IRC log of #maemo-ssu for Monday, 2012-01-02

*** _rd has quit IRC00:22
*** _rd has joined #maemo-ssu00:22
*** wmarone_ has joined #maemo-ssu00:53
*** wmarone has quit IRC00:53
*** wmarone has joined #maemo-ssu00:58
*** wmarone_ has quit IRC00:58
*** trbs has quit IRC01:22
*** jonwil has joined #maemo-ssu02:19
*** jonwil has quit IRC02:23
*** jonwil has joined #maemo-ssu02:24
*** M4rtinK has quit IRC02:34
*** M4rtinK has joined #maemo-ssu03:05
*** M4rtinK has quit IRC03:26
*** M4rtinK has joined #maemo-ssu03:26
*** mirandir has quit IRC03:35
*** M4rtinK has quit IRC03:58
*** amiconn has quit IRC05:20
*** amiconn_ has joined #maemo-ssu05:20
*** amiconn_ is now known as amiconn05:20
*** nox- has quit IRC06:07
*** DocScrutinizer has quit IRC08:02
*** DocScrutinizer has joined #maemo-ssu08:02
*** mirandir has joined #maemo-ssu10:48
*** freemangordon has joined #maemo-ssu10:57
*** MohammadAG has joined #maemo-ssu11:03
freemangordonMohammadAG, ping11:11
DocScrutinizer51moo freemangordon11:13
freemangordonHappy new year11:13
DocScrutinizer51happy new year to you too11:13
MohammadAGfreemangordon: Pong, had some life hiccups11:13
MohammadAGHappy new year everyone11:13
freemangordonDocScrutinizer, still running thumb compiled binaries without a single crash ;)11:14
MohammadAGThe update will be out today11:14
DocScrutinizer51good11:14
MohammadAGfreemangordon: This requires a patched kernel right?11:14
freemangordonMohammadAG, yeah, that is what I was going to ask11:14
DocScrutinizer51any hard numbers on size/speed?11:14
freemangordonMohammadAG, yea11:15
MohammadAGI'm more interested in ABI compatibility11:15
freemangordonwhat do you mean?11:15
DocScrutinizer51indeed, we won't be able to bring it to CSSU11:15
DocScrutinizer51but maybe for powerkernel it's nice11:15
freemangordonDocScrutinizer, the size difference is exactly 19 MB11:16
MohammadAGNo I meant something else11:16
DocScrutinizer51for...?11:16
freemangordonbinaries11:16
DocScrutinizer51errr11:16
MohammadAGWhat binaries are we talking11:16
DocScrutinizer51:-D11:16
MohammadAGWe will eventually have kernel updates in the cssu11:17
freemangordonMohammadAG, Qt, gtk, hildon-desktop, hildon-home, hildon-status-menu11:17
DocScrutinizer51oooh, will we?11:17
MohammadAGBut if we push a thumb kernel, will we break all apps?11:18
freemangordonsure, soon or later11:18
freemangordonMohammadAG, we don't need thumb kernel11:18
MohammadAGIs it like softfp and hardfp11:18
DocScrutinizer51nope, afaik11:18
freemangordonnope11:18
DocScrutinizer51it's just a "bugfix"11:18
DocScrutinizer51enabling thumb binaries11:19
MohammadAGI thought it needs an updated kernel?11:19
DocScrutinizer51yes11:19
DocScrutinizer51bugfix in kernel11:19
freemangordonwell, not updated, just one that has thumb errata workaround11:19
MohammadAGAh11:19
MohammadAGIf we fix it will it require updated bins?11:20
DocScrutinizer51bbl11:20
MohammadAGOr are they optional?11:20
MohammadAGIe11:20
MohammadAGCan we push the fix in pk for non cssu users?11:20
freemangordonno, we don't need updated binaries11:21
freemangordonyou mean push as in?11:21
freemangordonto push omap1 with thumb patch?11:21
freemangordonor to push kp in cssu?11:22
freemangordonMohammadAG ^^^11:22
MohammadAGOmap in casu pk in extras11:22
freemangordonWho will maintain it?11:22
MohammadAGNope11:23
MohammadAGOh11:23
MohammadAGRead that as will we11:23
MohammadAGPali?11:23
freemangordonIt is better to name kp community-kernel and to push it in CSSU11:23
freemangordonhe maintains KP11:23
freemangordon:D11:23
MohammadAGThat needs updated fcam frivrd11:23
MohammadAGYeah11:23
freemangordonAFAIK no11:23
MohammadAGhe'll maintain kp11:24
freemangordonor we can just include fcam drivers in KP11:24
MohammadAGWe'll maintain omap11:24
freemangordonwhy is that?11:24
freemangordonKP has many m,any bugs fixed compared to omap111:24
MohammadAGHmm11:25
freemangordonBTW that wil solve repo saga too11:25
MohammadAGI actually want kp in cssu11:25
freemangordonme too :D11:25
MohammadAGBut feat we'll mess up stuff like fcam11:25
freemangordonand Pali too11:25
MohammadAGFeel11:25
MohammadAGFear11:25
freemangordonwell, that should be easy to be solved11:25
MohammadAGFfs iOS11:25
*** Pali has joined #maemo-ssu11:25
*** Pali has quit IRC11:25
*** Pali has joined #maemo-ssu11:25
freemangordonPali, ping11:25
MohammadAGSomeone needs to maintainfcam11:26
Palipong freemangordon11:26
freemangordonlook at the logs for the last 15 minutes11:26
freemangordonwe're discussing KP in CSSU11:26
Paliok11:28
freemangordonPali, what is ok? :D11:28
PaliI read log11:28
freemangordonok11:29
PaliIn KP is patches directory where are stored all patches11:29
Paliso in CSSU can be used patches which are bug fix only11:29
freemangordonSure, but why?11:29
Paliand what do you want? Include full KP into CSSU?11:30
freemangordonyep, noone is obliged to use OC or anything from KP. And hopefully SR is stable ;)11:31
freemangordonthumb too, all those goodies11:31
freemangordonWhat problems (besides fcam) do you expect?11:32
PaliI'm not using fcam, so I do not know fcam state...11:32
PaliNow I do not see problems in KP11:33
freemangordonMe too, but AIUI it is a matter of re-compilation against KP11:33
Paliit is problem to include fcam drivers into kp (as patch)?11:34
PaliMohammadAG, did you tested patched osso-wlan (wlancond daemon) without packet injection drivers?11:35
freemangordonHave no clue, but I think it is better to keep them separate, as external package. Could make them a part of KP of course11:35
Paliif it will be included in KP, there will be no problems with kernel versions...11:36
freemangordonwell, that external package could be in CSSU too, so an upgrade to take place when is needed, otherwise we could break something for users which don't have fcam installed11:37
Paliok11:37
freemangordonMohammadAG, if we don't plan/intend/want to use thumb in CSSU, it is better to disable thumb errata in CSSU kernel as it leads to loss of performance11:39
freemangordonBut I still think it is better to use thumb, it frees a massive amounts of RAM, and flashspace11:40
freemangordonif Qt is thumb-compiled we can move it back to NAND, having an additional performance gain11:40
freemangordonOn the side note I will have to give you my VMWARE image with gcc 4.6.2 :D11:41
freemangordonMohammadAG ^^^11:42
freemangordonmia again :(11:43
*** FireFly has quit IRC11:49
MohammadAGfreemangordon: I'm in class11:53
MohammadAGAnyway11:53
freemangordonok, sorry11:53
MohammadAGI plan to use thumb, but why VMware?11:53
MohammadAGDo we have to upgrade gcc?11:53
freemangordonbecause that is what I have :D11:53
freemangordonyesm no option11:53
freemangordonyes,*11:53
MohammadAGGot gcc debs?11:54
MohammadAGI dislike virtualizatiob11:54
MohammadAGN11:54
freemangordonno, there are no such debs11:54
MohammadAGAnd how do we plan to upgrade it?11:54
freemangordoni've compiled gcc and binutils from sources11:54
MohammadAGWe'll need to pack them up into debs11:55
freemangordonhmm11:55
freemangordonnot sure about that, we're talking about scratchbox11:56
MohammadAGYes11:56
freemangordonand AFAIK there are no gcc debs for scratchbox11:56
MohammadAGHmm?11:56
MohammadAGHow is it installed then?11:56
freemangordonusing python ot whatever scripts, as a part of scratchbox installation11:57
MohammadAGOh, that'll be a problem11:57
freemangordonwhy?11:57
MohammadAGIs gcc in /usr/bin11:57
freemangordonnope11:57
freemangordondamn11:58
MohammadAGWhat?11:58
freemangordon"/scratchbox/compilers/"11:58
freemangordonanyway11:58
freemangordoncannot write the above without quotes11:59
*** Pali has quit IRC11:59
MohammadAGCan someone dpkg -S /scratchbox/compilers/gcc11:59
MohammadAGYou can, use //11:59
MohammadAG/usr11:59
MohammadAGThat's //usr11:59
freemangordonok, thanks11:59
freemangordonMohammadAG, it is not so simple12:00
*** Pali has joined #maemo-ssu12:00
*** Pali has quit IRC12:00
*** Pali has joined #maemo-ssu12:00
freemangordongcc is not installed via .deb12:00
freemangordonit is a patr of cross-toolchain12:00
MohammadAGI think it's part of scratchbox-compilers12:00
MohammadAGI installed all of scratxhbox from deba12:01
MohammadAGDebs12:01
freemangordoncan you ls -al your //scratchbox//compilers?12:01
MohammadAGLaptop isn't with me12:01
freemangordonyou will see what I mean12:01
MohammadAGAfaik theyre symlinks12:01
*** wmarone_ has joined #maemo-ssu12:01
MohammadAGNot sure to where though, can't remember12:01
freemangordonno, those are real binaries (most of them)12:02
MohammadAGCan you file gcc12:02
*** wmarone has quit IRC12:02
freemangordonunder scratchbox armel?12:02
MohammadAGNo, rhe one in compilers12:03
MohammadAGThr12:03
MohammadAGThe12:03
PaliI tried to replace some scratchbox included programs. I copied program to /usr/bin/<something>, but when I started <something> it still called scraptchbox tool not in /usr/bin/12:03
freemangordonTrust me, it is not so simple12:03
Paliso upgrading gcc via deb packages I think is not possible in scratchbox12:03
freemangordonagree12:03
freemangordonso you have to use my scratchbox as it is12:04
freemangordonthere is an option in sb-menu to setup a new target12:04
freemangordonwhere you choose where toolchain for that target is located12:04
*** DocScrutinizer51 has quit IRC12:09
*** DocScrutinizer51 has joined #maemo-ssu12:10
*** Pali has quit IRC12:13
*** andre__ has joined #maemo-ssu12:13
*** andre__ has joined #maemo-ssu12:13
*** ekze_note has quit IRC12:25
MohammadAGWe'll discuss this when I'm home12:30
freemangordonok12:30
MohammadAGfreemangordon: Should this be in the next update or future ones?12:30
freemangordonfuture12:31
freemangordonthere is too much work to be done12:31
freemangordonand new kernel is a must12:31
freemangordonso we need a new kernel in CSSU first12:32
MohammadAGK again we'll discuss this in 2h1512:33
MohammadAGActually 2h12:33
freemangordonBTW I have a feeling that n900 has only 256 RAM because the initial Nokia intent was to use thumb-compiled binaries :D12:33
MohammadAGUmm12:33
freemangordonok12:33
MohammadAGCan you test open mediaplayer?12:33
MohammadAGWith qt thumb12:33
MohammadAGHow much ram does it use on startup12:34
freemangordonI could try, if you point me to git repo12:34
freemangordonand tell me the command to check the memory usage :D12:35
freemangordonOops, sorry, misunderstand you12:36
freemangordonjust tell me the command12:36
MohammadAGJust use top for an approximation12:38
freemangordonok, but I think OMP memory usage highly depends on amount of music you have, ain't?12:39
freemangordonisn't sociality a better test?12:39
MohammadAGNo12:40
freemangordonok, give me a minute12:40
MohammadAGOn startup without any lists it uses 8 afaik12:40
freemangordonthe one in extras-devel?12:42
MohammadAGYes12:43
freemangordonok12:43
freemangordonexactly 8.0%12:47
freemangordonMohammadAG ^^^12:47
MohammadAGfreemangordon: In mbs12:47
freemangordonhow to chech that12:48
freemangordoncheck*12:48
MohammadAGAh nvm12:48
freemangordonbtw, now is 7.912:49
Lava_Croftbetween 9 and 10% here12:50
Lava_Croft601 songs, 10 videos, 4 radiostations12:50
Lava_Croftoh, nvm12:50
freemangordoncan you start calc?12:51
Lava_Crofti can12:51
freemangordonok, osso-calculator :D12:51
freemangordonand tell me the memory usage12:51
freemangordonin %12:51
Lava_Croft7.5%12:52
Lava_Croftsteady12:52
Lava_Croft7.4%12:52
Lava_Croft7.3, heh12:53
freemangordon6.7 here, steady12:53
Lava_Croftodd12:53
freemangordonMohammadAG ^^^12:53
Lava_Crofti think this is the first time is started the calculator12:53
Lava_Croftever12:53
freemangordon:D :D :D12:54
Lava_Croftbc > *12:54
MohammadAGCool12:54
Lava_Croftyeah, it sticks to 7.3% now12:54
Lava_Crofti think i misread omp usage tho, lemme recheck12:54
MohammadAGCan someone check a QML app12:54
Lava_Croft9%12:55
freemangordonhelium-mobile-browser?12:55
Lava_Croftfor omp12:55
Lava_Croftso i guess i saw it right12:55
freemangordonor sociality?12:55
Lava_Croftname me a qml app12:55
Lava_Croftsociality?12:55
freemangordonsociality12:55
MohammadAGIt's not QML12:55
freemangordoncutetube-qml12:55
MohammadAGI hate QML12:56
freemangordon:D12:56
freemangordonLava_Croft cutetube-qml12:56
Lava_Croft15%12:56
Lava_Croftfor sociality12:56
Lava_Croftok12:56
Lava_Crofti happen to have that installed too, i wonder why12:56
Lava_Croftoh now i remember12:56
Lava_Croftthats the app that keeps telling me theres stuff in my download queueu12:56
Lava_Crofteven tho its been empty for months12:57
Lava_Croft12.%%12:57
Lava_Crofter12:57
Lava_Croft12.512:57
freemangordoncutetube-qml - steady 10.3%12:57
Lava_Croftsteady12:57
Lava_Croftok ill send my n900 back to nokia and put memory usage unsatisfactory in the note12:58
* Lava_Croft rolls his eyes12:58
MohammadAGSociality depends on the number of notifications you have I guess12:58
Lava_Croftyes12:58
Lava_Croftsociality gets really slow with a lot of notifications12:58
freemangordonMohammadAG, have in mind I don't have thumb-compiled QML, so there will be additional memory gain from it12:58
Lava_Croftat least, thats how it feels12:58
MohammadAGOh really?12:59
Lava_Croftneed anymore readings?12:59
MohammadAGfreemangordon: QML is part of Qt12:59
freemangordonhmm12:59
freemangordonsure?12:59
MohammadAGlibqt4-declarative12:59
MohammadAGYes12:59
freemangordonaah, ok, i was thinking of those qt-mobility things12:59
Lava_Croftomp uses more power(cpu/ram) than the default player, right?12:59
MohammadAGThe memory benefit seems huge12:59
freemangordoni want to check some browser13:00
MohammadAGLava_Croft: Ram yes, CPU, shouldn't13:00
Lava_CroftMohammadAG: i use OMP basically all day, i noticed a little less battery life13:00
MohammadAGRam usage is due to Qt and Gtk13:00
Lava_Croftbut it might be a placebo-ish effect13:00
MohammadAGSomeone picked up omp13:00
Lava_Croftanother one?13:01
MohammadAGAfaik he fixed the high CPU usage13:01
Lava_Crofti know this gizzz or whatever was working on it last i know13:01
freemangordonno, it isn't, OMP is highly unoptimized AIUI13:01
Lava_Croftstill waiting for someone to pick up sociality13:01
MohammadAGIf not I should really patch it13:01
freemangordonit segfaults every now and then, and stucks CPU to 100% from time to time13:01
Lava_Croftits funny how you, kind of half-assed, made basically the only usable FB client for n90013:01
Lava_Croftyeah13:01
Lava_Croftomp tends to crash now and then13:02
MohammadAGLava_Croft: I'm writing it for harmattan so expect a huge update13:02
Lava_Croftit seems to happen the most when selecting a specific album13:02
MohammadAGFor Fremantle13:02
Lava_Croftnice13:02
freemangordonLava_Croft, what was sociality memory usage?13:02
MohammadAGWhat does it need btw?13:02
Lava_Croft15%13:02
MohammadAGFixes that is13:02
Lava_Croftwell, if you have the notifications open13:03
Lava_Croftand then switch to another app13:03
Lava_Croftthe graphics get all funky13:03
MohammadAGHildon desktop issue13:03
Lava_Croftthat doesnt matter for a poor end-user like myself13:03
Lava_Croftno offense, but just saying:)13:04
MohammadAGIt matters who's developing what :p13:04
Lava_Croftmaybe try to do em differently13:04
Lava_Croftthe notifications13:04
Lava_Crofteven tho the current system works nicely13:04
Lava_Crofti think thats the thing that stands out the most as needing a 'fix'13:04
Lava_Croftto be very honest, i dont really come across a lot of brokenness in sociality13:05
Lava_Croftbut, i mostly just read and update statuses13:05
Lava_Crofti dont really use any of the other features besides sporadic Places usage13:05
Lava_Croftit seems you cannot accept friend requests either13:05
Lava_Croftbut i guess thats an API problem13:05
MohammadAGI really need to limit initial status count to 20 and add a more button13:06
Lava_Croftmaybe less than 2013:06
Lava_Croft20 is still quite a list to scroll for that window size13:06
Lava_Croftwhy not make it automagically load more13:06
Lava_Croftafter scrolling 10, it starts to load more13:06
MohammadAG20 is what the harmattan and iOS apps use13:07
Lava_Croftbut the ios app is not portrait13:07
Lava_Croftidk about the harmattan app13:07
freemangordonsociality - steady 17.3%13:07
Lava_Croftyes, i won! \o/13:07
MohammadAGHarmattan sociality13:07
MohammadAGUnfair tests13:07
MohammadAGLogout and test13:07
freemangordonyep13:07
freemangordonok13:08
MohammadAGWait for login page to load13:08
Lava_Croftone thing is great about sociality13:08
Lava_Croftit adheres your themes etc13:08
Lava_Croftthat is a very big and underestimated plus13:08
MohammadAGColours are read from the theme engine13:09
Lava_Crofti figured13:09
MohammadAGThe rest is handled by Qt13:09
Lava_Crofttoo many apps use funky colours schemes etc13:09
MohammadAGQML broke the integration concept13:10
Lava_Croftah13:10
freemangordon15.8% after logout and restart13:10
Lava_Croftand qml is the future, right13:10
MohammadAGI hate the future13:10
Lava_Croftswallow razorblades13:10
MohammadAGHeh13:10
freemangordonLava_Croft, could you please logout and restart sociality13:10
Lava_Croftreboot?13:11
freemangordonno, just exit the application after logout13:11
Lava_Croftah, logout of FB?13:11
freemangordonyeah13:11
Lava_Croftdoes it matter that im hooked on USB charging?13:12
MohammadAGLogging out really needs to clear cache13:12
Lava_Croft3g radio streamning all morning ruins battery13:12
MohammadAGJust a note to self13:12
Lava_Croftoh13:12
Lava_Crofti clicked Back after it logged me out13:12
Lava_Croftthat seemed to close the app13:12
MohammadAGIt does13:12
freemangordonhave no clue, but i have a charger connected, so it should be the same13:12
Lava_Croftoh, yeah MohammadAG13:13
Lava_Croftif possible, make it remember your login name13:13
Lava_Croftso i dont have to type it again right now, after logging out13:13
freemangordondon't13:13
Lava_Croftbut since it seems a website, i guess thats out of your controls13:13
MohammadAGFacebook disallows that13:13
freemangordontell us the memory usage13:13
freemangordonbefore login13:13
MohammadAGWithout logging in13:13
MohammadAG:)13:14
Lava_Croftoh fuck you guys13:14
Lava_Croft:D13:14
Lava_Croftnow i have to logout and type my shit email adress again13:14
Lava_Croft@trioptimum.com gets old you know>_>13:14
freemangordonno, don't type your login13:14
Lava_Crofti wont13:14
freemangordonjust start sociality, wait until it loads FB page and report13:15
Lava_Croft17.713:15
MohammadAGWoo13:15
freemangordonyeah, 15.9 here13:15
MohammadAGYou lose13:15
Lava_Croftis this about the thumb compiling?13:15
freemangordonyep13:15
Lava_Croftyou guys made me read about it some time ago13:15
MohammadAGYeah13:15
Lava_Croftneato, but also seemingly randomly broken stuff13:15
freemangordonnope13:15
Lava_Croftfriends of mine say do not touch with a 10ft pole, but they are all these hardcore coder nerds13:16
MohammadAGOk13:16
Lava_Croftim certainly not13:16
MohammadAGSo can we push the thumb patch in kp ASAP?13:16
freemangordonI am running it on both my devices for about a month and a half13:16
Lava_Croftgood work on the memory saving then13:16
MohammadAGWe'll get to cssu and bin updates later13:16
freemangordonthere is such patch in KP for about an year :D13:16
MohammadAGThumb works on kp?13:17
freemangordonthe missing part was latest gcc and binutils13:17
MohammadAGAh13:17
MohammadAGAsk javispedro about upgrading gcc13:17
DocScrutinizer51hey guys, we won'r roll out owCSSU kernelm13:17
MohammadAGWhy?13:18
freemangordonyeah, there is another thumb-related patch in lates binutils13:18
freemangordonDocScrutinizer: why?13:18
DocScrutinizer51hey guys, we won'r roll our own CSSU kernel!13:18
Lava_CroftKP seems to drain my battery way faster than default13:18
Lava_CroftDocScrutinizer51: good second try, almost there!13:18
freemangordonLava_Croft, no way, there is something broken oon your system13:18
MohammadAGOc config?13:18
Lava_Croft:)13:18
Lava_Croftfreemangordon: i doubt it13:18
Lava_Croftnope13:19
MohammadAGKp should reduce battery usage13:19
Lava_Crofti quit using it about 2 weeks ago, the battery really lasts longer13:19
MohammadAGAt least when charging13:19
Lava_Croftthats what i read too13:19
DocScrutinizer51if sth wrong with KP, goddam FIX IT!!!13:19
Lava_Croftim not sure if its kp :)13:19
MohammadAGWhat?13:19
MohammadAGDocScrutinizer51: Why won't we roll kp in cssu?13:20
*** Effilry has joined #maemo-ssu13:20
DocScrutinizer51we could maybe adopt KP and include it as dependency13:21
freemangordonLava_Croft, with gtalk, facebook chat, skype and IRC connected constantly, plus BT handsfree, plus a lot (2,3,4) hours of browsing, I have about 50-60 % remaining battery at 7 p.m.13:21
MohammadAGThat's what we were talking about DocScrutinizer13:22
MohammadAGWe wont fork any kernel13:22
freemangordonmy sister's phone have about 3-4 days of battery life, but she is not using it hard13:22
DocScrutinizer51we don't need a 3th kernel flavour13:22
DocScrutinizer514th even13:22
freemangordonyeah, that is what we all agree on13:22
freemangordonwill put KP in CSSU, without any forks13:23
DocScrutinizer51bbl13:23
DocScrutinizer51work13:23
*** Effilry is now known as FireFly13:23
MohammadAGDoesnt kernel power include wlan injection patches?13:23
*** FireFly has quit IRC13:23
*** FireFly has joined #maemo-ssu13:23
freemangordonLava_Croft, there is something sucking your battery, but it is not KP, trust me on that13:23
Lava_Crofti have a second, new n900 that i could use for testing13:24
Lava_Croftbut im kind of lazy13:24
freemangordonMohammadAG, I think so, but not sure. Better ask Pali13:24
MohammadAGOh shit13:24
freemangordonLava_Croft, well on all device there is SR turned on13:25
MohammadAGI think I forgot my n90013:25
Lava_Croftfreemangordon: SR>13:25
Lava_Croft?13:25
freemangordonSmartReflex13:25
Lava_Croftah13:25
MohammadAGSmartRe... Nvm13:25
Lava_Croftthe shit that defaults to off13:25
freemangordonyeah, but i've fixed it in KP4913:26
Lava_Crofthmm13:26
Lava_Croftill have to check that out then13:26
Lava_Croftthanks13:26
freemangordonand in KP50 (which is still not realeased) it should be rock stable13:26
*** scoobertron has joined #maemo-ssu13:37
*** fw190 has joined #maemo-ssu13:39
freemangordonLava_Croft, ping13:51
*** lizardo has joined #maemo-ssu13:59
*** M4rtinK has joined #maemo-ssu14:43
*** M4rtinK has quit IRC15:02
MohammadAGfreemangordon, I'm home15:04
*** M4rtinK has joined #maemo-ssu15:06
freemangordonMohammadAG, good, so you are going to push a new update?15:07
MohammadAGyes15:08
MohammadAGI'll need you test some binaries cause I don't have my N90015:08
MohammadAGunless the changes are subtle15:08
MohammadAGby I don't have it I mean it's about 140km away15:08
freemangordonok. I won't be available all the time, but I will help with that15:09
MohammadAGno worries15:09
* MohammadAG reads merlin1991's email15:09
MohammadAGk, starting with hildon-desktop15:10
MohammadAGmerlin1991, ping15:10
freemangordonwell, I am using latest version from git for a month or so, no problems found. But it is thumb-compiled.15:11
MohammadAGfreemangordon, what do you make of http://gitorious.org/community-ssu/hildon-desktop/merge_requests/24 ?15:12
freemangordonnever tested that. And it seems it will need a rebase15:12
freemangordonBut maybe it is arcean to take that, i was focused on lots of different things last couple of months and forget lots of things re hildon-desktop15:14
freemangordonanyway, I don't think portrait lock should make it in this update15:14
freemangordonI am away for an hour or so, bbl15:16
*** m0use has quit IRC15:21
*** mirandir has left #maemo-ssu15:21
*** m0use has joined #maemo-ssu15:36
*** jonwil has quit IRC16:55
*** andre__ has quit IRC17:20
*** scoobertron has quit IRC17:23
*** scoobertron has joined #maemo-ssu17:59
*** BCMM has joined #maemo-ssu18:16
*** BCMM has quit IRC18:18
freemangordonMohammadAG, how is the update going?19:01
*** NIN101 has joined #maemo-ssu19:13
*** javispedro has joined #maemo-ssu19:17
MohammadAGjavispedro, what do you think about upgrading gcc?19:18
MohammadAGwe need it for thumb but apparently scratchbox is a bit f'd up19:18
*** NIN101 has quit IRC19:18
javispedrowhat do you want thumb for?19:18
javispedroavoid it on the n90019:18
MohammadAGit reduces memory footprint19:18
MohammadAGjavispedro, freemangordon fixed it19:18
javispedroin what sense? its performance sucked, i am not aware of any errata19:20
javispedrothat could be fixed I mean19:20
freemangordonthere are erratas19:21
freemangordonand as it is thumb2 there is no performance penalty19:21
freemangordonor at least performance penalty as something about 3-5 %19:22
javispedroI am always talking about thumb219:22
*** Misocainea has joined #maemo-ssu19:22
*** Misocainea has left #maemo-ssu19:22
javispedrothe problem is that the performance on the n900 for thumb2 code was horribly slower than on any other device19:22
javispedroif this is something than could be workarounded in the kernel, I don't know19:22
freemangordonhow did you test it when there is errata to be fixed?19:22
javispedroit worked quite well, what is the errata?19:23
javispedromy test was the dosbox JIT thumb backend, you can see the days I was testing this on the main chan logs19:24
freemangordonerrata 43097319:24
javispedroon most other devices it is the preferred target, but on n900 it was an order of magnitude slower than plain arm target.19:24
freemangordonthe other errata is 65741719:25
*** NIN101 has joined #maemo-ssu19:26
freemangordonwell, according to TI docs, there is almost no performance penalty for using thumb2 code instead of pure arm, have no idea why it was slow for you19:26
freemangordonI am usint thumb-compiled Qt,gtk and all of the hildon stuff on my primary device for more than a month, no visible slowdowns, actually it is faster because of the lighter memory requirements19:27
freemangordonjavispedro ^^^19:28
*** fw190 has quit IRC19:42
MohammadAGjavispedro, ?19:42
javispedroI am looking through DosBox code19:42
MohammadAGfreemangordon, scratchbox-libs: /scratchbox/compilers/bin/gcc19:43
javispedrono hint of T2, either way doesn't explain why it worked despite 43097319:43
javispedro(I mean, lots of T1)19:43
freemangordonwhich kernel are you using?19:43
freemangordonbecause KP has errata fixed for an year or so19:44
javispedrostock19:44
freemangordonwell the errata is such that it may run for hours and even days without a problem, and then out of the blue SIGILL19:46
javispedrohardly on a JIT..19:46
freemangordonself-modifying code? or Jazelle?19:48
javispedrolots of self modifying code on a small address space19:48
freemangordondo you have some cache/TLB/BTB or whatever flush?19:49
freemangordonas errata 430973 is for real, there should be something in your code that prevents it from triggering.19:50
javispedrofreemangordon: sys_cacheflush counts?19:53
freemangordoncould be :D19:54
freemangordonhave to look at the cernel code, bu most probably yes19:54
javispedrowell there are lots of it for obvious reasons19:54
freemangordonkernel*19:54
merlin1991freemangordon: re gcc debs19:55
merlin1991those are debs too19:55
freemangordonwell, you don't need that on cortex-a8 :p. no need to flush caches for self-modifiyng AIUI19:55
merlin1991though installed to the host system19:55
freemangordonwhich debs are those?19:55
MohammadAGfreemangordon, I told you above19:55
MohammadAGscratchbox-libs19:55
freemangordonahaa, ok. source?19:56
MohammadAGapt-get source I guess? :P19:56
MohammadAGjavispedro, should know19:56
merlin1991MohammadAG:  ping19:56
MohammadAGpong19:56
merlin1991err rather pong :D19:56
merlin1991you pinged me :)19:56
MohammadAGping19:56
javispedroMohammadAG: what do you want to do, upgrade to a new codesourcery release, or something more exotic?19:56
merlin1991apart from that, please push the modest-l10n to gitorious19:57
freemangordongcc 4.6.219:57
*** scoobertron has quit IRC19:57
MohammadAGjavispedro, thumb and gcc are on hold19:57
merlin1991I'd like to get the stable update with modest out :)19:57
MohammadAGbut we'd like to have gcc 4.6.219:57
MohammadAGwasn't modest fixed?19:57
freemangordonjavispedro: and latests binutils19:57
freemangordonlatest*19:58
MohammadAGbinutils is easy19:58
merlin1991MohammadAG: modest-l10n on gitorious still is cssu2 whilst we have cssu3 in the repo19:58
freemangordonMohammadAG: easy as in?19:58
merlin1991btw MohammadAG did you want anything in particular when you pinged me?19:59
MohammadAGit's not as complex as gcc afaik19:59
MohammadAGscratchbox-wise19:59
MohammadAGmerlin1991, yes, I wanted to bitch about osso-calculator-engine19:59
MohammadAGbut I think I'll just push it anyway19:59
freemangordonMohammadAG: ever done it(binutils)?19:59
*** trbs has joined #maemo-ssu19:59
MohammadAGno20:00
MohammadAGbut there are no complex symlinks in scratchbox afaik20:00
MohammadAGgcc is linked from the host and wrapped and stuff20:00
freemangordonsame for ld and other stuff AFAIK20:01
merlin1991MohammadAG: push modest-l10n to gitorious NAO!20:01
merlin1991:D20:01
freemangordonok, lets leave gcc and thumb on a background and make merlin1991 happy :D20:02
MohammadAGhe's already happy20:03
merlin1991oh I am, yay :)20:03
merlin1991MohammadAG: pm20:03
javispedroMohammadAG: binutils and gcc is the same for sbox20:04
javispedroboth are in the same "compiler" package20:04
freemangordongood :D. MohammadAG, if you or javispedro know how to raise gcc and binutils version I will be more than happy20:04
freemangordonI mean using "correct" way20:05
javispedroI would say that the easiest way would be to just use the harmattan version, but it does not seem to be 4.6...20:05
freemangordonit is not20:05
freemangordonand it does not have errata fixed20:05
javispedrothe ld one is quite probably a few lines of patch, could be backported20:06
freemangordonand libstdc++.so is not build against correct headers and libs20:06
freemangordonjavispedro, nope, it is not http://www.cygwin.com/ml/binutils/2009-05/msg00354.html20:07
merlin1991kubuntu stalled on boot, yay20:09
*** m0use has quit IRC20:26
javispedroseems that the patches to cs2009 in sbox are nowhere to be seen, and I remember they once ported cs2010 but it's also nowhere either20:28
javispedrothe idea is to simply follow the guide in the scratchbox-toolchain-extras package aka http://www.scratchbox.org/download/files/sbox-releases/hathor/src/scratchbox-toolchain-extras-1.0.25.tar.gz20:31
javispedroit was a lot of pain years ago (basically required sarge to build), and it is probably still some pain these days, but dunno how much20:32
freemangordonjavispedro, that won't work AIUI, we want ABI compatibility20:32
freemangordoni.e. keeping libc version 2.520:33
javispedrois it even build with thumb-interwork?20:33
freemangordonwhich one?20:33
javispedrothe current one, and binaries20:33
freemangordonI think so20:33
javispedroI am quite probably biased against thumb, but I think this will be a lot of a pain for a very small performance advantage20:34
freemangordonsmall?!?20:34
freemangordonThumb-compiled QT is about 14 MBs smaller than arm compiled20:35
javispedroit does not mean anything20:35
freemangordonyes, it means20:35
freemangordonyou have to put those in RAM sometimes :D20:35
javispedro"sometimes" being the keyword here20:36
freemangordonless size means less swapping20:36
javispedroit would be like saying -Os firefox is faster than -O2 firefox because it is 4 times smaller20:36
freemangordonmeans less lag20:36
freemangordonfaster as in?20:36
freemangordonbecause bottleneck on n900 is the amount of the RAM, not the CPU speed20:36
javispedroaka flash speed I guess20:37
freemangordonand now Qt is in /opt20:37
freemangordonwhich leads to additional penalty20:38
freemangordonthumb compiled libs and bins will lead to more free space on NAND, so we can put Qt there20:38
freemangordongot my points?20:39
javispedrowell, that is interesting indeed...20:39
freemangordonAnd I am able to use helium-browser for the first time without lag since I am using gcc-4.6.2-thumb-compiled Qt, that means something ;). Same for qmltube, sociality, all that qt stuff which is almost perfectly working, if you close your eyes for the lag because of constant swaping and reading pages from /opt/lib20:42
freemangordoneven Qt libs are still in opt. Hmm, maybe I should move them to NAND20:44
javispedrowell, not wanting to argue, if it does really mean the change between fitting something in NAND or not...20:59
javispedroI'd also suggest trying profile guided reordering or some other stuff, or even manually building part of Qt with -Os to try and squeeze it in nand21:00
freemangordoninstead of thumb?21:00
javispedroso, you can keep the same libc version if you want21:00
javispedroinstead of thumb, yes21:00
javispedroif you change the toolchain a lot of stuff is going to break21:00
freemangordonwell, I kept the same glibc version21:00
freemangordonnothing is broken so far21:01
freemangordonsorry, libc, not glibc21:01
javispedroif nothing is broken, why are you doing this? =)21:01
freemangordonwell, still have hope that something will explode :D21:02
javispedrooh, then you should indeed follow the guide =)21:02
freemangordonRTFM? :D :D :D21:03
javispedrothere's a .conf file you can change, and tell it where to pull the source files for all the involved components21:03
freemangordonHave you done it?21:04
javispedroi.e. headers, binutils, libc, gcc, and some other stuff (autotools iirc)21:04
javispedroI did it once for the n8x0, back when a gcc bug caused some failures building the snes emulator21:04
javispedrochanging the libgcc caused most c++ applications to crash on exit21:05
javispedroI eventually got tired of it :(21:05
freemangordonAnd that is the result of following TFM :)21:05
javispedrowell you will be for sure forced to upgrade libgcc21:06
javispedroas 4.4 to 4.6 is a rather large change..21:06
freemangordon4.2.3 ;)21:06
javispedrooh, 4.4 is harm =)21:06
freemangordonyep21:06
MohammadAGdid you try thumb compiled gtk?21:07
javispedroin n8x0 days I had to use a Debian Sarge VM, otherwise scratchbox-core had a few thousand build errors21:07
javispedroI think that this has been fixed in hathor21:08
freemangordonMohammadAG, who did you ask?21:08
javispedrothe problem is that in the source of the latest sbox I do not see the patches for cs2009, dunno if its means there are none, or if they have been "forgotten"21:08
*** fw190 has joined #maemo-ssu21:08
freemangordonjavispedro, the only library that needs change is libstdc++.so.6.0 from .16 to .19, but that brings no problem, as we keep libc version and it is compiled against it AIUI21:10
MohammadAGfreemangordon, you21:10
javispedrolibstdc++? or libsupc++?21:11
javispedrolibstdc++ comes with glibc version21:11
freemangordonyep, as I already told you the thumb-compiled libs and bins I am using are: Qt, gtk, hildon-desktop,hildon-home,hildon-status-menu21:11
freemangordonMohammadAG ^^^21:11
freemangordonjavispedro, it is compiled when you build gcc21:11
freemangordonand yes, libstdc++21:12
freemangordonjavispedro, versions error, original is libstdc++.so.6.0.9, replacement libstdc++.so.6.0.1621:15
javispedroah well21:16
freemangordonAnd by statically linking gcc and binutils I get rid of unresolved imports (in scratchbox)21:16
javispedrocross figners..21:16
javispedro!21:17
javispedrobtw21:17
fw190ja zawsze daje to pierwsze zdanie21:17
javispedroyou are probably also getting quite of an improvement by building qt with the new toolchain21:17
freemangordonyep21:18
freemangordonthats for sure21:18
javispedrowell, try it, it cant hurt21:18
freemangordonwhich one?21:18
javispedroI have little time today but I'll set this channel to autojoin21:18
freemangordonjavispedro, which one to try?21:19
javispedrofreemangordon: personally I would try something less recent such as cs201021:19
javispedrobut I guess you want 4.6, so..21:19
javispedroyou can find the config file nokia used in /scratchbox/compilers/*/config21:20
freemangordonlatest cortex-a8 fixes are from june/august 2011 or so, so cs2010 just wont do the job21:20
javispedroerr .. */build-config21:24
freemangordonwell, I think that using what I have already done will be much easier than RTFM21:25
javispedronot saying otherwise, or even moving to sb221:25
*** scoobertron has joined #maemo-ssu21:25
freemangordonI have fully working gcc 4.6.2 and binutils 2.22, will just share that21:26
freemangordonI mean fremantle armel target in sb21:27
javispedroyou replaced the binaries and it worked?21:28
javispedroif so, there are no patches, so it should also be easy to build it the sb way21:28
freemangordonnot so easy, but in general yes21:28
javispedroif its just rpath things the sbox makefiles take care of that21:29
freemangordondon't see the point21:29
javispedroautobuilder21:29
javispedrowell, doesn't matter I guess21:29
freemangordonwe are talking about Qt, gtk and stuff that anyway will be distributed through CSSU21:29
freemangordonyeah21:30
*** nox- has joined #maemo-ssu21:41
*** nox- has joined #maemo-ssu21:42
*** fw190 has left #maemo-ssu21:53
merlin1991fsckd modest22:12
merlin1991anyway, Stable release #2 is finally out22:16
merlin1991We now haz modest and upgraded certman in cssu-stable22:16
Sicelo\0/22:19
freemangordongood, good22:23
freemangordonmerlin1991, did you thumb compile it ? :p22:23
*** lizardo has quit IRC22:29
*** javispedro has quit IRC22:44
merlin1991freemangordon: no? :D22:49
*** ekze_note has joined #maemo-ssu23:03
merlin1991MohammadAG: ping me when the update is out23:07
*** trbs has quit IRC23:12
*** scoobertron has quit IRC23:21
*** javispedro has joined #maemo-ssu23:43

Generated by irclog2html.py 2.15.1 by Marius Gedminas - find it at mg.pov.lt!