 04 16:56:45 can't find it Mar 04 17:09:36 can anyone advice me how to add libsigc++ into toolchain? It is kind of tricky. It is not in Om package repository and i am not able to compile it by hand Mar 04 17:11:36 mokel: no need to compile the kernel... just pass it loglevel=2 on the commandline Mar 04 17:11:56 and shr has no own kernel... it is using the openmoko kernel Mar 04 17:12:01 ok Mar 04 17:12:22 kernel source is here: git://git.openmoko.org/git/kernel.git Mar 04 17:14:55 zeemokus Mar 04 17:48:19 mrmoku: can you explain why the kernel inside the .tgz is different to the kernel uImage-latest? they are the same size but have different md5's.. Mar 04 17:48:46 could it just be the build date? Mar 04 18:02:04 toggles_w: hmm... don't know that Mar 04 18:02:24 maybe mickeyl can sched some light on it :-) Mar 04 18:03:16 morning Mar 04 18:03:27 morning Mar 04 18:03:56 probably build date, ya Mar 04 18:08:45 mickeyl: whaazzzup! Mar 04 18:09:02 mickeyl: thanks mate, so i can use either then! Mar 04 18:09:27 now if only i could build -testing.... ;-) Mar 04 18:34:45 Ainulindale: ping Mar 04 18:34:51 * mrmoku would have a question now :-) Mar 04 18:38:52 Ainulindale: I have to do different branches for oe in the Makefile now Mar 04 18:39:06 Ainulindale: and I don't know what is the better way to handle the patches Mar 04 18:39:29 Ainulindale: a) branches in shr-overlay Mar 04 18:40:03 Ainulindale: b) two (or three when milestone1 comes one day) patchsets in the patches dir Mar 04 18:42:23 Ainulindale: hmm... thinking about it I guess you will tell me to ask the ML... so will I do now :-) Mar 04 18:55:17 hmm Mar 04 18:55:40 * bumbl_ was tempted to buy a working phone (g1/iphone/...) today Mar 04 18:56:36 what did you get? Mar 04 18:56:41 * bumbl_ resisted Mar 04 18:56:45 oh Mar 04 18:56:47 good Mar 04 18:56:54 anyone knows where i can change the fontsize for the font under the application icons? couldn't find it in the edje file Mar 04 18:57:11 i was playing with iPhone and G1 yesterday Mar 04 18:57:17 * mrmoku is proud of bumbl_'s resistance :-) Mar 04 18:57:25 and I really don't like that touch screens... Mar 04 18:57:47 me neither, thats why i bought the openmoko Mar 04 18:57:52 but i have to admit that i use my 6230i when i have to do "important" stuff Mar 04 18:58:08 bumbl_: exactly the same as I - 6230i too ;) Mar 04 18:58:27 but it starts breaking apart Mar 04 18:58:35 * mrmoku gets red when clients complain about buzz and low volume Mar 04 18:58:41 bumbl_: as mine... Mar 04 18:58:47 but if those two issues would be fixed :D Mar 04 18:59:13 hmm i didn't have buzz and low volume with fso ms5 Mar 04 18:59:51 which i have been using since before my trip to switzerland Mar 04 19:00:32 bumbl_: maybe you were phoning only polite people that did not complain :P Mar 04 19:00:48 mrmoku: nah Mar 04 19:01:08 i actually ask everytime if there is a buzz/echo/low volume Mar 04 19:01:12 bumbl_: what other things are the showstoppers for you? Mar 04 19:02:05 updating the gsm firmware was a huge step ahead for me Mar 04 19:02:20 (and of course frameworkd progress :-) Mar 04 19:03:15 mrmoku: slow moko, strange nsod when using qi and the fact that i am unable to get my contacts and calendar data from my 6230i via gnokii + bluetooth (authentication fails) Mar 04 19:04:53 ah + shr being unstable recently Mar 04 19:07:38 i've been using my GTA01 as my primary phone since launch. i'm just *very* careful about revisions of software i put on it. currently running SHR unstable 02162009. Mar 04 19:07:41 -> most showstoppers are hardware or low-level software ones + one is more or less my fault Mar 04 19:07:54 juri_: wow Mar 04 19:11:21 i did cd /usr/share; ln -s elementary-alarm elementary_alarm Mar 04 19:11:24 on shr testing Mar 04 19:11:28 but alarm still doesn't ring Mar 04 19:11:33 known problem? Mar 04 19:12:31 i might get one of those touch books though they look interesting Mar 04 19:15:43 ~seen quickdev Mar 04 19:15:43 quickdev was last seen on IRC in channel #openmoko-cdevel, 7d 4h 40m 23s ago, saying: 'and for you?'. Mar 04 19:17:20 hehe... remember that :-) Mar 04 19:17:26 heh... Mar 04 19:17:44 gmail isn't working right with shr maillist... Mar 04 19:17:57 for instance now i replied only for mrmoku :P Mar 04 19:18:16 and mrmoku agrees with option a) Mar 04 19:18:54 i'm playing with new icons for shr-settings Mar 04 19:21:31 I'm still having trouble with http://trac.freesmartphone.org/ticket/345 Mar 04 19:21:47 retrieving the whole book works but trying to retrieve individual binary messages is still failing Mar 04 19:21:50 mickeyl: Hey :) Wanted to report that i was using ogsmd-new-timeouts for 2-3 days and i can't say i really like it ;) Picking up the call become so much slower that it's no good for everyday use. And ringtone (mp3, after resume) sounds with pauses. Mar 04 19:25:20 that's sad Mar 04 19:25:30 i didn't see all these problems Mar 04 19:25:51 this means we need to go faster with the vala reimplementation Mar 04 19:26:21 we should profile first though Mar 04 19:27:43 mickeyl: do you plan to ship a vala python mix or vala only when every subsystem is finnished? Mar 04 19:27:47 mickeyl: Using mp3 instead of sid probably contributes. But really, too slow, the other side complains they hear nothing after i pickup but i can't say anything for like 2 sec because alsa state is not loaded yet (and the phone keeps vibrating) even after ringtone stopped. Picking up was really slow, but now it's even slower. That's how i feel it. Mar 04 19:28:37 Probably that's even more than 2 sec, as i never heard it myself. Mar 04 19:29:45 bumbl_: we will mix. Once the first vala subsystem is ready, it's going to ship Mar 04 19:30:28 mickeyl: using 858c8d58d1f7e807f2c09532787c4e7b1a5daa52 without modifications. Can provide frameworkd log with ogsmd DEBUG if needed. Mar 04 19:30:32 PaulFert`: I have that two on shr... since we ship a ogg file as ringtone Mar 04 19:30:38 s/two/too/ Mar 04 19:30:38 mrmoku meant: PaulFert`: I have that too on shr... since we ship a ogg file as ringtone Mar 04 19:30:53 mrmoku: what decoder do you use for ogg? Tremor, i hope? Mar 04 19:31:00 ok and is the performance boost with vala that big? Mar 04 19:31:41 mrmoku: So, you basically confirm that my observations are correct and you can reproduce? Mar 04 19:32:05 PaulFert`: yes, please. I'd like to see the timestmpa Mar 04 19:32:20 PaulFert`: though... I'm not using ogsmd-new-timeouts Mar 04 19:32:53 bumbl_: depends on the area, i guess between 3x and 10x Mar 04 19:33:00 overall Mar 04 19:33:26 mrmoku: than probably ogg is not really a good choice unless you're using tremor or tremolo decoder. Mar 04 19:33:31 PaulFert`: how can I determine if shr is using tremor? I don't think so... Mar 04 19:33:45 its using gst-plugin-ogg... Mar 04 19:37:50 PaulFert`: could you tme alsactl with sth like strace to find the timehogs please? Mar 04 19:38:59 mickeyl: http://pastebin.com/m3d3bd674 Mar 04 19:39:29 thanks Mar 04 19:40:02 DocScrutinizer: i'd like to, but i think i never did anything like that before. Do you think it's something worth optimizing? It seems to me that the main reason is slowness of ogsmd, not alsactl :) Mar 04 19:40:42 time alsactl restore -f xyz gave me some 0.8sec :-/ Mar 04 19:41:14 argh Mar 04 19:41:16 well I surely can get same results as you might get. So no need for you to try and strace it Mar 04 19:41:16 that's one second! Mar 04 19:41:20 mickeyl: according to gsm interference i heard i resumed by pressing power just before someone called me. Feel free to ask additional questions about the situation :) Mar 04 19:41:26 yup Mar 04 19:41:28 1 sec Mar 04 19:42:24 mrmoku: we should be using tremor for ogg anyway Mar 04 19:42:31 ln the FR, that is Mar 04 19:42:41 on the FR Mar 04 19:42:57 root@om-gta02:~# time alsactl -f /usr/share/openmoko/scenarios/gsmhandset.state restore Mar 04 19:42:59 real 0m 0.78s Mar 04 19:43:00 user 0m 0.06s Mar 04 19:43:02 sys 0m 0.07s Mar 04 19:43:03 root@om-gta02:~# Mar 04 19:43:53 I asume it's mainly filesystem I/O Mar 04 19:44:04 DocScrutinizer: hm, repeating the same state gives 0.2 Mar 04 19:44:07 anyway will trace it Mar 04 19:44:10 0.3 Mar 04 19:44:11 spaetz: hmm... tremor is in OE... I could just build it... but will the gst plugin use it? Mar 04 19:44:22 ogg is just the container Mar 04 19:44:41 PaulFert`: sure, as file is in sectors buffer Mar 04 19:44:42 vorbis and tremor exists in gstreamer Mar 04 19:44:57 exe-text as well as state file Mar 04 19:45:12 DocScrutinizer: looks more like in-kernel delays to me... Mar 04 19:45:15 ugh, RetrieveMessage is actually failing on all messages for me Mar 04 19:45:40 I do wonder now if this is related to the firmware upgrade Mar 04 19:45:59 strace and you'll know ;-) Mar 04 19:46:20 the error is in mediator.py Mar 04 19:48:27 PaulFert`: btw (0.3): real 0m 0.16s Mar 04 19:48:38 for 2nd time Mar 04 19:48:40 according to some maemo howto it's sufficient to have liboil and tremor installed wheren building gstreamer to have it build tremor Mar 04 19:48:46 DocScrutinizer: well, it depends :) Mar 04 19:49:53 http://wiki.wjsullivan.net/collaboration.cgi/FsoEl/Bugs/RetrieveMessageChannelMediatorError Mar 04 19:49:58 is the backtrace Mar 04 19:50:08 but you can select to use vorbis or tremor at runtime, so the app using gstreamer needs to adapt to that Mar 04 19:51:02 0.16s 0.15s 0.16s 0.16s 0.16s 0.16s Mar 04 19:52:04 real 0m 0.30s for fetching new different statefile from uSD Mar 04 19:52:21 :o Mar 04 19:52:26 DocScrutinizer: Are you sure that the cap near internal mic won't help to eliminate ripple on MICBIAS for HS mic? Or is the problem that connecting headset introduces so much RF that no cap helps? Mar 04 19:52:41 DocScrutinizer: yup, i did it from uSD, that pretty much explains it :) Mar 04 19:52:58 nah cap CAN'T help for hs Mar 04 19:53:07 nothing can help for headset Mar 04 19:53:46 DocScrutinizer: could you please explain why? Mar 04 19:54:00 as hs mic is basically one single line hardwired to GND Mar 04 19:54:43 of course you could insert a LPF somewhere to hs-mic as well Mar 04 19:55:15 Dave: ~_~ Mar 04 19:55:39 but you had to place a R *before* C, for the LPF, and another one *behind* C for getting some signal at all (source impedance) Mar 04 19:56:05 * johnsu01 tries again with log_level=DEBUG Mar 04 19:56:18 DocScrutinizer: when will moko 11 go stable? and does om still work on the gsm firmware? Mar 04 19:56:25 and we didn't even achieve to insert a bead in this line :-/ Mar 04 19:57:41 bumbl: rumor is it's already considered "official". And yes it seems OM is accepting further improvements of GSM-FW as they come along from Dieter Mar 04 19:58:44 (one single line) I.E: it's not differential, it's single-ended Mar 04 19:58:57 http://wiki.wjsullivan.net/collaboration.cgi/FsoEl/Bugs/RetrieveMessageChannelMediatorError Mar 04 19:58:58 DocScrutinizer: _now_ i get it. Sorry for being stupid Mar 04 19:58:59 now with more debug log Mar 04 19:59:23 PaulFert`: nah! Mar 04 20:00:18 PaulFert`: http://rafb.net/p/Dc9cYL24.html Mar 04 20:00:30 this should improve it Mar 04 20:00:38 can you check when you have some time? Mar 04 20:00:39 argh Mar 04 20:00:41 PaulFert`: source impedance of 8753 MICBIAS output is very low, so it doesn't help to place even 800uF on there ;-) someone in TPE actually did the test :P Mar 04 20:00:42 NOTE: Not creating empty archive for shr-settings-locale-pl-0.1.0+r70+3c9c1d4b744499ad3b17ca91e022bb2c30b73c58-r2 Mar 04 20:00:43 why? :X Mar 04 20:04:10 PaulFert`: A8 actually has beads on all 3 lines from hs-recetacle to AMP. So we might get rid of buzz the correct orthodox way finally Mar 04 20:04:29 mickeyl: Ok, do you want me to do it now? ;) Mar 04 20:04:48 no, it's not urgent Mar 04 20:04:52 am testing it atm. on my own Mar 04 20:04:58 i need millisecond granularity in the logs Mar 04 20:05:02 mickeyl: it doesn't look that autocomputing prefixes every time makes sense anyway, or does it? Mar 04 20:05:13 yes, that's why there's a cache now :) Mar 04 20:05:54 PaulFert`: honestly the stress and hassle to get those 3 beads in there cost me 6months of my live. Everybody accused me I broke GSM-FW upload :P Mar 04 20:06:32 :O Mar 04 20:08:21 Until I made clear it's just a matter of disabling AMP using AMP-SHUT to ddrive the outputs high-Z (as I thought every reasonable coder of fab-test-sw would have done from beginning) Mar 04 20:08:49 :-( Mar 04 20:09:09 mickeyl: are you interested in investigating why sometimes oeventsd doesn't change from lock to busy and doesn't enable backlight despite the user is doing his best to press the touchscreen and AUX? ;) Mar 04 20:09:36 We like you, Joerg. Mar 04 20:09:37 DocScrutinizer: that's sad... Mar 04 20:09:40 in the end it's historic fact: I broke it and they needed to patch fab-test sw to cope with it Mar 04 20:09:50 loool Mar 04 20:09:51 and we certainly don't blame you :P Mar 04 20:10:09 well, I filed http://trac.freesmartphone.org/ticket/369 with the info Mar 04 20:10:12 Some of us are just good at breaking things... and well, it's part of the process :) Mar 04 20:10:52 PaulFert`: sure, have a log for that? Mar 04 20:11:15 btw: hi mickey :) Mar 04 20:11:19 Dave: i can't call it breaking. I'd say it's exposing bugs. And some people don't like it when their bugs are exposed. Too bad OM still has such employees :( Mar 04 20:11:20 hi DocScrutinizer Mar 04 20:12:02 PaulFert`: ack Mar 04 20:13:26 ok, with that cache and logging to INFO, it's bearable again Mar 04 20:13:33 loggin = DEBUG will always be superslow Mar 04 20:13:49 i'm afraid python's logging infrastructure is wasting tons of cycles Mar 04 20:14:04 string handling ;) Mar 04 20:14:06 That's totally unsurprising :D Mar 04 20:14:07 mickeyl: how am i supposed to run ogsmd then to help finding bugs? ;) Mar 04 20:14:09 i need to think about removing the calls by preprocessing or so Mar 04 20:14:23 PaulFert`: i know Mar 04 20:14:37 it's just that there's a tradeoff between debugging info and latency Mar 04 20:14:57 one which i can't remove in py Mar 04 20:15:52 freesmartphone.org: 03mickey 07framework * r437b3564fbf2 10/framework/subsystems/ogsmd/gsm/channel.py: ogsmd: channel.py: cache autogenerated prefixes since re.findall is an expensive call Mar 04 20:15:57 dump "trace" to an array internally, avoiding all the cost of "normal" logging? Mar 04 20:16:48 hmm? Mar 04 20:16:56 i'm afraid the time is spent formatting the messages Mar 04 20:17:09 then print out the whole array after profiling the interesting code segment Mar 04 20:17:11 ah, now i understand Mar 04 20:17:27 so you would move formatting time to afterwards Mar 04 20:17:32 only saving internal state somewhere Mar 04 20:17:36 nah, don't format. store integer keys and reformat at print time Mar 04 20:17:51 yup, exactly Mar 04 20:17:55 i see Mar 04 20:18:03