**** BEGIN LOGGING AT Wed Jul 04 02:59:58 2012 Jul 04 05:12:46 where in audio files located in n900 folders Jul 04 05:12:52 where is audio files located in n900 folders Jul 04 05:16:26 /home/user/MyDocs/.sounds ? Jul 04 05:16:41 or .sound Jul 04 05:27:04 ShadowJK, thanks Jul 04 06:16:46 what's up with the hidden directories for that, anyway? Jul 04 06:16:50 what's wrong with "Sounds"? Jul 04 06:23:30 look at file manager Jul 04 06:23:34 it is sounds Jul 04 06:23:38 on english locale Jul 04 06:23:42 Sound* Jul 04 06:23:48 but if you go to any other locale Jul 04 06:24:02 maemo file manager will make the name look like it's in your local lingo Jul 04 06:32:53 jacekowski: os x does that Jul 04 06:33:04 but the directories are still called in a sensible way Jul 04 06:47:04 osx sucks Jul 04 06:47:15 lol why u mad tho Jul 04 06:47:28 and really, it's doing the same thing maemo filebrowser does Jul 04 06:47:36 except without hiding the actual directories Jul 04 06:49:48 I agree with kerio FWIW Jul 04 06:50:02 there could at least be symlinks Jul 04 06:51:06 luke-jr: that's actually even worse Jul 04 06:51:22 hm, how does windows access those directories? Jul 04 06:51:31 there are some "aliases" for os x Jul 04 06:52:32 screw Windows Jul 04 06:52:37 that's half the problem with Maemo Jul 04 06:52:44 trying to be Windows-friendly Jul 04 06:52:45 i know :s Jul 04 06:52:49 fucking fat32 mydocs Jul 04 06:53:00 but to answer, Windows actually names your directories in your locale Jul 04 06:53:09 and has registry keys that tell programs what they're called Jul 04 06:53:35 (that's for the OS user dirs, not accessing Maemo's… not sure on the latter) Jul 04 06:53:55 no, i don't really care about its directories :) Jul 04 06:54:18 os x has empty ".localized" files in the directories that have to be renamed Jul 04 06:54:46 and then it has its own lookup table Jul 04 06:54:53 by name Jul 04 08:34:33 zeq: there is a massive back out in fennec from yesterday, seems not only we were hit by the scrolling/performance problem. could you pull the latest code from mercurial and push it on github? Jul 04 09:08:44 freemangordon: there's a mirror of mozilla-central on github btw. I've rebased it on the current upstream. Jul 04 09:15:39 zeq: great. btw you may want to join #mobile, romaxa is there and there is an interesting discussion going on ;) Jul 04 09:50:51 I went to #mobile to read interestings discussions. Jul 04 09:50:57 I was dissapointed. Jul 04 09:51:24 freemangordon: not everything for CSSU-thumb installs cleanly in the SDK. Jul 04 09:51:42 zeq: for example? Jul 04 09:51:55 I'll go through it all again and take proper notes :) Jul 04 09:52:12 once you have the correct repos in place it should be ok Jul 04 09:52:18 it was something to do with the flasher Jul 04 09:52:28 vi__: #mobile in mozilla? Jul 04 09:52:29 not even relevent to the SDK Jul 04 09:52:57 vi__: irc.mozilla.org Jul 04 09:53:12 vi__: you're not interested to get fennec running on n900? Jul 04 09:53:15 :P Jul 04 09:53:27 @60fps ;) Jul 04 10:03:19 freemangordon: Yes I am very interested. I went to #mobile on freenode. Jul 04 10:03:23 I was the only one there. Jul 04 10:03:30 Forever alone. Jul 04 10:03:57 vi__, you meant ##mobile :) Jul 04 10:03:58 vi__: irc.mozilla.org Jul 04 10:04:08 oh wait Jul 04 10:04:41 freemangordon: email yet? Jul 04 10:06:38 vi__: no Jul 04 10:07:16 freemangordon: soo many mozilla bugs... looking to see if there's a suitable on for our fremantle patch Jul 04 10:07:36 zeq: ok. Jul 04 10:07:45 zeq: building the latest Jul 04 10:08:10 and going off for lunch :) Jul 04 10:08:40 I'm kinda hungry too now you mention it... Jul 04 10:09:08 freemangordon: how do I enable backing swap? I tried insmod /path/to/ramzswap.ko backing_swap=/path/to/backingswap disksize=soem_disksize Jul 04 10:09:39 but I have NFC ho to see if it has worked without stats from rzcontrol. Jul 04 10:11:55 also, how does it compress without the LZO module? Jul 04 10:17:20 ok. There seems to be some useful discussion on the google code page: Jul 04 10:17:23 https://code.google.com/p/compcache/wiki/CompilingAndUsingNew Jul 04 10:17:35 although the lzo question still stands. Jul 04 10:47:29 insmod /lib/modules/current/ramzswap.ko backing_swap=/dev/mmcbl Jul 04 10:47:31 k0p3 memlimit_kb=63556 Jul 04 11:23:04 vi__:lzo iz compiled in kernel Jul 04 11:23:08 *is Jul 04 11:30:03 zeq: latest m-c fennec works ok Jul 04 11:30:45 freemangordon: hw accel? Jul 04 11:34:01 without. and it is FASTER than before, lemme check if HW accel works Jul 04 11:36:59 faster is good :) Jul 04 11:37:05 how's memory usage? Jul 04 11:38:02 hmm, HW accel works too, despite those "couldn't create EGL image errors", and it is FAST Jul 04 11:38:19 cannot assess memory usage :( Jul 04 11:38:49 nice! :D Jul 04 11:39:12 well, fwnnwc uses 47.6 MB, TMO loaded Jul 04 11:39:16 *fennec Jul 04 11:39:32 plugin-container uses 31.3MB Jul 04 11:39:38 that is according to conky Jul 04 11:39:59 why so much for plugin-container? that's flash? Jul 04 11:40:09 NFC Jul 04 11:40:38 there is no flash content on TMO, so I don't know. lemme check yotube Jul 04 11:42:13 crashed :D Jul 04 11:42:32 oops! Jul 04 11:42:36 trying without hw accel Jul 04 11:43:20 freemangordon: ty Jul 04 11:44:12 vi__: np Jul 04 11:45:03 zeq: SF rendering seems still to be faster than HW, will have to wai romaxa to fix that Jul 04 11:45:48 s/SF/SW/ Jul 04 11:45:48 freemangordon meant: zeq: SW rendering seems still to be faster than HW, will have to wai romaxa to fix that Jul 04 11:53:24 https://bugzilla.mozilla.org/show_bug.cgi?id=770845 Jul 04 11:53:26 04Bug 770845: was not found. Jul 04 11:53:41 freemangordon: ^ Jul 04 11:53:54 Platform: x86_64 Linux Jul 04 11:53:56 :D:D:D Jul 04 11:54:07 oops! :/ Jul 04 11:54:41 better :) Jul 04 11:55:24 yeah, much better :) Jul 04 12:01:11 linaro 4.7-2012.07 is expected on the 12th Jul 04 12:03:09 freemangordon: do you think it's worth trying to build a "proper" scratchbox toolchain from it, or just update the kernel headers and C lib from CSSU-thumb and build another "alien toolchain"? Jul 04 12:03:35 (Definitely need to update the kernel headers!) Jul 04 12:04:48 zeq: I am not the correct person to answer that :(. It seems you're much more experienced than me on that matter - take my "toolchain" as a proof :D. Jul 04 12:05:09 But anything that works should be ok Jul 04 12:05:55 backingswap...ouch. Jul 04 12:06:25 vi__: toldya ;) Jul 04 12:06:32 but why?? Jul 04 12:06:46 wtf did nokia do to the VM? Jul 04 12:06:55 NFC Jul 04 12:21:50 zeq: well, flash is totally broken in latest :(. Jul 04 12:26:31 freemangordon: did you look at what romaxa pointed us at? Jul 04 12:26:58 zeq: which one? Jul 04 12:27:04 for the flash plugin Jul 04 12:27:17 that should be enabled Jul 04 12:27:25 it probably uses a completely different codepath Jul 04 12:27:31 maybe Jul 04 12:27:41 and while it was slow before, now it is totally broken Jul 04 12:28:34 freemangordon: looking for what romaxa said... Jul 04 12:28:47 ok Jul 04 12:29:27 https://bugzilla.mozilla.org/show_bug.cgi?id=598868 Jul 04 12:29:29 04Bug 598868: was not found. Jul 04 12:30:28 freemangordon: see mMaemoImageRendering in dom/plugins/ipc/ Jul 04 12:30:40 (romaxa said ^) Jul 04 12:32:19 yeah, but what about it? Jul 04 12:32:24 huh, how does maemo figure out what's the sim's number? Jul 04 12:32:31 freemangordon: looks like a MAEMO=6 needs changing Jul 04 12:32:43 freemangordon: it only enables qt stuff for HARM Jul 04 12:32:53 zeq: naah Jul 04 12:33:04 it is ok Jul 04 12:33:05 dom/plugins/ipc/PluginInstanceChild.cpp Jul 04 12:33:20 #if (MOZ_PLATFORM_MAEMO == 5) || (MOZ_PLATFORM_MAEMO == 6) Jul 04 12:33:24 it isok Jul 04 12:33:32 if defined(MOZ_WIDGET_QT) && (MOZ_PLATFORM_MAEMO == 6) Jul 04 12:33:47 at the top Jul 04 12:33:56 re Qt includes Jul 04 12:34:26 vi__, backing swap of ramzswap seems to use very random access patterns for write requests, whereas the nokia modified default swapout tries to keep writes as sequential as possible. This is a bit different from mainline linux, which tries to put writes near last read OR write access. Jul 04 12:34:36 random write requests == slow on flash Jul 04 12:35:30 freemangordon: I guess you're right, it mostly seems to refer to keyboard modifiers Jul 04 12:35:34 zeq: yes, but that is some key handling, on all other places it is #if (MOZ_PLATFORM_MAEMO == 5) || (MOZ_PLATFORM_MAEMO == 6) Jul 04 12:35:37 yep Jul 04 12:36:15 BTW sunspider scores even better, now it is 5095 Jul 04 12:37:03 N900 hw is pretty nice when running optimal code :) Jul 04 12:37:27 cant wait for nonsucking fennec Jul 04 12:38:23 yep. dang, my desktop scores 6 times faster. 4 phenom2 955 3.2G cores. it is pretty impressive result Jul 04 12:38:49 ShadowJK: ty Jul 04 12:38:52 yeah.. embarassing really Jul 04 12:39:29 ARM Cortex-A8 that much better clock for clock? Jul 04 12:39:31 mainline linux is optimized for hd, nokia mods optimized for flash, and ramzswap backing swap seems unoptimized ;) Jul 04 12:39:43 ShadowJK: bah! Jul 04 12:40:05 ~awesome Jul 04 12:40:05 hmm... awesome is JoshBang Jul 04 12:40:14 zeq: who knows. Jul 04 12:40:39 but values are those Jul 04 12:40:59 you may try to run it too Jul 04 12:41:11 I will :) Jul 04 12:41:46 going to re-merge my gstreamer config patch too, see if that works Jul 04 12:42:06 what i cannot understand is why it loads pages so slow. Jul 04 12:42:10 I had to move the Maemo config block above the codecs in configure.in Jul 04 12:42:11 zeq: pull it first Jul 04 12:42:27 i've commited the .patch Jul 04 12:42:55 freemangordon: I think something is timing out on page GET Jul 04 12:43:06 DocScrutinizer, btw, signatures aren't displayed in default maemo.org theme Jul 04 12:43:16 need to check what goes over the wire Jul 04 12:43:38 it's fast once the page starts loading Jul 04 12:44:24 freemangordon: I need to walk my dog now... I'll be back later :) Jul 04 12:57:57 in the meantime: ACTA is doomed in EU Jul 04 12:58:17 yippee! Jul 04 13:03:50 dammit, jake lloyd Jul 04 13:50:23 wtf is faster N9? Jul 04 13:50:29 Is it like the new speedpatch? Jul 04 13:56:01 vi__: tweaks and transitions for n9 Jul 04 14:01:16 dont talk about "speedpatch" :( Jul 04 14:13:22 yo vi__ Jul 04 14:13:27 gimme a prebuilt reaver-wps Jul 04 14:14:14 freemangordon: fennec build underway... Jul 04 14:21:18 DocScrutinizer et al.: what's the "correct" way of compiling stuff on the n900? Jul 04 14:22:38 kerio: you what? Jul 04 14:23:08 vi__: nevermind, found one Jul 04 14:23:17 a thing to bruteforce wps pins Jul 04 14:36:47 I know what it is. Jul 04 14:36:58 1. Hello Jul 04 14:37:04 2. Banal platitude. Jul 04 14:37:13 3. Jul 04 14:51:08 zeq: maps.google.com does not turn gps on. are you aware of something specioal we should do to enable geolocation? Jul 04 15:07:20 freemangordon: I would have thought it should work. Jul 04 15:07:27 Anybody here with an N9? Jul 04 15:07:46 zeq #harmattan Jul 04 15:07:52 zeq: #harmattan Jul 04 15:08:01 vi__: thx Jul 04 15:10:26 lots of the same faces over there :) Jul 04 15:16:59 seems my previous messages never got posted Jul 04 15:17:04 freemangordon: do you have your builds hosted somewhere? I'd love to do some testing. Jul 04 15:17:12 btw, using cssu-thumb for about a week and I haven't got any single problem or crash Jul 04 15:18:32 zeq: have a N9 what can I do? Jul 04 15:20:57 qwazix_900: if you could install fennec, and report back whether gps works on maps.google.com Jul 04 15:21:11 just a minute Jul 04 15:24:27 zeq: asks for location only after tapping on the dot above the little guy in the map Jul 04 15:24:42 freemangordon: did you do that? ^ Jul 04 15:24:53 above the zoom bar (desktop website) Jul 04 15:25:15 then position works as expected Jul 04 15:25:20 qwazix_900: using gps? Jul 04 15:25:53 trying. I'm inside Jul 04 15:26:04 let me get on the balcony Jul 04 15:26:45 ShadowJK: but then how did backing swap work on the N8X0? Jul 04 15:29:24 I enabled mapsgl to see what'll happen and now I can't press the dot. Other controls work Jul 04 15:30:00 got it, its pretty small and without a stylus... Jul 04 15:32:28 qwazix_900: it's using gps though, right? Jul 04 15:32:43 I'll test it if my fennec build ever completes :) Jul 04 15:33:03 Can't say. The icon in status bar has not appeared Jul 04 15:33:28 Im trying to guess from accuracy but handling is a pain Jul 04 15:34:13 Kozzi on #harmattan tested and seems to think it's working..? Jul 04 15:34:54 apparently had a couple of confirmations for sharing location Jul 04 15:36:11 from accuracy it does NOT seem to be using gps Jul 04 15:36:33 gives me about 50 meters radius circle Jul 04 15:37:03 qwazix_900: any gps indicator? Jul 04 15:37:23 no Jul 04 15:37:33 freemangordon: did it attempt to get location at all? Jul 04 15:38:21 I opened nokia maps to force the gps on, but accuracy on google maps still the same. It does ask for location, I answered yes and gives me an approximation Jul 04 15:39:30 thanks qwazix_900 Jul 04 15:42:28 np Jul 04 15:45:12 btw MapsGL works but it's slow and now I can boil an egg on the N9 Jul 04 15:46:45 Probably no webgl hw acceleration Jul 04 16:20:27 <_freemangordon> zeq: just clicked on that dot, lets see Jul 04 16:20:47 <_freemangordon> yep, GPS is on :) Jul 04 16:21:33 :D Jul 04 16:21:52 <_freemangordon> what's MapsGL? Jul 04 16:21:57 libqtm works then :) Jul 04 16:22:41 it's supposed to be 3d hw accelerated google maps Jul 04 16:22:45 <_freemangordon> yep, it geave me my correct location, even I am indoors, let me go outside to check the accuracy Jul 04 16:23:00 like a cut down web version of googleearth Jul 04 16:23:36 _freemangordon: zeq you are too stuborn ants, working all the time i see you here Jul 04 16:23:39 =) Jul 04 16:23:45 <_freemangordon> hmm"your computer is running webgl slowly..." Jul 04 16:24:22 I saw a bug in moz bugzilla mentioning lack of hw accel on harm Jul 04 16:24:22 <_freemangordon> I hope to find a way to disable it afterwards :D Jul 04 16:24:35 ZogG_laptop: LOL Jul 04 16:25:13 <_freemangordon> I still faile to see why it loads the sites so slow Jul 04 16:25:20 btw zeq _freemangordon you update the newest fennec? afaik new one for android has better UI Jul 04 16:25:45 ZogG_laptop: new one for Android has a native Android UI Jul 04 16:26:28 so in order to have one we need to reqrite the UI only? Jul 04 16:26:30 _freemangordon: did you try running a packet sniffer? Jul 04 16:26:46 <_freemangordon> no Jul 04 16:26:50 in similuar way damagedspine is doing for wazr Jul 04 16:28:35 ZogG_laptop: microb is pretty much equivilent, but stuck with an old version of Gecko Jul 04 16:28:56 not much we can do with that unless Nokia release the source Jul 04 16:29:21 so, yes a new UI would have to be written to make it like the Android version. Jul 04 16:29:25 "your computer is running webgl slowly..." also on N9 Jul 04 16:29:29 <_freemangordon> zeq: did you see romaxa's questions for native Qt UI? Jul 04 16:29:51 no, what did he say? Jul 04 16:30:06 aah :P Jul 04 16:30:08 is he writing one? Jul 04 16:30:58 [12:07] freemangordon: not sure how much you interested in qt development UI, but I have qtmozembed project which basically does things similar to webkit2, startup equivalent to microb or normal qt application, scrolling zooming 60fps, similar to android native fennec, but using Qt embedding APIS Jul 04 16:30:58 [12:08] romaxa: having microb replaced with a modern browser is our ultimate goal ;) Jul 04 16:31:27 [12:08] not that microb is bad, its just gecko behind it getting old Jul 04 16:31:27 [12:09] plus I prefer Qt over GTK for new development Jul 04 16:31:27 [12:09] freemangordon: yep, main problem with qtembedding APIs is that someone need to write Qt browser UI... like moz guys did with android native UI Jul 04 16:31:27 [12:10] yep, I saw your patches, any intention to upstream that? Jul 04 16:31:27 [12:10] freemangordon: I was talking about qtmozembed which is using all latest tech from mozilla central, off main thread compositing et.c. Jul 04 16:31:27 [12:10] yes, I look at your wiki page where are the instruction for building for harm Jul 04 16:31:27 [12:10] freemangordon: I tried to push it into upstream, but bsmedberg by some unclear reason against that Jul 04 16:31:29 [12:11] yeah, saw the bug :) Jul 04 16:31:29 [12:11] *I saw Jul 04 16:33:06 It is not very clear what intensions romaxa has, but the fact that he set up my toolchain in his SB sounds promising Jul 04 16:33:11 I wonder why the upstream resistance? Jul 04 16:33:50 I'm interested in building the ui if you're looking for hands, that is if it's possible to do it in QML Jul 04 16:33:50 you ask? why upstream guys reject patches? comeon buddy :P Jul 04 16:34:09 politics :) Jul 04 16:34:30 qwazix_900: More hands are certainly welcome Jul 04 16:35:05 freemangordon: do you think QML would be fast enough? Jul 04 16:35:26 but romaxa deffinitely shows interest in helpng us with n900 port Jul 04 16:35:28 zeq Jul 04 16:35:32 oops Jul 04 16:35:50 zeq: I don't think we need anything but Qt for that Jul 04 16:35:56 qwazix_900: ^^^ Jul 04 16:36:11 you mean qtwidget? Jul 04 16:36:26 Qt in CSSU is hildonized to the point you only need to setup the correct object name Jul 04 16:36:34 qwazix_900: yes Jul 04 16:36:46 or QtGraphicsWidget more precisely Jul 04 16:36:54 I'll have to brush my skills on that Jul 04 16:36:56 foe egl HW acceleration Jul 04 16:37:19 it's great we've encouraged romaxa to rejoin the N900 efforts :) Jul 04 16:37:26 yes Jul 04 16:37:29 wasted on Harmattan ;) Jul 04 16:37:33 :P Jul 04 16:37:51 I just thought that qml would be portable to harmattan and nemo too Jul 04 16:38:02 freemangordon: where's your gravity gun :P Jul 04 16:38:04 (And I like it more) Jul 04 16:38:06 qwazix_900: yeah, you have your point Jul 04 16:38:27 lets see, what theme QML apps use? Jul 04 16:38:36 qwazix_900: qt is just a portable, considering the rest has to be compiled anyway Jul 04 16:38:54 freemangordon: afaik they're native too Jul 04 16:39:02 (might be wrong????) Jul 04 16:39:08 zeq: Nokia in their ethernal wisdom made pure Qt apps not looking very well on Harm AFAIK Jul 04 16:39:08 hildonizing qt-components is another good-looking project Jul 04 16:39:17 oh Jul 04 16:39:31 Nokia.... Jul 04 16:39:43 freemangordon: yep, that's the problem. We'd have to skin the whole UI by hand Jul 04 16:39:49 dhbiker: in my hand of course Jul 04 16:40:47 qwazix_900: and it does not make sense, as we will have enough problems even without that ;) Jul 04 16:41:29 hmm, on the other hand... right now fennec xul Qt uses pure QWidget Jul 04 16:41:56 and that's why all the mess with rotation Jul 04 16:42:00 yes Jul 04 16:42:33 qwazix_900: we (freemangordon) finally got that working properly Jul 04 16:42:34 zeq: Qt on harm is not aware of orientation changes. Smart , ain't? Jul 04 16:43:17 freemangordon: really? explains why we hit problems Jul 04 16:43:24 zeq: yeah, I noticed, congrats Jul 04 16:43:34 so you either use MT or listen to the sensors Jul 04 16:43:43 *MTI Jul 04 16:43:48 how does that work with MT? Jul 04 16:44:09 NFC, I never write an app for Harm Jul 04 16:44:51 I remember seeing in the code som MT 180deg rotation thing Jul 04 16:44:52 * freemangordon feels hes English skills reduced by half as a result of his birthday night :D Jul 04 16:45:01 :D Jul 04 16:45:08 it was good then? :) Jul 04 16:45:50 fennec is finally built Jul 04 16:45:57 zeq: but of course, a rock bar with friends, JD and tabble football, what could be better Jul 04 16:46:04 *table Jul 04 16:46:12 sounds good Jul 04 16:47:02 hmm, i am going to try egl with the changes romaxa proposed Jul 04 16:47:05 anyway, congrats on your birthday :) Jul 04 16:47:13 thanks Jul 04 16:48:06 freemangordon: what changes? Jul 04 16:48:32 freemangordon: shipit Jul 04 16:48:50 vi__: what? Jul 04 16:48:56 fennec Jul 04 16:49:07 zeq: why wasted on harmattan, it can be made for both in Qt i think =) Jul 04 16:49:18 ZogG_laptop: not really Jul 04 16:49:27 for harm it must use MTI Jul 04 16:49:27 on the ui front, qml is much faster to code, runs in simulator so no need for building each time you move something a pixel left/right, doesn't need scratchbox, and doesn't have all the c++ limitations. AFAIK we can load assets from maemo theme so it should look native. If it turns out to be too slow Jul 04 16:49:31 freemangordon: because of hildon and MTF thing? Jul 04 16:49:46 yep Jul 04 16:49:49 we'll know early enough to move to Qt-widget Jul 04 16:49:54 oh hey, there's actually a gcc for the n900 now Jul 04 16:49:54 neat Jul 04 16:50:06 * freemangordon is afk Jul 04 16:50:17 freemangordon: but i think it can still be done with less sweat than writting everything separetly Jul 04 16:50:32 wtf, why is gcc not installed as "gcc"? Jul 04 16:51:00 ZogG_laptop: the git is public, go right ahead. Jul 04 16:51:05 by c++ limitations I mean having the widges parented correctly etc, not actual programming limitations Jul 04 16:51:14 vi__: after you sir Jul 04 16:51:22 ZogG_laptop: o0 Jul 04 16:51:23 s/widges/widgets Jul 04 16:52:16 person1: I want you to do this thing my way! Jul 04 16:52:30 person2: You are free to do it your way if you wish. Jul 04 16:52:37 vi__: i was discussing my point of view, i have noot enuf skills for doing anything useful, and just asked few questions. And i don't believe that — you can always fork or do it yourslef is always useful, that's all Jul 04 16:52:38 person1: NO, U DO IT. Jul 04 16:52:44 person2: eh? Jul 04 16:53:03 vi__: i didn't enforce or asked to do it for me Jul 04 16:54:24 vi__: afaik for now harmattan ui was rewritten or partly rewritten as understand for harmattan and maemo part is missing. I made a proposal and just said my opinion. no more no less. i think you got me wrong, pal =) Jul 04 16:55:34 Another crazy idea would be to go chromeless and write the ui in html/js Jul 04 16:56:36 don't know how better that would be than xul though. I suspect xul's heaviness is due to all the multiplatform code legacy it has acquired Jul 04 16:56:55 If that's the case we would see major improvement Jul 04 17:03:15 it's not that XUL can't be, or isn't fast. It's just different. It's pretty much another whole toolkit. Jul 04 17:05:01 they say that camino is pretty faster than ff on mac (camino is native cacao+gecko) Jul 04 17:05:17 microb faster than fennec Jul 04 17:05:59 native android ui faster by scales than fennec-xul on android Jul 04 17:06:30 seems to me that xul is the culprit here Jul 04 17:06:31 qwazix_900: actually, once we get hw accel working properly in fennec, it should be faster than microb Jul 04 17:06:45 at the moment, there isn't much it it Jul 04 17:06:57 microb is slightly faster Jul 04 17:07:21 although that depends heavily on what you're measuring Jul 04 17:07:29 for JS fennec blows microb away Jul 04 17:07:38 zeq: of course but yo have to take into consideration how much faster gecko is now vs 1.9x Jul 04 17:07:51 *you Jul 04 17:07:52 for just interactive response microb is slightly ahead Jul 04 17:08:38 ofc microb is preloaded so something for the microb scale there Jul 04 17:09:03 and uses less memory Jul 04 17:09:17 or at least, less additional memory Jul 04 17:09:28 since the backend is always loaded Jul 04 17:10:42 how hard would it be to try building chromeless and seeing what's the relative speed? Jul 04 17:13:52 why is microb closed-source, btw? Jul 04 17:15:09 because it seems nokia closed down every maemo version more and more Jul 04 17:15:32 is it elop's fault? Jul 04 17:15:45 it was fully open back in the N8x0 days Jul 04 17:16:55 kerio, it's a differentiation thing. Jul 04 17:18:12 It was to be opened, but never did. Could be his fault but I doubt he'll even know the word microb. Jul 04 17:19:02 he probably did set policy though Jul 04 17:19:25 otherwise, where did meltemi go? Jul 04 17:19:38 GeneralAntilles: It could be argued that hildon-desktop is a differentiation thing but that's open Jul 04 17:19:59 zeq: south Jul 04 17:20:43 meltemi is a north wind blowing late summer over the aegean sea Jul 04 17:20:54 :) Jul 04 17:21:21 perfectly named Jul 04 17:21:34 indeed Jul 04 17:22:28 infobot, what is meltemi? Jul 04 17:22:30 qwazix_900: what are you talking about? Jul 04 17:22:56 infobot, meltemi is is a north wind blowing late summer over the aegean sea Jul 04 17:22:56 qwazix_900: okay Jul 04 17:24:09 infobot no, meltemi is a north wind blowing late summer over the aegean sea Jul 04 17:24:09 qwazix_900: okay Jul 04 17:24:18 infobot, what is meltemi? Jul 04 17:24:18 [meltemi] a north wind blowing late summer over the aegean sea Jul 04 17:25:34 qwazix_900, the widgets aren't open. Jul 04 17:26:24 zeq: any idea if we could hack cursor mode into fennec? Saw any relevant code? Jul 04 17:27:12 maybe fennec has code upstreamed from microb, Jul 04 17:28:14 I didn't see anything, but you can select and copy text Jul 04 17:28:23 by touch-hold Jul 04 17:29:21 yes but you cant use websites that need hover (and not click eg pulldown menus) or drag (eg drawing) Jul 04 17:29:42 qwazix_900: I do like that feature in microb Jul 04 17:30:03 although, you could always use a mouse ;) Jul 04 17:30:52 does a mouse behave like a mouse in fennec? or like a finger? never tried that Jul 04 17:32:20 I think my gf has a bt mouse, maybe I'll give it a go later. Jul 04 17:33:41 I bet it works like a finger. Hover will work but not drag. Jul 04 17:37:16 qwazix_900: it works like finger for scrolling, otherwise it works like mouse. Not tested it much though Jul 04 17:46:42 ok, thumb2-compiled busybox-power is in cssu-thumb repo, not tested it, just compiled the code iDont sent me Jul 04 17:51:55 webm doesn't seem to be working Jul 04 17:53:32 freemangordon: although it has revealed something interesting Jul 04 17:54:10 I selected fullscreen, an when I managed to exit, fennec was running with Hildon status bar, and reduced chrome. Jul 04 17:54:13 ?!? Jul 04 17:54:29 both portrait and landscape Jul 04 17:54:33 zeq: any idea which .desktop to edit so fennec to start from applications? Jul 04 17:54:33 vi__: let me correct that .desktop entry and I will upload it in cssu-thumb for you ;) Jul 04 17:55:10 I'll have a look Jul 04 17:55:19 zeq: something bad is happening with my inet, not sure I have all of your mesages Jul 04 17:55:28 [20:51] zeq: any idea which .desktop to edit so fennec to start from applications? Jul 04 17:55:29 [20:52] vi__: let me correct that .desktop entry and I will upload it in cssu-thumb for you ;) Jul 04 17:55:40 zeq: where did you select fulscreen? Jul 04 17:55:42 I got those Jul 04 17:55:49 aah, ok Jul 04 17:55:54 I was testing webm Jul 04 17:56:02 video never started, just the spinner Jul 04 17:56:14 how to test that? Jul 04 17:56:18 I held on the "video" and selected fullscreen Jul 04 17:56:49 http://www.google.co.uk/url?sa=t&rct=j&q=&esrc=s&source=web&cd=6&ved=0CF8QFjAF&url=http%3A%2F%2Fwww.wellesley.edu%2F~kenny%2Fhtml5%2Fvp8.html&ei=E4L0T_WWC8yL4gTLyb3SBg&usg=AFQjCNFD7JSVPU5iH6-or6xX-q9Nmz5jsA&sig2=ZOqfQhBi76zO-MHvJ5EAeQ Jul 04 17:56:57 yuck Jul 04 17:57:13 I googled for a webm test :) Jul 04 17:57:27 now it's better :D Jul 04 17:57:29 tested in firefox on my laptop first Jul 04 17:57:42 you did the same? Jul 04 17:58:06 weird huh? Jul 04 17:58:15 how can we toggle that??? Jul 04 17:58:25 no, i meant telling me "how" instead of pasting the link Jul 04 17:59:12 http://www.webmfiles.org/demo-files/ Jul 04 17:59:13 another Jul 04 17:59:30 I did supposedly compile it with gstreamer enabled though Jul 04 18:00:43 My chrome is now like the Hamattan version without the close and dashboard buttons Jul 04 18:02:36 ./mobile/xul/installer/debian/fennec/usr/share/applications/hildon/fennec.desktop Jul 04 18:02:40 guys, am I allowed to put LD_LIBRARY_PATH in exec line in .desktop file? Jul 04 18:03:11 doesn't maemo use ld.so.conf? Jul 04 18:03:33 it uses, but qtm12 is not there Jul 04 18:03:57 shouldn't it be? or not? Jul 04 18:04:12 what is the policy? Jul 04 18:04:17 rpath? Jul 04 18:05:01 zeq: NFC, don;t want to lose time on that now, just want it working Jul 04 18:06:56 can i disable the Ovi repository safely? Jul 04 18:09:11 well LD_LIBRARY_PATH works Jul 04 18:09:30 do we have a webm codec? Jul 04 18:10:00 hmm, but I don't know which file to edit, seems .desktop file is generated in runtime Jul 04 18:10:07 zeq: don't know Jul 04 18:10:58 /mobile/xul/installer/debian/fennec.desktop.in Jul 04 18:11:01 zeq: it could be that missing signal ;) Jul 04 18:11:39 freemangordon: could be Jul 04 18:11:52 the video player runs, but the video never starts Jul 04 18:12:03 ooh, blind me, i looked in that directory 10 times, didn't see that file :D Jul 04 18:12:14 :) Jul 04 18:14:23 meh, HAM just found some updates that aren't updates Jul 04 18:14:33 silly HAM Jul 04 18:17:48 freemangordon: I think it must be the missing signal Jul 04 18:18:02 yep Jul 04 18:18:03 what a pita Jul 04 18:19:04 http://lists.freedesktop.org/archives/gstreamer-commits/2011-February/047159.html Jul 04 18:19:08 freemangordon: what speed did you say you got for your desktop? Jul 04 18:19:27 zeq: should not be, just find a way to workaround it with what we have Jul 04 18:19:36 zeq: 820ms Jul 04 18:19:47 we could patch the support into our gstreamer Jul 04 18:19:58 but today i tested google chrome, it did about 450 ms Jul 04 18:20:06 zeq: not, don't :) Jul 04 18:20:15 better patch fennec Jul 04 18:20:24 I got 715.2ms with firefox on my laptop Jul 04 18:20:36 that will become unmaintainable Jul 04 18:20:39 this is highly optimized Gentoo though :) Jul 04 18:20:54 well, this is wondoze xp here Jul 04 18:21:13 could be the reason Jul 04 18:21:17 core2duo 2.4GHz Jul 04 18:21:44 hmm, it could be optimized for intel, not AMD ;) Jul 04 18:22:03 probably Jul 04 18:22:16 you don't want to see my CFLAGS/LDFLAGS ;) Jul 04 18:22:39 as phenom II should go circles around core2 Jul 04 18:23:26 gcc-4.7.1 with every go-faster switch turned on :) Jul 04 18:23:53 zeq: I am compiling right now, could you grep for "source-setup"? Jul 04 18:24:03 ok Jul 04 18:24:09 to see which file to modify Jul 04 18:24:48 depend on >= 0.10.33 as that's when the playbin2 source-setup signal was Jul 04 18:24:54 yup Jul 04 18:25:08 content/media/gstreamer/nsGStreamerReader.cpp: Jul 04 18:25:19 thanks Jul 04 18:25:45 you think this would be better than patching in support maemo side? Jul 04 18:26:00 yep Jul 04 18:26:12 lemme try to workaround it Jul 04 18:26:17 ok, just just checking :) Jul 04 18:26:52 http://talk.maemo.org/showthread.php?t=85242 Jul 04 18:27:19 no more maemo device from now on? Jul 04 18:27:37 jean_brat: you are surprised? Jul 04 18:28:09 yes kinda Jul 04 18:28:23 fennec 9a1 did 3207 on the N900 oc'ed to 1Ghz Jul 04 18:28:24 was hoping N950 official release Jul 04 18:28:53 qwazix_900: sunspider? Jul 04 18:29:05 qwazix_900: 16a1 does 5000 OC to 805 Jul 04 18:29:18 I'm not going to test at 1Ghz Jul 04 18:29:22 yep Jul 04 18:29:23 :D Jul 04 18:29:39 I like my N900 working ;) Jul 04 18:29:45 stock got to 5140 Jul 04 18:29:52 ok Jul 04 18:30:07 so we're slightly behind the old js jit Jul 04 18:30:44 but this is with software only rendering atm Jul 04 18:30:58 hw offload will give the CPU more time Jul 04 18:31:26 zeq: there is no rendering in sunspider ;) Jul 04 18:31:39 freemangordon: but if the CPU is busy rendering... Jul 04 18:32:03 yeah Jul 04 18:32:12 for some reason fennec never sleeps... Jul 04 18:32:25 we'll have to look into that at some point Jul 04 18:32:42 let's see how does the N9 Jul 04 18:32:59 I'm going to see if I can get that fullscreen exit button working from the qt example code Jul 04 18:33:27 (since I now know there is a non-fullscreen mode already there!) Jul 04 18:47:54 5632 on the N9, 4328 on the N950 Jul 04 18:50:31 hmm, that's funny Jul 04 18:50:40 1GHz,meh Jul 04 18:50:44 ??? Jul 04 18:50:49 doesn't it use hardfloat? Jul 04 18:50:58 yes, it uses Jul 04 18:51:07 I expected it to be faster Jul 04 18:51:18 BTW i think I workarounded gstreamer Jul 04 18:51:23 :D Jul 04 18:51:44 shall I push the configure.in change? Jul 04 18:51:48 lemme check it Jul 04 18:51:55 zeq: yes Jul 04 18:53:09 ok pushed Jul 04 18:53:34 it just moves things around a little so it knows it's building for maemo5 before it tests for gstreamer Jul 04 18:53:35 thanks Jul 04 18:54:41 zeq: WTF? 1 files changed, 154 insertions(+), 150 deletions(-) Jul 04 18:54:55 like I said, moves things around a little :) Jul 04 18:55:03 a little moves around? Jul 04 18:55:07 :D Jul 04 18:56:37 samsung galaxy note 1539 !!! Jul 04 18:56:40 * zeq is afk for a bit Jul 04 18:57:06 qwazix_900: ff? Jul 04 18:57:10 qwazix_900: what CPU is that? Jul 04 18:57:37 1.5 dual exynos Jul 04 18:57:56 well, seems reasonable then Jul 04 18:57:56 zeq: yep Jul 04 19:02:06 after a reboot, N9 dropped insignificantly to 4973 Jul 04 19:02:57 ok, lets check google maps with egl enabled Jul 04 19:06:13 is the ovi* stuff removable? Jul 04 19:06:44 what about mediaplayer? Jul 04 19:06:47 cant you just stop removing packages Jul 04 19:06:50 no, i better keep that Jul 04 19:06:54 Lava_Croft: but... ovi is evil! Jul 04 19:07:00 Lava_Croft: and i'm keeping backups now Jul 04 19:07:04 its not evil, its just badly made Jul 04 19:07:08 EVIL Jul 04 19:07:09 well, badly presented mostly Jul 04 19:07:27 also, having a store on an apt repo is silly Jul 04 19:09:24 Lava_Croft: ovi store, aka angry birds download server Jul 04 19:09:48 and did anyone ever use any of the other stuff? Jul 04 19:09:55 whats wrong with angry birds, apart from being a bit of a silly game? Jul 04 19:10:13 well, yeah, thats another point:) Jul 04 19:10:20 angry birds is great Jul 04 19:10:32 was angry birds space released for maemo? Jul 04 19:11:33 well, with egl enabled fennec crashen in a 1-2 minutes Jul 04 19:12:31 ooh, what's "usb console" in backupmenu? Jul 04 19:12:46 ftdi serial console? Jul 04 19:24:08 <_freemangordon> WTF is with my inet? Jul 04 19:31:14 when you guys say to file down the "teeth" of the microusb part of the cables Jul 04 19:31:20 does it have to be completely flat at the end? Jul 04 19:31:59 mine is Jul 04 19:32:05 mine is not Jul 04 19:32:22 i still have two kinda bumps Jul 04 19:32:24 :) Jul 04 19:32:30 but they're small Jul 04 19:32:37 at least, much smaller than the standard teeth Jul 04 19:32:46 btw, today marks 1 year of ownership of my N900 Jul 04 19:32:49 \m/ Jul 04 19:34:32 my usb is still good. battery is now just about 1200mAh, and N900 is still my primary internet & phone device ;) Jul 04 19:34:40 cssu-stable + kp5- Jul 04 19:34:46 kp50 Jul 04 19:35:11 http://wiki.maemo.org/BrowserBenchmarks Jul 04 19:39:42 hmm, this N900 I have here also turned 1 year recently (24 june). So far, USB seems to be ok (unlike the first one which died after 6 months). But I also filed those barbs on the cable in the hope it increases the lifetime a bit. No moving around when it is charging etc. Jul 04 19:45:53 N9 stock browser is fast Jul 04 19:46:19 any tips for dealing with technology and sand? Jul 04 19:46:28 it is but it has a nasty bug Jul 04 19:46:30 and most importantly, the mix of both Jul 04 19:47:31 qwazix_900: nice, you can add n900 fennec16a1 6438ms @ stock frequency Jul 04 19:47:32 when you delete a word with backspace it comes back from the grave after you've deleted the space before it and that goes into endless loop Jul 04 19:47:54 freemangordon: doing it asap Jul 04 19:47:56 i might add, this baby has been flashed only once (the day it was bought). in the meantime, i had a couple of kernel upgrades and downgrades, loss of many /dev/ nodes (while playing with the bootmenu+uboot combo) .. for the last issue, i'm eternally grateful for RescueOS Jul 04 19:48:25 zeq: why it is so slow compared to v9? Jul 04 19:49:30 AFAIK they changed the jit engine Jul 04 19:50:09 s/jit/JIT/ Jul 04 19:50:10 zeq meant: AFAIK they changed the JIT engine Jul 04 19:50:22 aah Jul 04 19:50:32 with a slower one? nice move :D Jul 04 19:50:45 depends on the CPU target Jul 04 19:50:58 omfg so smooth Jul 04 19:51:07 i filed down the barbs on the cable Jul 04 19:51:11 I just retested my laptop set at highest (stock) frequency Jul 04 19:51:16 634.7ms +/- 39.5% Jul 04 19:51:37 (rather than ondemand) Jul 04 19:51:43 freemangordon: not a slower one, but some c++ version of some piece of code used for arm are not as fast as ssx version of the same code used for x86 Jul 04 19:51:54 freemangordon: are you testing the N900 with ondemand governor? Jul 04 19:52:04 peetah: but it is supposed to be NEON optimized Jul 04 19:52:07 zeq: yes Jul 04 19:52:19 maybe I should lock the frequency? Jul 04 19:52:26 freemangordon: yes but for x86 it is directly written in assembly Jul 04 19:52:27 kerio: i think you lose nothing by filing them smooth (unless you want to hoist your N900 ;) Jul 04 19:52:37 it' not just c++ compiler optimization Jul 04 19:52:37 peetah: the same for arm Jul 04 19:52:41 I think the published benchmarks are with locked frequency Jul 04 19:52:52 aah, lemme try then Jul 04 19:53:33 Sicelo: it's not completely smooth though Jul 04 19:53:38 there's still a tiny bump Jul 04 19:53:41 is it ok? Jul 04 19:54:20 it is, otherwise you risk cable falling when you move the device Jul 04 19:54:34 I am using mine more than 2 years that way Jul 04 19:55:40 so... let's say i have a n8's OTG usb adapter Jul 04 19:55:51 how would i make it fit on the n900's port? Jul 04 19:56:22 freemangordon: done Jul 04 19:56:59 kerio: order one that fits (galaxy S2) Jul 04 19:57:05 qwazix_900: failing that? Jul 04 19:57:29 qwazix_900: thanks, I will edit it if there is a difference with locked frequency Jul 04 19:57:36 I wouldn't mess with DIY plugs in the fragile N900 port Jul 04 19:57:57 given that they're almost free Jul 04 19:58:29 qwazix_900: k Jul 04 19:58:44 wait, how can the galaxy s2 have such an adapter? Jul 04 19:58:50 isn't that against the specs? Jul 04 19:59:56 qwazix_900: yes, there is, 6133 ms Jul 04 20:00:23 okay, shall I change it? Jul 04 20:00:32 I will Jul 04 20:00:50 kerio: I think it is, they don't have an official otg logo Jul 04 20:01:38 but the extra pin is there cause the normal cable with F-F adapter didn't work (not recognised) Jul 04 20:01:43 ready Jul 04 20:02:10 however you won't need it on the N900 Jul 04 20:02:28 I mean the 5th cable in the port Jul 04 20:02:39 lemme check if it actually works Jul 04 20:02:57 I know it fits, but haven't run hen with it Jul 04 20:06:24 freemangordon: it's slower? Jul 04 20:06:46 zeq: no, faster, by about 300 ms Jul 04 20:07:01 I thought you had 5xxx? Jul 04 20:07:21 was 6438ms Jul 04 20:07:44 seems gcc 4.7 will be of benefit here Jul 04 20:08:36 unfortunately I am on a vacation from 10th to 16th Jul 04 20:08:55 hmm, well, not unfortunately :D Jul 04 20:16:51 I installed the latest stable CSSU firmware. No orientation lock? I had previously installed CSSU and it had orientation lock. Is that because that was not a stable version? I should not limit myself to the stable version if I am to get the orientation lock features? Jul 04 20:17:23 it's a separate install Jul 04 20:18:07 hen is in devel? Jul 04 20:21:08 freemangordon: going somewhere nice? Jul 04 20:21:59 should have the compiler ready for when you return :) Jul 04 20:22:10 yeah, kavarna rock fest. though no good bands this year :(. But still, sea, sand, beer, stuff like that :P Jul 04 20:22:16 zeq: but of course Jul 04 20:29:07 freemangordon: did your gstreamer compatibilty changes work? Jul 04 20:30:10 not from the first try, i used the wrong function, g_object_connect instead of g_signal_connect, now re-compiling with correct one Jul 04 20:30:36 BTW how we exit fullscreen? Jul 04 20:31:26 I used fn-backspace I think Jul 04 20:31:45 though I am not sure whether to do g_object_unref in callback for source object, zeq, any thoughts? Jul 04 20:33:43 I'll take another look at the patch introducing the feature Jul 04 20:34:35 zeq: https://bugzilla.mozilla.org/show_bug.cgi?id=758338 Jul 04 20:34:37 04Bug 758338: was not found. Jul 04 20:35:07 kerio: I can confirm it works with a $2 galaxy S2 hostmode cable Jul 04 20:35:16 neat Jul 04 20:35:32 well, i suppose that it doesn't violate usb specs if it's actually proper host mode Jul 04 20:35:38 zeq: http://pastebin.com/qZJgJUYY Jul 04 20:35:49 it sounds weird though, if it's the same port then it should be OTG Jul 04 20:38:56 serialisation of requests? Jul 04 20:39:18 re. 758338 Jul 04 20:40:04 zeq: i have sound :P Jul 04 20:40:30 but mmedia is broken in nighly anywa, flash behaves in the same way Jul 04 20:40:38 i think it violates in in the sense that hosts should be usb-a not usb-b but I havent read the spec Jul 04 20:40:48 only a stuttering sound, the same for webm Jul 04 20:41:02 :D Jul 04 20:41:29 going to commit and push Jul 04 20:41:38 could well be the same bug Jul 04 20:41:52 could be Jul 04 20:42:03 otherwise it behaves according to spec: if it detects the extra pin it enables host Jul 04 20:43:17 looks ok to me. Jul 04 20:46:15 zeq: pushed Jul 04 20:59:21 zeq; could you pull the latest changes from m-c? lets hope mmedia is fixed. Jul 04 21:01:39 ...i just did a pstree Jul 04 21:01:42 what the FUCK Jul 04 21:05:07 vi__, I haven't investigated what it does on N8x0 Jul 04 21:06:02 kwtm: you can install orientation-lock on CSSU-S Jul 04 21:06:18 ooh, missed Sicelo telling that Jul 04 21:06:39 freemangordon: ok Jul 04 21:08:16 freemangordon: need to merge Jul 04 21:08:23 be a few minutes Jul 04 21:08:29 ok Jul 04 21:19:28 freemangordon: something weird happening with the merge... I'll get it fixed Jul 04 21:22:57 ok Jul 04 21:23:45 the last rebase didn't go cleanly did it? you fixed it up. Jul 04 21:24:00 I obviously pushed without noticing Jul 04 21:24:08 hmm, it went ok Jul 04 21:24:18 zeq: check your branch Jul 04 21:25:21 zeq: i did some pushes, it could be that, i.e. .patch I sent you is merged. but there should be no problem aiui Jul 04 21:26:17 with the rebase some patches are "empty" with their changes already applied.... I don't know, it's weird as I said. Jul 04 21:26:46 ok Jul 04 21:27:56 you can try to reset to HEAD and pull after that. but I am not exactly git guru, so might be talking nonsense Jul 04 21:30:19 hang on, trouble with the washing machine :S Jul 04 21:32:43 :) Jul 04 21:34:00 do you guys use something to watch youtube videos? Jul 04 21:34:05 fuse Jul 04 21:34:33 kerio: yep, microb, qmltube Jul 04 21:34:49 freemangordon: i meant apart from microb ¬.¬ Jul 04 21:35:07 qmltube Jul 04 21:35:53 actually I've been meaning to mention this, but a CSSU update some time back broke the built-in qmltube-widget player Jul 04 21:35:54 zeq: automatic fuse? Jul 04 21:36:03 in the plug, 13A! Jul 04 21:36:14 a little worrying Jul 04 21:36:26 yeah :( Jul 04 21:36:52 seems to be working Jul 04 21:37:01 washing machine, that is :P Jul 04 21:37:19 zeq: well, it has something to do with the fact marxian was trying to do it compatible for bot At 4.7.0 (stock) and 4.7.4(cssu) Jul 04 21:37:39 zeq: if it is working, don't touch it, you know :P Jul 04 21:45:58 Hi, (thought it would be time to pop up here again) Jul 04 21:46:58 helo desu Jul 04 21:47:19 wai wha Jul 04 21:52:13 freemangordon: pushing it now, can you check it carefully when you pull please Jul 04 21:53:44 I think it was your prepare for upstream patches conflicting Jul 04 21:54:00 ok Jul 04 21:54:05 not sure how that happened Jul 04 21:54:37 I think I took the right versions of the code, but please do check. Jul 04 21:55:09 ok Jul 04 21:56:17 CONFLICT (content): Merge conflict in widget/qt/nsWindow.cpp ;). going to check it Jul 04 21:56:38 ok :) Jul 04 21:58:14 just push whatever needs merging and I'll pull straight away so we're in sync Jul 04 22:00:43 zeq: pushed Jul 04 22:02:03 zeq: why do you think orientationChanged is defined only when QTM is defined? Jul 04 22:03:46 zeq: it seems your commits removed mine :( Jul 04 22:05:23 I'll recheck the whole code Jul 04 22:08:25 freemangordon: somehow git confused them and tried to apply them at the same time?!? Jul 04 22:08:48 yeah, furtunately it is not a big change Jul 04 22:09:17 anyway, as for your question, it *was* because of the orientation API usage from libqtm. For us there's no good reason actually. Jul 04 22:10:09 we should move our rotation code outside QTMOBILITY and just protect it with MAEMO==5 Jul 04 22:10:35 we don't use any QTMOBILITY for rotation support Jul 04 22:11:37 but it needs to be in every file, otherwise there's build breakage if it's undefined, which is what my commit fixed. Jul 04 22:12:45 the header declaration wasn't applied because QTMOBILTY wasn't set, while it was used in the cpp files Jul 04 22:13:22 of course QTMOBILITY should have been set, but that was because I'd lost my PKG_CONFIG_PATH variable Jul 04 22:13:44 makes sense? Jul 04 22:16:14 decaration should be QTMOBILITY || MAEMO==5 Jul 04 22:16:46 that's what I should have done :) Jul 04 22:17:06 freemangordon: ^ Jul 04 22:34:20 zeq: yeah Jul 04 22:34:40 ok, going to check what's happened with the code Jul 04 22:38:31 zeq: it seems I should reapply some changes Jul 04 22:44:36 freemangordon: go ahead Jul 04 22:44:51 i'm going :) Jul 04 22:51:18 zeq: I still fails to see what is the reason for #if (MOZ_PLATFORM_MAEMO == 5) && defined(MOZ_ENABLE_QTMOBILITY) Jul 04 22:51:18 QObject::connect(QApplication::desktop(), SIGNAL(resized(int)), widget, SLOT(orientationChanged())); Jul 04 22:51:36 #if (MOZ_PLATFORM_MAEMO == 5) should be enough Jul 04 22:55:22 zeq: pushed (without && defined(MOZ_ENABLE_QTMOBILITY) :) ) Jul 04 22:55:25 Is that what I put i there? Jul 04 22:55:32 yep Jul 04 22:55:32 It should be || Jul 04 22:55:48 but why? we don't want this for maemo6 Jul 04 22:55:50 sorry, wrong file Jul 04 22:55:58 :S Jul 04 22:56:06 that is in nsWindow Jul 04 22:56:07 yes that should just be MAEMO5 Jul 04 22:56:27 well, you may want to pull, so we have the same code Jul 04 22:56:33 and where the slot is defined should be || Jul 04 22:56:47 pulling now Jul 04 22:56:54 in mozqwidget.h? Jul 04 22:56:59 lemme check Jul 04 22:56:59 yes Jul 04 22:58:17 aah, i see, now it is #ifdef MOZ_ENABLE_QTMOBILITY only. But dows it make sense to build for maemo5 without QTM? Jul 04 22:58:21 *does Jul 04 22:58:46 zeq: ^^^ Jul 04 22:58:47 it breaks the build if QTMOBILITY isn't enabled Jul 04 22:59:22 logically it should be || even if it doesn't really make sense to build for Maemo5 without mobility Jul 04 22:59:47 agree, will change and push it Jul 04 23:02:09 there wasn't too much broken :) Jul 04 23:02:26 yep Jul 04 23:03:25 there certainly are frequent updates to m-c Jul 04 23:03:42 constant flux Jul 04 23:03:45 higgs Jul 04 23:03:45 yeah, lots of work is going on Jul 04 23:03:51 * DocScrutinizer higgs Jul 04 23:03:58 DocScrutinizer: surprising news Jul 04 23:04:02 bozon? Jul 04 23:04:11 booz on Jul 04 23:04:22 did they find it? Jul 04 23:04:31 freemangordon: they believe so Jul 04 23:04:41 * DocScrutinizer fires a higgs bozon gun at ACTA Jul 04 23:04:43 or certainly a particle with the expected properties Jul 04 23:04:49 aah, i missed that Jul 04 23:04:58 they don't yet don't if it's the only one... Jul 04 23:05:18 such is particle physics Jul 04 23:05:24 * freemangordon opens LHC wiki page Jul 04 23:11:57 zeq: pushed Jul 04 23:12:20 pulled Jul 04 23:12:51 BTW Exec=LD_LIBRARY_PATH=/opt/fennec-16.0a1:/opt/qtm12/lib /opt/fennec-16.0a1/fennec in .desktop file does not work :( Jul 04 23:13:09 the same works from shell Jul 04 23:13:25 ok, so we'll need to decide whether to: Jul 04 23:13:34 1) use a script launcher Jul 04 23:13:43 2) link with rpath Jul 04 23:13:57 rpath is better I think Jul 04 23:14:02 3) put /opt/qtm12/lib into ld.so.conf Jul 04 23:14:24 1 or 2, but 2 seems better Jul 04 23:14:34 if you know how to do it :) Jul 04 23:14:56 I can look it up, can't remember off-hand :) Jul 04 23:14:57 cflags in debian/rules? Jul 04 23:15:04 or ldflags Jul 04 23:15:14 ldflags probably Jul 04 23:15:28 almost certainly infact Jul 04 23:15:56 hmm, jscript is faster than before, mmedia still broken, lets check flas Jul 04 23:16:00 *flash Jul 04 23:16:55 http://en.wikipedia.org/wiki/Rpath Jul 04 23:18:59 we just need to put -Wl,-rpath=/opt/libqtm12/lib into LDFLAGS I think Jul 04 23:19:21 s/libqtm/qtm/ Jul 04 23:19:22 zeq meant: we just need to put -Wl,-rpath=/opt/qtm12/lib into LDFLAGS I think Jul 04 23:19:35 best get it right! :P Jul 04 23:19:43 dammit man, that is faster than microb :) Jul 04 23:19:59 hw accel? Jul 04 23:20:03 no Jul 04 23:20:14 lemme try it Jul 04 23:20:14 it's going to be crazy quick Jul 04 23:20:22 when it's all working Jul 04 23:20:30 yep Jul 04 23:21:23 actually we already have an roath-link in mozconfig Jul 04 23:21:32 s/roath/rpath/ Jul 04 23:21:32 zeq meant: actually we already have an rpath-link in mozconfig Jul 04 23:23:10 freemangordon: should be able to just add qtm to the rpath-link in mozconfig Jul 04 23:23:33 will you do that? Jul 04 23:23:43 yes, and I'll test it first Jul 04 23:23:51 ok Jul 04 23:24:13 still faster without egl Jul 04 23:24:33 building now Jul 04 23:24:46 egl still broken then Jul 04 23:25:01 probably falling back to sw Jul 04 23:25:14 we need romaxa to fix it Jul 04 23:25:21 yup Jul 04 23:25:52 flash still broken Jul 04 23:25:54 :( Jul 04 23:25:58 shame :( Jul 04 23:26:06 yeah Jul 04 23:27:26 freemangordon: do we need the fennec dir in rpath too? Jul 04 23:27:37 yes Jul 04 23:27:47 I've been adding it to my LD_LIBRARY_PATH, not tried without it Jul 04 23:27:50 ok Jul 04 23:28:00 better stop the build and figure that out Jul 04 23:28:01 needs libxul.so Jul 04 23:33:58 freemangordon: does https work for you Jul 04 23:34:02 ? Jul 04 23:34:25 i think so, but not sure, lemme check Jul 04 23:35:27 I think it might be linking to some of the microb libs Jul 04 23:35:47 check ldd /opt/fennec*/libxul.so Jul 04 23:36:06 what? Jul 04 23:36:43 let me check with LD_LIBRARY_PATH Jul 04 23:36:56 yes, that is what you need Jul 04 23:36:58 ;) Jul 04 23:37:09 https works though? Jul 04 23:37:56 for me just typing something into the url bar then selecting google search fails when it tries to use https://www.google.com Jul 04 23:39:26 yeah, seems https is broken Jul 04 23:40:33 I've been just building with make -f client.mk && make deb. That seems to work fine, but since it automatically sets the version I can't use it with rpath since I won't know the install path in advance... Jul 04 23:41:09 @installdir@ ? Jul 04 23:41:22 or something like that Jul 04 23:41:33 that is what .desktop.in uses Jul 04 23:41:36 maybe... I'l try it Jul 04 23:42:01 have in mind that file starts with #filter substitution Jul 04 23:43:29 hmm, but https to mozilla site works, wtf? Jul 04 23:43:46 interesting Jul 04 23:44:50 is it just https://www.google.com that's broken then? Jul 04 23:44:57 no Jul 04 23:45:34 different encryption type? Jul 04 23:46:03 google uses RC4, 128bit appaerently Jul 04 23:46:24 its is enabled in about:config Jul 04 23:46:59 same for mozilla.org Jul 04 23:47:12 not that then Jul 04 23:47:46 well, that will be fixed upstream soon or later Jul 04 23:48:13 @installdir@ hasn't been expanded :( Jul 04 23:49:20 portrait mode is broken, damn Jul 04 23:49:40 :( Jul 04 23:51:10 hmm, it is working now, wtf? Jul 04 23:51:39 yep, it's ok Jul 04 23:51:51 you sure? Jul 04 23:52:09 yees Jul 04 23:52:24 hmm Jul 04 23:52:28 seems i was runnung it with egl Jul 04 23:52:35 googled issued certificates to themself Jul 04 23:53:09 the right bar was not appearing when scrolled to de left Jul 04 23:53:29 FIQ|n900: that could be a reason Jul 04 23:54:30 but it fails with sec_error_no_memory :D Jul 04 23:57:08 lemme check gitorious Jul 04 23:57:57 is it possible to upgrade the gecko version used by microB? Jul 04 23:58:20 "possible" as in "without the microB source and reverse-engineering doesn't count" Jul 04 23:58:47 should be Jul 04 23:59:10 would it be trivial? Jul 04 23:59:14 well, gitorious https works, could be something with google certificates Jul 04 23:59:25 trivial? don't think so Jul 04 23:59:44 FIQ|n900: you need to match an undocumented ABI Jul 05 00:00:18 ok fine Jul 05 00:00:19 maybe there's enought information it what's available, maybe not Jul 05 00:00:36 FIQ|n900: if we get fennec working ok, I think there will be no need for updated microb. though if we update it, we will free lots of RAM Jul 05 00:00:37 there are some sources Jul 05 00:01:04 freemangordon: this was my idea Jul 05 00:01:18 I think it is possible, but not trivial, we should write our own eal Jul 05 00:01:52 what else depends upon it? modest? Jul 05 00:01:53 the gecko was heavily modified too :/ Jul 05 00:01:56 conversations? Jul 05 00:02:02 Maps Jul 05 00:02:06 ok Jul 05 00:02:11 yeah, lots of stuff Jul 05 00:02:20 right Jul 05 00:02:27 other things used it Jul 05 00:02:29 some we have sources for though? Jul 05 00:02:31 I forgot that Jul 05 00:02:48 isn't maps just JS Jul 05 00:02:54 it is Jul 05 00:02:55 obfuscated JS but still Js Jul 05 00:03:03 JS, even Jul 05 00:03:18 how does it do os calls? Jul 05 00:03:27 which calls? Jul 05 00:03:34 Maps is js? Jul 05 00:03:34 well Jul 05 00:03:40 It doesn't have a binary component? Jul 05 00:03:45 gps, accessing taskswitcher, etc Jul 05 00:03:57 directly from it Jul 05 00:04:06 I think maps is all js? Jul 05 00:04:15 gps is through browser location plugin afaik Jul 05 00:04:21 It paints the screen from the maps files, and decodes them in JS? Jul 05 00:04:42 not sure about task switcher, dbus call maybe Jul 05 00:04:55 yes, but *how* Jul 05 00:05:05 JS can't do that on its' own Jul 05 00:05:22 unless it's modified Jul 05 00:05:38 NFC **** ENDING LOGGING AT Thu Jul 05 02:59:58 2012