**** BEGIN LOGGING AT Wed May 05 02:59:56 2010 May 05 06:20:58 I meet belowing Error while build meego iso from trunk repo: May 05 06:21:02 Error: failed to create image : Failed to build transaction : system-config-date-1.9.36-20.1.noarch requires gnome-python2-canvas May 05 06:22:17 I checked repo.meego.com, there is no gnome-python2-canvas package. May 05 08:27:59 morning slaine May 05 08:32:49 morning Stskeeps May 05 08:32:58 refreshed I hope May 05 08:34:21 yeah, a bit better May 05 08:39:13 Oh, wednesday already May 05 08:39:19 TSG meeting tonight May 05 08:52:44 WAHa_06x36: jkirdner is also looking at beagle/meego i think - you two should coordinate May 05 09:02:35 Did you get any further with your watchdog problem WAHa_06x36 May 05 09:11:39 o/ May 05 09:11:45 slaine: pin May 05 09:11:46 g May 05 09:11:57 ack May 05 09:12:28 we're making progress on the RWG front as per emails May 05 09:12:43 I'm building a system that will be a community OBS May 05 09:12:51 awesome May 05 09:12:54 :) May 05 09:12:59 is this all happening on the community list ? May 05 09:13:21 it will be on -dev AFAI May 05 09:13:23 K May 05 09:13:37 grand, that's the only one I'm signed up to at the moment May 05 09:13:57 although quim seems to have re-defined -dev.... which was a surprise May 05 09:14:18 indeed, he seems to have turned into a moderator of late May 05 09:14:30 remind me ... were we discussing rpm 'training' ? May 05 09:14:32 looks like my registration problem only occurred with yahoo mail May 05 09:14:34 ages ago? May 05 09:14:36 redefined -dev? May 05 09:14:47 Stskeeps: as "platform dev only" May 05 09:14:59 he redirected app-dev to -sdk/forum May 05 09:15:05 as in not a developers list, but platform dev list May 05 09:15:05 which kinda blew me away May 05 09:15:18 well, that might make sense to some degree May 05 09:15:21 But isn't that the original intent of -dev? May 05 09:15:32 I thought it was a list for developers May 05 09:15:37 not just platform May 05 09:15:39 yeah, except -sdk was about developing an sdk I thought May 05 09:15:47 and what defines 'platform' May 05 09:15:49 I've the same understanding May 05 09:16:08 plus it ain't excatly bulging at the seams :) May 05 09:16:16 lbt, re the rpm training, I think it was th0br0 that was mentioning getting some Fedora heads together to offer some training online May 05 09:16:27 if things pick up more, -dev would be a horrid place to subscribe to May 05 09:16:46 Stskeeps: yes. but that's the time to start splitting it May 05 09:16:48 Wait 'til this time next month May 05 09:16:57 lbt: old habits die hard :/ May 05 09:17:23 ie, split is better to do early than late May 05 09:17:25 it's a ml, not a forum... we have tools you know ;) May 05 09:17:27 we should all have meego 1.0 in our hands (or on our laps, depending on hardware) and the list will explode May 05 09:17:44 Stskeeps: I disagree - a split increases the chances of missing something May 05 09:17:54 well, each his own opinion :) May 05 09:17:56 right now we need to maximise awareness of activity May 05 09:18:00 i don't like the lack of Reply-to: for instance May 05 09:18:00 :P May 05 09:18:10 * lbt reply-alls May 05 09:18:17 and i also like splits because it aids metrics May 05 09:18:21 * lbt likes lkml May 05 09:18:32 if it works for them... May 05 09:18:33 which will come back to us again as a valuable tool May 05 09:18:34 :P May 05 09:19:47 when will the garage be opened ? May 05 09:19:58 * lbt pings th0br0 for rpm training support :) May 05 09:20:07 TheBootroo: I'm working on it May 05 09:20:09 TheBootroo: you keep asking the same questions May 05 09:20:43 lbt: you re charged of garage admin ? May 05 09:20:59 slaine: :-) May 05 09:21:30 I'm not "in charge" of anything... but X-Fade and I are working on getting something up and running May 05 09:21:40 ok May 05 09:21:52 lbt, who supplied the hardware in the end ? May 05 09:21:55 how is that possible without access to server May 05 09:22:16 pupnik: snap, haha May 05 09:22:25 we will be looking for people to alpha/beta-test the processes. They'll need a damned good justification to be part of the alpha-test group though May 05 09:22:35 I rock May 05 09:22:38 does that count ? May 05 09:22:40 slaine: we have a small server from maemo.org May 05 09:22:53 slaine: submit and see... you only get one submission :) May 05 09:23:03 hehehe May 05 09:23:14 i'm a roxxxin hard Qt dev, can I test ? ;-) May 05 09:23:53 sigh. that was 2 potential testers down at the first hurdle.... "What do you do to make lbt's life easier?" May 05 09:24:05 coffee ? May 05 09:24:10 :D May 05 09:24:49 lol, So it's favors your after then lbt. I didn't know it was gonna be that kinda party May 05 09:25:20 XD May 05 09:25:42 I can't remember exactly who was involved but there was a thread on the mailing list where someone said they had an n900 to help, and someone replied "I think Nokia has enough n900s" (or something like that). I lol' May 05 09:25:46 lol'ed* May 05 09:26:06 why ? May 05 09:26:12 heh... I will put out an email. And yes, I'm looking for people who can bring something significant to the party: process knowledge, experience with relevant systems/tools etc etc May 05 09:26:29 TheBootroo, can't you see the funny in that? :P May 05 09:26:36 because the poster apparently thought having a N900 was sufficient to assist with meego May 05 09:26:40 lbt, I would actually be interested in repackaging my moblin repo's rpms for x86 meego, so if there's space and that's something you'd like to test, give me a shout when you're ready. May 05 09:26:44 not really .... that's true and false May 05 09:27:15 Termana: it was probably TheBootroo that sent that email May 05 09:27:18 ;) May 05 09:27:21 no May 05 09:27:26 :P meany May 05 09:27:28 i'm not on the mailing list May 05 09:27:39 probably TheBootroo's dad May 05 09:27:43 slaine: OK. I'll forget - but hopefully we'll get a thread going and there'll be a point to jump in. May 05 09:27:43 i have an N900 and i'm keeping it May 05 09:27:56 I'll keep an eye out so May 05 09:28:10 pupnik: my father doesn't have a N900, even if he really wanted to May 05 09:28:16 I'm somewhat familiar with .spec files too, so I may be able to offer some help there May 05 09:30:00 do you think that there are some Nokia devs who have already a Meego 1.0 preview release running with preliminary Ui on their N900 ? May 05 09:30:44 TheBootroo: of course, it's already be discussed by our benevolent dictators May 05 09:30:53 s/be /been / May 05 09:30:53 slaine meant: TheBootroo: of course, it's already been discussed by our benevolent dictators May 05 09:31:20 so ? May 05 09:31:46 it may happen a big leak like iPhone 4G one ? May 05 09:32:08 TheBootroo: no, everyone is running xterm with emacs inside ;) May 05 09:32:20 they're not testing the UX ? May 05 09:32:27 emacs is a UX ;) May 05 09:32:30 or have they all just an Iphone ? May 05 09:32:33 :D May 05 09:32:51 TheBootroo: http://carrypad.com/2010/04/13/meego-at-idf-netbook-and-handheld-eye-candy-chrome-fennec-and-lots-of-developer-details/ May 05 09:33:19 Stskeeps: Emacs is not the handheld UX May 05 09:33:41 i would like to wager that emacs would be an excellent handheld u May 05 09:33:41 x May 05 09:33:41 :P May 05 09:34:27 slaine: i know these 3 screenshots but are they the only one on the whole web ? May 05 09:35:10 yes May 05 09:35:20 :-( May 05 09:35:43 and they weren't properly "released" by the meego team, so it might not even be accurate May 05 09:36:16 i hate 'artistic blur' May 05 09:36:25 Stskeeps: I'd rather use vim May 05 09:36:42 slaine: shame on you ^^^ May 05 09:37:00 i would rather use QtCreator 2.0 May 05 09:37:05 ;-) May 05 10:51:58 hrm, didn't kwork May 05 12:10:48 forum.meego.com down? May 05 12:11:06 yeah, reggies upgrading them both May 05 12:11:28 both? May 05 12:11:43 both of one? May 05 12:11:49 meego.com and maemo.org one May 05 12:17:19 * Stskeeps yawns May 05 12:18:55 ah May 05 13:48:11 heya people May 05 13:51:51 moo May 05 14:12:19 if the released browser is going to be the loosely integrated chromium May 05 14:12:21 like i am running here May 05 14:12:24 i am going to throw up May 05 14:12:39 * TSCHAKeee wonders why the hell they got rid of the browser as a tab May 05 14:13:52 jedix: ask in here jedix May 05 14:14:14 does anyone know where the armv7 repo is? May 05 14:14:46 you can lie to it and use the armv5el rpms in the :armv7el though May 05 14:15:18 then, create a project and build armv7el using armv5el to bootstrap May 05 14:15:48 okay May 05 14:15:56 Stskeeps may know about armv7el May 05 14:16:10 so I need to grab an entire copy of armv5 May 05 14:16:14 so um, where's the next version of Mer? May 05 14:16:21 * TSCHAKeee chuckles and dodges the nerf darts May 05 14:16:24 ;) May 05 14:16:45 TSCHAKeee: OT, take it to #mer May 05 14:16:51 May 05 14:16:57 AIIIEEE! *DODGE-BAZOOKA-FROM-stskeeps* just kidding! just kiddiinggg!! May 05 14:17:00 :) May 05 14:17:03 * w00t_ hands lbt a bigger stick May 05 14:17:25 is there an rsync host or do I have to use wget? May 05 14:17:48 rsync? what centrury do you think this is? May 05 14:17:58 we're talking Intel here.... May 05 14:18:23 * lbt hands jedix some 10baseT May 05 14:18:33 well, what's the best way to grab the entire package set? May 05 14:18:33 what's wrong with rsync now? May 05 14:18:55 wget --mirror --no-parent --some other crap May 05 14:19:06 * w00t_ senses that someone's sarcasm detector is failing May 05 14:19:15 oh.. god... May 05 14:19:25 jedix: that link I gave you.... look in there May 05 14:21:22 okay, it's going for armv5 May 05 14:21:41 so I should follow that entire site? May 05 14:23:02 lbt: you mentioned a static busybox? May 05 14:24:48 jedix: not the entire site... that's a complex mishmash for fremantle that's WIP at the moment May 05 14:25:11 It uses the rpm cross build process which I think needs static busybox. May 05 14:25:41 NB if you get this working can you document what you do in intimate detail please? May 05 14:26:19 I will try, for sure May 05 14:26:28 but to get to this point was hell May 05 14:26:30 http://wiki.maemo.org/OpenSuse_Build_Service/Installation May 05 14:26:44 ie: vmware->qemu->vbox to just get the image going May 05 14:27:02 then vbox wasn't working right for the obs May 05 14:27:10 the web stuff was not reachable May 05 14:27:11 mmm xen :) May 05 14:27:18 man May 05 14:27:27 and i thought the linuxmce build system was insane May 05 14:27:29 ;) May 05 14:27:55 * jedix -> coffee May 05 14:33:08 TSCHAKeee: OBS is wonderful May 05 14:33:16 * lbt is serious :) May 05 14:33:23 * lbt goes for coffee too May 05 14:34:13 ;) May 05 14:36:56 oh sweet coffee May 05 14:37:00 OBS is wonderful? May 05 14:37:01 jedix, you should be able to write a script to pull down all the packages May 05 14:37:13 slaine: wget can do it.. May 05 14:37:16 I did one last year to get all the moblin stuff May 05 14:37:20 not sure what you mean May 05 14:37:43 I probably just didn't know how to use wget to get the directory I wanted May 05 14:37:44 I have the src.rpm's May 05 14:37:47 grand May 05 14:38:01 just grabbing the amrv5 stuff May 05 14:38:07 so I can bootstrap the puppy May 05 14:38:10 I pulled down all the .src.rpm's to rebuild them May 05 14:38:25 epic fail May 05 14:38:34 hopefully they've sorted their packaging this time around May 05 14:38:37 oh god May 05 14:38:54 don't tell me I'm setup for an epic fail May 05 14:41:55 jedix: here's an example May 05 14:41:56 http://lists.moblin.org/pipermail/dev/2009-August/005889.html May 05 14:42:05 jedix: i'd wait personally May 05 14:44:57 what I don't get is this maemo.org/OBS page says it's uisng a repositories May 05 14:45:00 I can't use them May 05 14:45:17 Stskeeps: it's either wait or try myself May 05 14:48:49 lbt: so are the repositories important in this setup? May 05 14:51:37 jedix: alot of it is probably still behind the iron curtain May 05 14:52:01 slaine: yeah, that's why I'm wondering if it's necessary May 05 14:52:38 i'd imagine it is as a lot of necessary stuff is still behind said curtain May 05 14:52:43 looking at the howto, they're using debian/ubuntu to bootstrap.. which doesn't really make sense but maybe it doesn't matter, maybe it's just for the inital build May 05 14:54:00 I see the Moorsetown chip is finally out, now called the Atom Z600 May 05 15:03:03 interesting stuff May 05 15:03:04 http://www.semiaccurate.com/2010/05/05/intel-announces-atom-z600-series-phones-tablets/ May 05 15:03:47 fascinating May 05 15:05:37 and what's that they show on the screen? May 05 15:05:40 android May 05 15:05:42 of course May 05 15:05:44 :( May 05 15:05:46 (#@(#@ May 05 15:06:45 Stock phone photo though May 05 15:11:10 it's an Aava Mobile phone May 05 15:13:59 how do I order one? May 05 15:15:03 jedix: that webpage was about building fremantle on obs, not meego. May 05 15:15:33 it's not a howto... it's a "oh, right. Maybe I could try something like that" guide May 05 15:16:26 lbt: yeah, just wondering if the repos are necessary May 05 15:16:33 becuase they seemed to be when I was playing with it May 05 15:17:14 what do you mean by repo. May 05 15:21:24 repository May 05 15:24:29 what repository May 05 15:24:42 meego arm repo's or maemo ones May 05 15:54:22 Oh god NO May 05 15:55:22 Intel GMA 600 Graphics – Integrated power-optimized 2-D/3-D graphics with up to 400MHz graphics core frequency, support for OpenGL ES2.0, Open GL 2.1, and OpenVG 1.1, and hardware-accelerated7 HD video7 decode (MPEG4 part 2, H.264, WMV & VC1) and encode May 05 15:55:22 (MPEG4 part2, H.264). Supports internal display up to 1366 x 768 LVDS or 1024 x 600 MIPI. May 05 15:55:35 That's what's on the Atom Z600 May 05 15:55:39 Poulsbo++ May 05 16:00:31 slaine: I was under the impression that the major problem with poulsbo/GMA500 was that Intel didn't acquire full rights to the system but only licensed the core; hence they never had specs to give out, and even the drivers were written by someone else May 05 16:08:00 Bostik: precisely. May 05 16:08:09 lovely hardware, little support May 05 16:09:41 one can hope they had learned the lesson May 05 16:09:59 but this is real life, so not much chance of that happening May 05 16:14:51 It's possible that the IEGD team have gotten upto speed with the driver code May 05 16:15:18 it was a long time ago that Tungsten Graphics wrote the original driver for them. May 05 16:20:33 * CosmoHill offers people cheese toasties May 05 16:21:08 Awe dude, that's just wrong May 05 16:21:59 no, bottle guy is wrong May 05 16:23:04 CosmoHill: May 05 16:23:05 Q: "Why don't doctors give cervical exams to women over 65 ? May 05 16:23:05 A: "Have you ever tried to open a cheese toastie ?" May 05 16:23:19 ewww wtf May 05 16:23:28 ok, people May 05 16:23:39 * slaine runs for cover May 05 16:23:55 Just letting you know i'm paying attention :) May 05 16:24:02 i was talking about the food May 05 16:24:28 CosmoHill - you're in the clear, but Slaine is lucky he ran for cover :) May 05 16:24:34 :) May 05 16:24:35 almost are rumbled as this guy, http://www.herald.ie/multimedia/archive/00563/2804_crispad_PA_563283t.jpg May 05 16:24:45 s/ are / as / May 05 16:24:46 slaine meant: almost as rumbled as this guy, http://www.herald.ie/multimedia/archive/00563/2804_crispad_PA_563283t.jpg May 05 16:26:35 I feel pretty tired May 05 16:26:51 you look pretty tired May 05 16:27:06 :| May 05 16:27:35 CosmoHill: yeah, me too. And I've another late night tonight May 05 16:27:53 what you working on? May 05 16:29:05 customized linux distro May 05 16:29:16 based on Fedora at the moment May 05 16:29:31 i have a small custom one for my server May 05 16:29:55 I need one for our hospital units and set top boxes May 05 16:30:22 we've got a hand rolled one that's mostly ok, I'm working on the tools to create, package, distribute it May 05 16:30:30 from scratch May 05 16:34:56 heya everyone May 05 16:35:07 hey th0br0 May 05 16:35:22 let's see what tonight's meeting will bring May 05 16:35:25 I'm watching Ultimate Police Chases: Nowwhere to run May 05 16:35:30 it's horrible editing May 05 16:35:49 ^^ May 05 16:36:06 even tho they've redone it with an english narrator May 05 16:36:23 what is / was the original language? May 05 16:36:28 American May 05 16:36:46 it's just toned down a bit May 05 16:36:49 and less annoying May 05 16:38:02 k May 05 16:38:20 still the editing is all over the place May 05 16:38:39 the story is linear but the video clips aren't May 05 16:40:39 th0br0: looking forward to the "* Upcoming Release" agenda item :) May 05 16:45:40 slaine: I have two assignments due in friday May 05 16:46:01 stop watching car crash tv then May 05 16:46:03 ;) May 05 16:46:23 my biggest problem now is the report May 05 16:46:35 the non-coding part ? May 05 16:46:39 yes May 05 16:46:54 yeah, no motivation to those parts May 05 16:47:37 i think I'll watch a movie May 05 16:47:42 haha May 05 16:47:46 then work on it after dinner May 05 16:47:50 my game is almost finished May 05 16:48:06 i have a strong urge to watch Roborts May 05 16:51:18 what I'd like is a good computer monitor that can also handle 16:9 stuff from the DVD or sky box May 05 16:52:19 I do like that the first thing you see is "Do turn your mobile phones off and reframe from talking" May 05 16:52:26 the only DVD I've seen that has it May 05 17:04:42 slaine: that's what I'm saying May 05 17:04:46 there are no meego repos May 05 17:04:54 and there's no choice in obs to use maemo May 05 17:04:58 so wth do I do? May 05 17:05:26 you wait, like we said earlier May 05 17:05:47 it's all still behind the iron curtain May 05 17:05:57 heh May 05 17:06:17 but the srpms are out May 05 17:06:28 I can't wait May 05 17:06:34 just for command line, jedix May 05 17:06:47 no ui yet May 05 17:07:01 there's lots of inconsistencies. there's kickstart files that are public that reference repo urls that are closed etc. May 05 17:07:17 so the equilivant n900 image x support? May 05 17:07:17 it's a bit of a mess atm. But should clear up over the coming weeks May 05 17:08:15 yes, i think everyone's pretty busy knitting together the codebase May 05 17:16:53 time to commute May 05 17:16:58 catch you all later May 05 17:17:01 bye May 05 17:20:49 so, am i correct in thinking that the meeting is in a little more than 30 minutes? May 05 17:21:10 no, you're not May 05 17:21:20 it's 99 minutes away May 05 17:21:38 19:00 UTC May 05 17:21:39 hey thiago_home May 05 17:21:45 hey CosmoHill May 05 17:22:39 I'm happy with the code for my assignment :) May 05 17:23:34 good May 05 17:23:44 but the important question is whether your professor is happy with it too :-P May 05 17:23:50 (or the assistant) May 05 17:24:02 i think he's alright with it :0 May 05 17:24:07 it's due in friday May 05 17:24:15 i need to do a 1500 word report on it May 05 17:24:43 ah, okay - I'm an hour off May 05 17:24:45 thanks May 05 17:25:04 i'm not correcting for daylight savings May 05 17:25:31 hey Cosmo, what are you writing? May 05 17:25:43 a text based rpg game May 05 17:26:05 that sounds like fun May 05 17:26:19 depends how my work is going May 05 17:26:30 like a mush? May 05 17:26:36 @ <-- the icon for a spiraling vortex of pop culture hell May 05 17:26:38 mush? May 05 17:27:06 yeah, it was a text-based rpg May 05 17:27:29 well more like a kind of software you could use to play a rpg May 05 17:27:48 there were mush'es, moo's, one called tinyMush... May 05 17:28:32 this was right before mosaic released May 05 17:28:43 probably no one uses them at all now May 05 17:28:51 * TSCHAKeee remembers May 05 17:29:04 <-- old grey haired bastard May 05 17:29:36 I'm not old or grey haired May 05 18:04:43 Quick reminder that the TSG meeting starts in 55 minutes (19:00 UTC). May 05 18:05:08 ok thx May 05 18:05:12 Agenda / logistics here: http://wiki.meego.com/Technical_Steering_Group_meetings May 05 18:05:13 Cheers, DawnFoster :) May 05 18:05:24 cheers th0br0 May 05 18:05:57 oh, i meant it as "thanks", not bye ;) May 05 18:06:42 btw, just *where* does the community come into play in that meego structure graph? May 05 18:10:54 th0br0 - people can participate in almost any area. May 05 18:11:37 right now, most of the people running those functions will be intel / nokia until people start contributing and rise up in the ranks based on merit May 05 18:12:29 For example, in the community office, we have people outside of Intel / Nokia in key positions - like Reggie who runs the forum May 05 18:12:50 In the program office, localization has tons of contributions from people doing community translations. May 05 18:13:02 anyone can submit bugs, contribute patches, etc. May 05 18:15:33 what about community-run initiatives? May 05 18:15:39 I don't really see any place for that in there. May 05 18:15:56 .seen lbt May 05 18:16:01 .any lbt May 05 18:16:02 mh. May 05 18:16:23 morning Dawn May 05 18:18:11 Community run initiatives can happen anywhere May 05 18:18:48 wherever they make the most sense. May 05 18:19:33 some might fall under the community office, while others, like localization, fall into other areas May 05 18:19:52 Dawn, I'm just getting ramped up - do the TSG's happen every week at the same time? May 05 18:20:51 DawnFoster: your topic got cut off in the meeting room May 05 18:23:01 the TSGs happen every week at the same time. May 05 18:23:24 if we cancel them, you'll see a note at the top of the TSG wiki page: http://wiki.meego.com/Technical_Steering_Group_meetings May 05 18:23:38 stskeeps - thanks - should be fixed now May 05 18:23:59 okay, thanks May 05 18:28:06 DawnFoster: when things step up something like ubuntu's 'fridge' meeting calendar could be nice May 05 18:28:24 ie, show when the meeting room is booked May 05 18:31:58 Stskeeps, what about mfe ? :D May 05 18:32:00 * VDVsx hides May 05 18:32:40 * thebootroo hurts VDVsx May 05 18:34:09 stskeeps: yeah, we'll need something like that May 05 18:49:36 /NICK mskarpne May 05 18:50:08 oops :) May 05 18:50:14 So.. I have a repo made, but I don't know how to setup a Cross compile for OBS May 05 18:50:39 hi all May 05 18:51:03 hello May 05 18:52:24 hello again jedix May 05 19:25:01 slaine: I propose we update the protocol to include a "raise hand for question" May 05 19:25:34 I thought the whole point of putting QUESTION in front of a comment was that it was noted and addressed in due course May 05 19:26:22 slaine the problem is that questions get lost. We've had times where we're like 10 questions behind - it's too hard to manage May 05 19:26:40 So how do I ask a question then ? May 05 19:26:56 wait until this question is wrapping up May 05 19:27:07 (it had gone quiet when I started typing it, so I thought everyone had said their piece) May 05 19:27:22 no worries - we'll wrap this one up May 05 19:27:26 slaine: that's why we need a raise hands :) May 05 19:28:36 we're still working out the bugs in the process :) May 05 19:30:53 slaine: I'll be looking for beta testers RSN for the meego-community OBS that we're building May 05 19:31:49 slaine: and make sure you cc -dev on the mail to Imad :) ... May 05 19:31:57 slaine the idea with extras is pretty much to host your type of thing May 05 19:32:14 totally, but as arjan points out, that might not always be the case May 05 19:32:42 legal questions naturally need to be looked at sure, but at least the wlan drivers are clean? May 05 19:33:50 well, that depends on who to you talk to May 05 19:34:34 GregKH took exception to me providing a binary rpm of the broadcom drivers as he sees any kernel drivers as derivative works and therefore covered by the GPL May 05 19:35:05 I respected his position and came up with a more complicated source rpm route May 05 19:35:50 which works nicely, but is hard to distribute May 05 19:36:27 yes, took a good bit work to make it as simple as possible May 05 19:37:05 * lbt stomps all over lcuk May 05 19:37:06 thanks lbt for posing that question earlier ;) May 05 19:37:35 yes, I was going to follow up, WHEN THE TIME WAS RIGHT, ;) May 05 19:37:44 its ok lbt i was OT - thought we were on other business and arjans point piqued interest May 05 19:39:26 but I'm happy to see some constructive tsg meeting today May 05 19:39:51 meeto :) May 05 19:39:59 :) May 05 19:40:17 formatting and balance is very comfortable and people are managing to speak clearly May 05 19:40:42 even when misfires occur May 05 19:41:10 arjan: I'm just prodding about mic2 vs kiwi... IMHO you should aim to move away from mic2 May 05 19:41:37 but clearly not until something meets the needs May 05 19:44:13 well, at least the project structure looks sane May 05 19:44:18 even though some annotation would be good May 05 19:45:18 lbt: I disagree. May 05 19:45:34 lbt: kiwi is not there yet. and also, sometimes it's ok to be leading in some area rather than following. May 05 19:45:38 OK but you did say "=0" May 05 19:46:31 I wasn't there when meeting started. But I have an impression now is that MeeGo will be a distribution only and any "dev projects" will be treated similar like other upstream projects like x.org. Am I right? May 05 19:47:10 ezjd__: more or less, but they're still under the meego name/project May 05 19:47:19 it's ok to take credit for the work we all do for meego May 05 19:47:22 then there are the apps May 05 19:47:29 we're not, say, ubuntu or some other integration-only distro May 05 19:47:34 sample and real apps for each segment May 05 19:47:36 ezjd__: that's probably good though May 05 19:47:43 does meego have better maps than maemo5 May 05 19:47:45 ezjd__: not quite... we think the Extras-ish will help May 05 19:47:53 like the desktop/homescreen May 05 19:48:04 dhq: not yet May 05 19:48:09 oh May 05 19:48:21 thankx for the info :) May 05 19:48:47 arjan: I know how important security is. I am not surprised it's being done right. I just want to see it explicitly :) May 05 19:49:24 That sounds OK to me. My next question is that for a dev project like UI framework I am interested, will it run pretty much independent to MeeGo? May 05 19:49:49 hmm... not so much May 05 19:50:01 its main goal is to provide a UI framework for meego, so it can't be that independent May 05 19:51:08 thiago_home: but it would be nice to ensure it is not "designed to be unuseable" May 05 19:51:33 ie it uses standard apis, is freedesktop compliant etc. May 05 19:51:43 lol May 05 19:51:51 thiago_home: But I though it will be easy to use MeeGo framework on other distribution since the underlying infrastructure is very similar. May 05 19:51:54 well, it might be designed to be used on meego only May 05 19:52:08 making it work on other distros may not be easy May 05 19:52:28 for example, current Maemo SDKs use Scratchbox. That's a PITA to get working outside of Debian 32-bit based distros. May 05 19:53:02 thiago_home: give me a week... May 05 19:53:22 we have a community OBS running May 05 19:53:54 I don't want an OBS. I want to build on my own machine. May 05 19:54:05 but fortunately, I've already built my own toolchain. It works just fine. May 05 19:54:07 then you don't understand OBS :) May 05 19:54:21 except Creator is hardcoded to think that "anything that isn't MADDE is Desktop" May 05 19:54:37 whose fault is that? May 05 19:54:40 :P May 05 19:54:58 the berlin guys, of course, but I didn't insist on a fix because I know they have more important things to do. May 05 19:56:09 ooh, shiney....we're getting a gui... May 05 19:56:27 and by the sound of it... xeyes too! May 05 19:56:33 what? no more xterm only? May 05 19:56:38 lbt: dang, and I was hoping we'd keep the fancy terminal window as the UI :) May 05 19:56:40 woohoo, xeyes! May 05 19:56:46 DawnFoster: can I ask a question now ? Or should I keep waiting ? May 05 19:56:49 xeyes for touch interface :-) May 05 19:56:50 they're letting in the riff raff now :) May 05 19:56:57 We're about out of time May 05 19:57:33 Looks like to me that the UI is Moblin UI only :) May 05 19:58:02 i think some questions or #meego afterdiscussion might be worthwhile personally May 05 19:58:15 agreed May 05 19:58:23 handset UX not being part is a bit of a shocker :) May 05 19:58:43 (even though it does lessen the pain a bit that it's coming out soon after.) May 05 19:58:45 sound like a scheduling issue? May 05 19:58:57 probably a team that didn't deliver May 05 19:59:04 lol May 05 19:59:05 slackers! May 05 19:59:10 (so it goes everywhere) May 05 19:59:16 wtf May 05 19:59:23 no handset UX? May 05 19:59:24 TSCHAKeee: +1 hr May 05 19:59:26 :( May 05 19:59:30 they couldn't get back from Beijing in time ;) May 05 19:59:34 TSCHAKeee: nah, just delayed May 05 19:59:47 shit i missed it again? May 05 19:59:50 :( May 05 19:59:55 today has totally sucked May 05 20:00:10 really pleased to see the "structure" diagram... thanks DawnFoster :) May 05 20:00:18 girlfriend dumped me, another job offer told me no, broke, electrics about to be turned off, rent not paid. May 05 20:00:35 ouch TSCHAKeee :( May 05 20:00:54 heh... I hear MeeGo is hiring.... May 05 20:01:06 TSCHAKeee: perfect time to apply for a job at meego devices May 05 20:01:06 :P May 05 20:01:14 Oh really ? May 05 20:01:15 stskeeps: we're having some issues with the meetbot - none of the topics appeared in the minutes May 05 20:01:22 DawnFoster: hmm. May 05 20:01:28 No loss tere May 05 20:01:32 yeah distmaster... go debug it May 05 20:01:38 DawnFoster: i'll take a look May 05 20:01:41 thiago_home, the first time I installed the xeyes widget on Fremantle I had my N900 sitting on my desk. Tapped all over the screen trying to get them to move before I realized it was accelerometer based. . . . May 05 20:01:47 TSCHAKeee: yikes - sorry about your day May 05 20:01:51 I saw them earlier ( was 10 mins late and looked at the online log) May 05 20:01:51 * lbt makes a note never to write a quick script to help someone out.... May 05 20:01:59 lbt he would debug it, but the OBS wont build the debug tools :p May 05 20:02:00 DawnFoster: i think one of the issues might have been that it wasn't opped from start on May 05 20:02:10 i'll quickly test that theory May 05 20:02:15 GAN900: hehehe May 05 20:02:17 DawnFoster: uh, as long as the logs are complete May 05 20:02:23 you can always just replay them May 05 20:02:35 I guess the <'s and >'s might have confused meetbot May 05 20:02:52 oh nvm he's already add that ,) May 05 20:02:52 GAN900, xeyes is the first time any of liqbase code was integrated into another app o_O May 05 20:03:00 s/add that/at it/ May 05 20:03:01 th0br0 meant: oh nvm he's already at it ,) May 05 20:03:12 DawnFoster: yeah, i think it was that :/ May 05 20:03:23 DawnFoster: i'll see if there's some commands for log replay May 05 20:03:28 btw, lbt, what's that about obs and you? May 05 20:03:31 th0br0: rpm school! May 05 20:03:33 Stskeeps: afair there are. May 05 20:03:36 lbt: yes? May 05 20:03:54 th0br0: I'm setting up an OBS with X-Fade May 05 20:04:06 for teacing packaging? May 05 20:04:14 *teaching May 05 20:04:26 it will be a prototype for the community OBS for Maemo and MeeGo May 05 20:04:39 so, 2 topics th0br0 :) May 05 20:04:45 :) ok, cool. May 05 20:04:47 1. Community OBS (as you asked about) May 05 20:04:53 community OBS MeeGo -> repository wg? May 05 20:04:56 2. RPM training (I need help) May 05 20:05:10 yes community OBS MeeGo -> repository wg May 05 20:05:12 go ahead, I'll try to help you as best as i can. May 05 20:05:18 DawnFoster: if you give me a bit i'll see if i can make a new log that fixes things :) May 05 20:05:35 god it's hard when you interleave irc with yourself May 05 20:05:37 that's great to hear. I've to admit i haven't been following the mailing lists recently; too much trolling. but everything seems to be taking on form now, I guess I'll start doing it again. May 05 20:06:21 so, my internal team at nokia is doing some work to help the internal devs learn how to transition from deb to rpm May 05 20:06:37 and we are getting permission to publish that to meego.com wiki May 05 20:06:39 huh, you're from nokia? May 05 20:06:49 with my other hat on, yes May 05 20:06:54 oh, didn't know that :) May 05 20:07:03 I'm a community bod ... they hired me May 05 20:07:13 ok May 05 20:07:18 :) May 05 20:07:23 recently? May 05 20:07:25 stskeeps: thanks so much for looking into it! May 05 20:07:32 so I advocate from the inside too ... gently May 05 20:07:36 ^^ ok May 05 20:07:44 6+ months May 05 20:07:49 oh right May 05 20:08:21 and seemingily i'm outed by now as working for meego too :P May 05 20:08:21 stskeeps: weird, I op'ed it about 15 minutes before the meeting May 05 20:08:29 so what do you need my help on? May 05 20:08:54 th0br0: basically the guys writing this stuff are learning as they go May 05 20:08:56 I'll only be around for about 30-45 minutes more tonite as i'm writing a test tomorrow, but i'll be around from 16:00 CEST tomorrow I guess. May 05 20:08:58 ok. May 05 20:09:06 DawnFoster: that is weird May 05 20:09:18 so would take any advice if you get time to keep an eye on it May 05 20:09:36 lbt: sure. May 05 20:09:49 eg http://wiki.meego.com/Packaging/Deb_conversion_example May 05 20:10:20 on a general note, we really need to look at meego policy May 05 20:10:29 it's a tad vague in places :) May 05 20:10:31 there is no really complete official one yet is there? May 05 20:10:31 k May 05 20:10:44 http://wiki.meego.com/Documentation_backlog May 05 20:10:51 no. May 05 20:11:03 lbt i thought make install was depreciated yet its squarly in the middle of the example? May 05 20:11:12 it's a crucial part of making it work May 05 20:11:16 lcuk: correct May 05 20:11:18 Ev'ning May 05 20:11:31 lcuk: just like distributing src is deprecated... :) May 05 20:11:33 and wont the rm -rf be dangerous depending on buildroot location May 05 20:11:37 lcuk: May 05 20:11:38 no May 05 20:11:45 buildroot is a chroot May 05 20:11:48 *must* be May 05 20:12:06 hopefully will be else ww3 occurs May 05 20:12:07 lcuk: you can be a guinea pig too ... you'll learn lots and you can help debug it May 05 20:12:38 lbt: Does the Nokia Qt SDK do anything with packaging? May 05 20:12:40 I also took a look at changelogs this week May 05 20:12:44 th0br0, i come from a simpler time when you didnt need external services and chroots just to build some cdoe May 05 20:12:45 code May 05 20:12:50 true May 05 20:12:52 Jaffa: yes, it makes .debs :) May 05 20:12:55 but you need that when creating packages ;) May 05 20:12:59 http://wiki.meego.com/Packaging/Guidelines#Changelogs May 05 20:13:04 thats just scary tho! May 05 20:13:19 lbt: Right, so targetted at Ovi and closed source stuff May 05 20:13:29 DawnFoster: new theory - qgil was chairing the meeting and hence it ignored all your #topic's May 05 20:13:30 lbt: Bah, humbug. May 05 20:13:34 Jaffa: I'm not up-to-date to be fair :) May 05 20:13:48 https://fedorahosted.org/rpmdevtools/ << any plans on getting that into meego repo? May 05 20:13:51 although I'd expect it to be aimed at VB-level coders :) May 05 20:13:53 Stskeeps: oh, interesting May 05 20:14:10 lbt: I'd be surprised if it spat out a .tar.gz with a debian/ and a .dsc (or a .tar.gz and a .spec) May 05 20:14:23 lcuk: that's how I made my toolchain. A cross-compiler and a --sysroot= switch May 05 20:14:31 Jaffa: I saw something on the ml to that effect May 05 20:14:33 lbt: don't you know - it's the future! May 05 20:14:35 I think May 05 20:14:41 bah humbug May 05 20:14:48 thiago_home, i just said sod it and built directly on device May 05 20:14:53 stskeeps: I'm in a meeting now, but we should try to confirm that May 05 20:14:59 make install on the n900 is surprisingly effective May 05 20:15:07 same as gcc May 05 20:15:08 and keep in mind that chair must set topics May 05 20:15:11 lcuk: my N900 has nowhere the processing power of the compile farm in the office May 05 20:15:30 sure but if you are modifying a single widget within your app its fine May 05 20:15:36 th0br0: need more info... but arjan is probably the man to ask. I'd suggest putting it into extras though May 05 20:15:37 most people do not require to build all of qt May 05 20:15:41 * Jaffa 's N900 seems to have the power of our build farm for the past few days :( May 05 20:15:42 DawnFoster: i just confirmed it by editing the log file and replaying it after editing all DawnFoster> #topic to qgil> #topic May 05 20:15:44 ok lbt May 05 20:15:46 Damn VMware May 05 20:15:48 its same pricniple as just using python May 05 20:15:53 lcuk: even libqt4-dev is maybe too big for the device May 05 20:16:11 sure - but pyqt works and includes all the same defines... May 05 20:16:17 stskeeps: thanks! that seems like a bug to me :) May 05 20:16:22 but one we can work around May 05 20:16:24 DawnFoster: http://trac.tspre.org/meetbot/meego-meeting/2010/meego-meeting.2010-05-05-18.58.html is fixed minutes, old ones left as http://trac.tspre.org/meetbot/meego-meeting/2010/meego-meeting.2010-05-05-18.58-broken.html May 05 20:16:24 lcuk: the problem is that make install is selfish :) May 05 20:16:26 lbt: http://pastebin.com/urrKi4MK May 05 20:16:27 ergo - it SHOULD work May 05 20:16:49 lcuk: that's different, that's python May 05 20:16:59 thiago_home, think about the principle tho May 05 20:17:05 Source0 upstream or debian/copyright << doesn't make much sense May 05 20:17:07 you say libqt4-dev wont work May 05 20:17:14 that is just a list of function defines etc May 05 20:17:21 DawnFoster: next time quim or you can do #chair DawnFoster May 05 20:17:23 th0br0: so we're looking at templating and accelerating package creation. So "hints and tips" like this that eventually become best practice May 05 20:17:27 DawnFoster: and it will accept you as chair as well May 05 20:17:31 *much sense May 05 20:17:35 sure lbt May 05 20:17:50 perfect! I'll add that to my notes & I'll let quim know about this May 05 20:18:03 lcuk: one more thing: the libqt4 build is a cross compilation, so uic, moc, rcc and qmake are x86 binaries May 05 20:18:03 IMHO it would be easier to semi-copy fedora's packaging guideline in the end tho... i doubt that packaging policy will differ that much May 05 20:18:06 ImadSousou: are you around? May 05 20:18:23 thiago_home, well compile those for arm May 05 20:18:26 th0br0: 100% agree! May 05 20:18:26 but nvm. May 05 20:18:27 DawnFoster: have a good meeting, i'm afk for tonight :) (and remember to post the minutes to mailing list) May 05 20:18:30 :) May 05 20:18:42 th0br0: no. please do mind.... May 05 20:18:46 lcuk: yeah, but we haven't done that yet May 05 20:18:49 yeah right :) May 05 20:18:51 lbt: did you see gcobb and I kicking around the idea of making mud target SRPM (on meego-dev)? May 05 20:19:00 thiago_home, whats gonna happen when theres arm chipset big enough to be usable as desktop May 05 20:19:05 that deb_conversion_example is just a draft tho, right? May 05 20:19:09 Jaffa: I want you + gcobb to help me proto my community OBS May 05 20:19:14 isnt it just a case of building full qt for arm May 05 20:19:20 lcuk: when that happens, desktops will be still 10x faster May 05 20:19:29 th0br0: it's a "release early" and the best we can do so far :) May 05 20:19:30 are you sure? May 05 20:19:31 (help) May 05 20:19:31 i mean, transmission might be a bad example if you don't build subpackages / transmissiond... after all, you don't have to use %config etc otherwise May 05 20:19:32 lbt: happy to :) May 05 20:19:44 stskeeps - have a good night, and thanks for the corrected minutes - I'll post to the mailing list shortly :) May 05 20:19:48 the best based upon the OBS status or your knowledge? May 05 20:20:00 Jaffa: I think that will replace/supercede mud May 05 20:20:04 kinda May 05 20:20:08 lcuk: pretty much. ARM is not targetting server business. May 05 20:20:08 besides, shouldn't Requires: meego-lsb be actually part of the default buildroot? May 05 20:20:10 thiago_home, all im saying is its technically feasible to do native compilation, it may be impractical in the qt sense May 05 20:20:14 lcuk: it's targetting the low power market May 05 20:20:27 lbt: Yeah, I think you're right. Probably. May 05 20:20:28 You can split buildrequires to multiple lines to facilitate readability May 05 20:20:29 th0br0: mmm May 05 20:20:30 lcuk: sure, technically it is feasible. May 05 20:20:45 lcuk: my whole point was practicality. It's a lot more practical to cross-compile and deploy. May 05 20:20:52 you forgot to remove the group in %postun lbt May 05 20:20:53 thiago_home, ifa cluster of arm chips can fit 100x more chips in a cabinet because of their power profile then its feasible May 05 20:20:58 th0br0: not me :) May 05 20:21:06 lbt: But then we'll want the app SDK to spit out something which can be submitted to the community OBS May 05 20:21:10 lcuk: the office next to mine is the S60 Test Lab May 05 20:21:13 th0br0: actually I admit, I've barely read it May 05 20:21:15 :D May 05 20:21:21 Jaffa: *nod* May 05 20:21:28 so, who's up for making a emacs based handset UX since we won't have one for n900? ;) May 05 20:21:29 lcuk: there are about 50 N97s, 5800, N86 and N95 in there May 05 20:21:30 Jaffa: and we get osc build too May 05 20:21:37 Stskeeps: woo hoo May 05 20:21:39 (until it's out) May 05 20:21:43 lcuk: they had to install an extra A/C May 05 20:21:48 ok May 05 20:21:53 Stskeeps: see org-mode! May 05 20:21:59 it's noisy inside May 05 20:22:03 lbt: *nod* - but remembering app SDK runs on more than Linux :) May 05 20:22:14 Stskeeps: Eclipse UX ;) May 05 20:22:15 though, truth be told, the noise and heat come from the base cradles and the routers May 05 20:22:27 th0br0: please feel free to edit the page/discussion page May 05 20:22:38 Jaffa: yes, so does osc :) May 05 20:22:41 heh thiago_home May 05 20:22:47 o/ InformatiQ May 05 20:22:52 lbt: will you be around tomorrow? May 05 20:23:05 Stskeeps: Eclipse Clearly Leads In People Superceding Emacs May 05 20:23:05 yes.. ping me to get my attention May 05 20:23:19 ok. i'll try to give the page some love too May 05 20:23:25 lbt: ah, cool. Last time I asked the response was less.... positive May 05 20:23:40 Jaffa: Eclipse. emacs for the 23rd century... May 05 20:23:49 lbt: zactly :) May 05 20:24:03 ie you need to wait that long for CPUs to run it in real time! May 05 20:24:25 does Eclipse still have a button to run the garbage collector? May 05 20:24:46 thiago_home: No, that's always been a third-party plugin May 05 20:24:47 * lbt is convinced java is a black-ops project by Kingston May 05 20:25:21 thiago_home: which I've never had need for, but some of my colleagues like it for its provision of a memory monitor May 05 20:25:25 lcuk: like "humans were invented by water to transport it uphill" ? May 05 20:25:42 ! May 05 20:25:51 thiago: bsd fortune cookie db? May 05 20:26:00 by kingston? huh? May 05 20:26:10 they make memory... May 05 20:26:11 th0br0: Memory manufacturer May 05 20:26:17 ah huh... :D May 05 20:26:43 * CosmoHill didn't fall asleep >.> May 05 20:27:06 Someone asked today if our product would run, in production, on a 1GHz Celeron with 18MB of RAM. "No" was the short answer. May 05 20:27:22 Ha, er, s/18/128/ May 05 20:27:56 just who still uses such old systems in production May 05 20:27:57 thats high spec! May 05 20:28:30 The relase has been branched and is being matured now (qgil, 19:50:30), anyone have a link? May 05 20:28:31 th0br0: Noone, it was a new grad employee looking foolish :) May 05 20:28:36 Jaffa: written in java? =) May 05 20:28:37 ah ok May 05 20:28:51 RST38h: Indeed May 05 20:29:02 Jaffa: well...you know... May 05 20:29:11 Java rocks! May 05 20:29:16 yeah, grads... May 05 20:29:16 th0br0: Trying to impress with his knowledge of CPU types or something May 05 20:29:20 Isn't that what you wanted to say, RST38h? ;) May 05 20:29:23 RST38h: Shush you. May 05 20:29:42 * lbt hugs java May 05 20:29:46 tightly May 05 20:29:50 very very tightly May 05 20:29:51 necrophile May 05 20:29:51 * lcuk drinks java May 05 20:29:55 RST38h: not at all. May 05 20:30:14 * Stskeeps glances at minutes May 05 20:30:16 lbt - round the neck? May 05 20:30:20 'director of nokia meego' May 05 20:30:29 is that meego devices or what does that mean? :P May 05 20:30:42 Stskeeps: "owner" would sound cooler May 05 20:30:43 I saw that Stskeeps... ain't irc conversation.... persistent May 05 20:30:51 it means a guy called meego is now director of nokia May 05 20:30:55 (I know we cannot wish for "emperor") May 05 20:30:56 Stskeeps: Presumably th former. Otherwwise: game over. May 05 20:31:27 lbt: could be worse, could be trying to be a cross-platform, web-based product ISV with C++ May 05 20:31:47 yeah. that'd never work! May 05 20:32:15 what about those suckers that built kernels and things in .. c ? May 05 20:32:15 not unless you had a mobile phone sugar daddy May 05 20:32:39 surely it wouldv been better in java :D May 05 20:32:48 lcuk: perhaps different technologies are better suited to different tasks? May 05 20:32:56 (there (is (only) (one) true) way) May 05 20:32:58 indeed May 05 20:33:14 perl for web apps May 05 20:33:39 lbt: There'll probably be a port of DUI to JavaScript output. GWT for Qt May 05 20:33:46 * suihkulokki remembers the days when sun was boldly claiming how every bulb will have a ip address and run on picojava... May 05 20:34:03 lcuk: Perl's the exception. It's perfect for everything. May 05 20:34:12 except reading? May 05 20:34:16 suihkulokki: LightFactory? May 05 20:34:20 Perl is outdated. May 05 20:34:25 suihkulokki: Well, yes :) May 05 20:34:25 write once, cry everywhere May 05 20:34:27 perl6 man!! May 05 20:34:32 C++ for webapps! Wt++! May 05 20:34:35 and demonstrating your fact jaffa, i compiled your comment up and its a working xml processor! May 05 20:34:39 (wihtout the ++ after Wt tho) May 05 20:35:01 http://www.webtoolkit.eu/wt May 05 20:35:02 ;) May 05 20:35:27 Jaffa: so... mud on an OBS... May 05 20:35:34 how does that work then? May 05 20:35:54 is it just about declaring a repo that is "meegoplus" May 05 20:35:59 lbt: On N900 so dodgy West Coast Main Line data May 05 20:36:13 and bunging a load of libraries+apps in there May 05 20:36:31 ah, commuting? May 05 20:36:48 What's wrong with C you guys May 05 20:36:48 lbt: can the repo have a pre-processor? "mud" would then munge whatever it got in with some heuristics May 05 20:37:01 lbt: yeah, late nights for next 3 months or so. May 05 20:37:20 I'm thinking that we should focus on a repo and some policy first May 05 20:37:36 with a strong eye to automating it with mud l8r May 05 20:38:02 I'm also writing an event system for nokia OBS/QA systems May 05 20:38:08 lbt: absolutely, you mentioned mus ;) May 05 20:38:17 bah. sodding keyboard May 05 20:38:28 well, just keeping you keen :) May 05 20:39:01 Intel state the z6xx (Moorestown) will support Meego so does anyone know if that means they are going to produce a Free "driver" for the powerVR based gma600? May 05 20:39:28 right, I'm really out now. bye. Jaffa I'd prefer to have a MeeGo MUD :) May 05 20:39:28 lbt: there are loads of comments on meego-* about providing h/w for repos etc. Didn't Mer use OpenSuSE's? What's the story there? May 05 20:39:41 we want our own IMHO May 05 20:39:51 we'd integrate with SSO May 05 20:40:00 (where MUD == "Multi User Dungeon") ;) /me signs off. May 05 20:40:01 automate promotion like the autobuilder May 05 20:40:02 th0br0: You have encountered a tmo troll waving a 770 at you. You have: no tea. May 05 20:40:36 plus the opensuse OBS is heavily used....ie slow May 05 20:40:45 lbt: Thought: EC2 or similar for scalability May 05 20:40:54 heh... I have an AWS account May 05 20:41:00 I'm working on it :) May 05 20:41:12 lbt: :) May 05 20:41:17 worried about security though May 05 20:41:37 we can't create a VM inside an EC2 VM May 05 20:41:42 lbt: They're doing VPN stuff now, AFAIK May 05 20:41:46 i hear the N word May 05 20:41:51 and chroot is escapable.... May 05 20:41:55 dragged in mud May 05 20:41:57 hey InformatiQ... yes indeed May 05 20:42:00 have respect May 05 20:42:10 lbt: Some h/w as core & master - then spawning EC2 instances when demand is high May 05 20:42:33 now EC2 May 05 20:42:34 the problem is that the EC2 systems would run user-provided stuff as root May 05 20:42:41 some one breif me in May 05 20:42:48 lbt: Ah. May 05 20:42:58 InformatiQ: possibly scaling the OBS using EC2 workers... May 05 20:43:03 I've been looking at it May 05 20:43:20 Jaffa: yes... ah May 05 20:43:26 hmmm not bad May 05 20:43:29 lbt: sbox uses chroot ultimately, we don't worry about escapage on Maemo autobuilder May 05 20:43:38 (o we?) May 05 20:43:39 on demand workers in cases of high load May 05 20:43:41 pas as you go May 05 20:43:41 Jaffa: stakes are higher May 05 20:43:46 InformatiQ: *nod* May 05 20:43:51 lbt: true May 05 20:43:58 InformatiQ: exactly May 05 20:44:19 Reduces capital expenditure for community repo too, especially at start May 05 20:44:31 Easier sell to sponsors May 05 20:45:41 and what is mud? May 05 20:45:43 it would be an interesting thing to make work... May 05 20:47:15 InformatiQ: http://mud-builder.garage.maemo.org - makes it easier to create Debian packages for Maemo from upstreams (e.g. tarballs, svn etc.) without upstream having Debian packaging support May 05 20:48:24 InformatiQ: And without maintaining a whole copy of upstream just to add that, especially when most of it is machine creatable. May 05 20:48:46 sounds interesting but depends on the implementation May 05 20:49:12 Jaffa: this is one reason I'd like us to adopt a policy of "when in doubt, Suse/Fedora/... is upstream" May 05 20:49:31 InformatiQ: I'd point you to my vim & vala packages in MUD if I wasn't on a slow, laggy link May 05 20:50:19 lbt: problem with a similarly packaged upstream is if it's base policies (dependency names, package responsibilities) aren't compatible. May 05 20:50:35 *exactly* May 05 20:50:40 lbt: found that with the Debian upstream mode of mud early on May 05 20:50:57 this is the reason I'm pissed that MeeGo didn't make that statement for the core May 05 20:51:05 it wouldn't have hurt... May 05 20:51:07 lbt: which is why mostly "upstream" for mud packages is real upstream May 05 20:51:56 yeah, which means you still need to follow a package chain, find the upstream etc May 05 20:52:26 you can't just adopt the package chain and create a -maemo1 variant :) May 05 20:52:59 lbt: indeed May 05 20:53:12 lbt: and it only gets worse over time :( May 05 20:53:37 and it adds zero value May 05 20:54:16 * lbt grinds teeth :) May 05 20:55:40 * lcuk shoots lbt, my teeth are sore again May 05 20:55:51 * lcuk needs a new mouth May 05 20:56:13 * lbt wonders just what would be possible on a community OBS... May 05 20:56:25 hmm May 05 20:56:56 lbt if i were developing at home May 05 20:57:04 how would i make a package to install/test myself May 05 20:57:09 does everything need to go through obs May 05 20:57:13 http://xkcd.com/ May 05 20:57:16 lol! May 05 20:57:24 I think your dad did that lcuk May 05 20:57:40 groan May 05 20:58:20 you mean for the OBS May 05 20:58:37 you know you were talking chroots earlier May 05 20:58:38 i mean, if im developing appXYZ May 05 20:58:44 and i want to test it on my device May 05 20:58:53 hmmm May 05 20:58:57 currently i dpkg_buildpackage and jobs done i copy .deb and install May 05 20:59:13 rpmbuild May 05 20:59:16 (assuming scratchbox style) May 05 20:59:20 lcuk: should be the same with osc build May 05 20:59:36 write a tarball, specfile, run rpmbuild and you get an rpm May 05 20:59:59 ok, and that rpmbuild is the central core function that all of OBS is built around? May 05 21:00:01 lbt: has to be in a meego chroot May 05 21:00:14 lcuk: kinda May 05 21:00:23 does it have protection against being run outside a chroot May 05 21:00:29 obs is built around "a command" May 05 21:00:37 crap, it"s eletrion day tomorrow May 05 21:00:43 so OBS can be used to build .debs? May 05 21:00:45 which can be rpmbuild or dpkg-buildpkg May 05 21:01:04 around the outside of that command it creates a chroot May 05 21:01:18 and it looks at your build-deps to decide what to put in the chroot May 05 21:01:41 all of which is done using the "build" command May 05 21:02:03 now the OBS *service* runs the build command for you on a big compile farm May 05 21:02:36 and (thiago_home are you listening) the OBS command "osc build" runs the self-same build command on your home machine May 05 21:02:57 so the build-service and your home machine run the same code.... guaranteed May 05 21:03:18 and that chroot even contains the toolchain May 05 21:03:43 yes? May 05 21:03:52 sounds reasonable May 05 21:04:02 but how does it know to build that package for 5 different arches May 05 21:04:28 for arch in arch1 arch2.... ; do osc build ; done May 05 21:04:29 win/lin32/lin64/mac/armel etc May 05 21:04:35 ok May 05 21:04:43 ah, careful May 05 21:04:48 so machines in the cluster wont be specifialised for arch May 05 21:04:49 they're OSes, not arches May 05 21:04:59 theres a difference? May 05 21:05:02 well, a mix actually May 05 21:05:07 yes. May 05 21:05:08 targets May 05 21:05:47 you can have debian lenny on 64/32/arm5/arm7/ppc/sparc.... May 05 21:05:58 yes but the same binary wont work in each May 05 21:06:06 or ubuntu/fedora/win32 on each of the same arches May 05 21:06:15 so targets is a better name but can encompas multiple flavours of silicon May 05 21:06:31 yes May 05 21:06:42 OBS has target+arch May 05 21:06:55 so debian_lenny/i586 May 05 21:07:11 or meego_1.0/armv5el May 05 21:07:17 ok, so a general app which was being written to run anywhere - nothing special, "make all" should build generically May 05 21:07:30 no May 05 21:07:33 hello world in native c with simple defines for instance May 05 21:07:56 on windows make install goes to c:\some crappy\ path May 05 21:08:00 will every app have to explicitely specify its targets+archs? May 05 21:08:13 ahh ok understood May 05 21:08:17 autotools solves that May 05 21:08:24 for sane systems May 05 21:08:25 but we havent got autotools May 05 21:08:32 or so i heard May 05 21:08:34 ie all versions of linux that matter May 05 21:08:37 Meego has May 05 21:08:47 meego is a strong autotools distro May 05 21:08:53 wasnt timeless having trouble getting it working? May 05 21:08:53 AFAIUI May 05 21:08:58 heh... May 05 21:09:00 I tried to follow the moblin guide on how to use 'build' to make native packages, but it never really worked out May 05 21:09:20 that was mostly due to packaging issues May 05 21:09:22 slaine: wait until we have an OBS... the SDK will fade into obscurity May 05 21:09:44 lbt, well, more thinking about me, on my local setup May 05 21:09:50 there will always be a need for bedroom coders to build natively without running through obs? May 05 21:10:03 lcuk: no May 05 21:10:11 * slaine is coding in his bedroom right now May 05 21:10:41 sure, but that's like saying there's always a need to be able to code in asm May 05 21:10:47 * lcuk turns off slaine's light - its way past your bedtime and its a schoolnight May 05 21:10:54 * thiago_home goes to bed May 05 21:11:18 lcuk: i agree to an extent May 05 21:11:21 lbt - theres not a lot of need to code in it, but you can be damn sure it needs to be compilable May 05 21:11:26 but then the rpm tools are there May 05 21:11:31 well, for some situations... but 99.999999% of the time... not really. it's just machismo May 05 21:11:43 lcuk, kids are in bed, it's the only time I'll get today to do something productive (was in meetings all day, no coding :( ) May 05 21:11:49 get a meego image use rpmbuild May 05 21:11:54 run it in kvm May 05 21:11:57 build inside May 05 21:12:12 kvm? May 05 21:12:17 InformatiQ: that unfortunately doesn't work May 05 21:12:29 slaine: why not? May 05 21:12:30 I tried exactly that for rebuilding the moblin source rpms May 05 21:12:35 arent virtual machines and systems within systems slower in general? May 05 21:12:37 InformatiQ: Poor packaging May 05 21:12:55 slaine: that was which ver of moblin? May 05 21:12:58 lcuk: do you overclock to get the extra 4%? May 05 21:13:04 2.0 and 2.1 May 05 21:13:17 lbt: he runs gentoo also :) May 05 21:13:25 slaine: do u remember what sort of issues? May 05 21:13:32 lbt was always hard work cross compiling maemo apps within scratchbox within vmware May 05 21:13:38 on windows May 05 21:13:39 it's all in the mail archive May 05 21:13:52 lcuk: I was going to say "then you won't notice the virtual machine slowdown" May 05 21:13:55 that was what led me towards native building and speedups May 05 21:13:58 but for that... May 05 21:14:09 well, i'll give it a try defenitely this week May 05 21:14:20 but even so... you lose more time than you save May 05 21:14:24 I was promised it would be sorted for moblin 2.2 May 05 21:14:35 lbt, sure within vmware native compiling ubuntu apps within the VM were reasonable May 05 21:14:40 but if it can be built in obs (which would create a chroot of meego rpms) then it should work in kvm image May 05 21:14:47 and i didnt for my small light apps May 05 21:14:53 but thats historical May 05 21:14:53 well meego is even better than 2.2 May 05 21:14:57 :) May 05 21:15:09 it has the best of both worlds May 05 21:15:21 InformatiQ: that was the theory. I setup the build as per the instructions on the site and it still failed on those packages May 05 21:15:42 I was told the packaging had been fixed for 2.2, but, as you said, 2.2 got delaye/reworked into meego 1.0 May 05 21:16:17 i would expect this issue to have been fixed May 05 21:16:55 the reason I was rebuilding everything was to have a generic x86 version of moblin that would run on non-ssse3 x86 cpu's May 05 21:17:17 AND IT WOULD ALSO RUN ON ONES WITH SSSE3 TOO May 05 21:17:24 -caps May 05 21:17:37 just a bit less optimal May 05 21:17:48 lcuk: PLEASE USE CAPS WHEN TALKING ABOUT THE SSSE3 CLUSTERFCKUP May 05 21:18:12 OK May 05 21:18:19 that's just intel being bloodyminded if you ask me :) May 05 21:19:14 indeed - especially considering excuse being graphics speedups - the lower spec machines are normally lower resolution thereby negating the boost May 05 21:23:46 I've been tempted to ask for some metrics that proved ssse3 improved real world performance on typical netbook hardware. May 05 21:24:24 well, start by not using the 8087 FP instructions May 05 21:24:39 that's a complete ABI change though May 05 21:25:13 With all the repackaged versions of Moblin UI for fedora spins, ubuntu, opensuse, mandrake etc. you can run the Moblin UX on all sorts of hardware May 05 21:25:16 and it's usable May 05 21:26:18 thiago_home, as lbt said, asm isnt routinely used now and people have obviously rebuilt the stack to systems without it May 05 21:26:23 as slaine says May 05 21:26:41 * slaine manages about 17000 machines that are either Via 1Ghz CLE266 boxes or Intel ULV Celeron with 855GM) May 05 21:26:48 so if we switch PR1.2 to -mfloat-abi=hard, no one will care? May 05 21:28:00 thiago_home, inverse: i would jump at the chance to run through all of maemo through the n8x0 spec builders May 05 21:28:37 so have separate builds? May 05 21:28:46 or run OMAP2-optimised software on the N900? May 05 21:28:54 i do May 05 21:29:04 sorry, let me rephrase May 05 21:29:07 which one are you proposing? May 05 21:29:09 its reasonably recently i upgraded my scratchbox to fremantle May 05 21:29:15 i used diablo for all my compiling May 05 21:29:29 never noticed a difference May 05 21:29:48 apps work the same in general May 05 21:30:01 I don't know of any differences between OMAP2 and 3, though May 05 21:30:14 we're not using hard FP ABI anyway May 05 21:31:20 i fail to understand why there is so much emphasis on atom specifics when intels primary concern for the last 30 years has been backwards compatability with the x86 chip May 05 21:31:56 $$ May 05 21:32:09 yeah but all that silicon and all those developers May 05 21:32:22 sure, people will continue to buy new machines May 05 21:32:29 grr, damn mouth May 05 21:32:32 mouce* May 05 21:32:35 the Atom is a different architecture May 05 21:32:48 yes, x86, but different from the rest May 05 21:33:47 it's special! May 05 21:33:52 iirc, Phoronix ran some tests against Moblin 2.1 Vs Ubuntu and ubuntu came out on tope for most things May 05 21:34:02 so the sse3 compile flag wasn't doing that much May 05 21:34:34 slaine: did you see this weeks Windows 7 vs Ubuntu 10.04 tests? May 05 21:35:01 yeah, only thing that surprised me was that ubuntu performed as well as it did against windows May 05 21:35:23 really? May 05 21:35:30 thiago_home, wouldnt you like to run meego on your desktop instead of other things May 05 21:35:39 they where game tests, and windows does games well May 05 21:35:53 lcuk: actually, no May 05 21:36:02 I'm not the target audience of meego May 05 21:36:16 I'm interested in the other tests which show system performance better, filesystem, process management etc. May 05 21:36:58 anyway, SSSE3 isn't going to get used by the compiler in most applications May 05 21:37:09 the one thing that matters is SSE2 for floating point May 05 21:37:49 slaine, so that includes everything from pentium4 upwards? May 05 21:37:53 thiago_home, sorry May 05 21:38:06 agreed. having ssse3 is just enought to make sure meego won't run (or build) on non-intel hardware May 05 21:38:08 P4 northwood? May 05 21:38:19 yes May 05 21:38:26 why don't we want meego on AMD? May 05 21:38:30 nite all, sweet dreams May 05 21:38:33 we doe May 05 21:38:34 cyas trem May 05 21:38:35 do May 05 21:38:41 Intel don't May 05 21:39:12 so, no handset ux for 1.0 May 05 21:39:23 am I the only one feeling a bit disappointed by that? May 05 21:39:25 actually the PITA isn't running MeeGo on AMD May 05 21:39:44 it's that you can't develop for an Intel Atom device on an AMD desktop May 05 21:39:47 "wait for day one", "wait for 1.0 in may", "wait for it coming Real Soon(tm)" is starting to show a bit of a pattern May 05 21:39:50 hehe bbc made a bobo May 05 21:40:05 you could hear the editor doing the count down in the background to the start and end of the news May 05 21:40:06 lcuk: x86-to-x86 cross compiler May 05 21:40:20 uh? May 05 21:40:36 what I said May 05 21:40:37 has x86 started diverging so much that different subarches now have their own compiler versions? May 05 21:40:56 ah this is true, intel and AMD are different May 05 21:40:58 besides, the toolchain doesn't need to be compiled with SSSE3 May 05 21:41:06 thiago_home: you need more... the chroot needed to run mic2 needs to be non-ssse3 May 05 21:41:10 but only if you use flags to compile it for one processor May 05 21:41:25 w00t_: yes, it's a tad disappointing. I'd imagine we'll see a whole raft of stuff come online between 1.0 and 1.1 May 05 21:41:29 if it can run on Windows and Mac, why can't it run on older x86? May 05 21:42:04 again the distinction between application SDK and platform SDK I guess May 05 21:42:06 slaine: *nod* May 05 21:42:15 thiago_home: yes May 05 21:42:16 the application SDK should pretty much run anywhere May 05 21:42:19 * javispedro envisions application SDK requiring SSSE3 ;) May 05 21:42:29 the platform SDK is for geeks like us May 05 21:42:31 but you can't make an image to put on your device. May 05 21:42:33 javispedro: For x86, it probalby will May 05 21:42:42 slaine: it's just a bit frustrating having things held up is all, I'm looking forward to playing with it :) May 05 21:42:52 aren't we all :) May 05 21:43:04 thank god I can rebuild it then :) May 05 21:43:35 The relase has been branched and is being matured now (qgil, 19:50:30), its available now May 05 21:43:52 I really don't understand what the big deal is. I don't see anyone complaining about Neon requirement. (aside from those who want to run on N800/N810) May 05 21:44:00 and Neon will never be supported on x86 May 05 21:44:07 it's ARM May 05 21:44:16 Neon? May 05 21:44:23 if we can run a completely different arch, why can't we run a slightly different one? May 05 21:44:24 thiago: exactly. they are complaining! May 05 21:44:29 nate@blue[1014]:~ $ rpm -q neon May 05 21:44:29 neon-0.28.2-2.i686 May 05 21:44:32 that program? May 05 21:44:37 no, ARM Neon May 05 21:44:41 my mistake May 05 21:44:57 http://en.wikipedia.org/wiki/ARM_architecture#Advanced_SIMD_.28NEON.29 May 05 21:45:00 thiago_home: because, AFAIK there's no quemu for ssse3 on non-sssse3 devices May 05 21:45:28 lbt: technical issue that can be solved May 05 21:45:45 again, if we can emulate a completely different arch, why not one that is only slightly different? May 05 21:45:56 oh, I agree May 05 21:46:14 I was asking about it months ago... May 05 21:46:20 zero interest. May 05 21:46:41 when MeeGo goes bigger then more devs will hit it when they try to make images May 05 21:46:58 maybe May 05 21:47:04 heck, I think the community OBS uses AMD cpus... that'll be interesting May 05 21:47:21 but my opinion is that, for a distro targetting low-powered processors, we should be squeezing the most out of performance May 05 21:47:47 thiago: agreed with that. now, development tools, on the other side... May 05 21:48:00 javispedro: application SDK vs platform SDK. May 05 21:48:00 that's not the point... the point is that there's no mechanism for non-Intel desktop owners to engage May 05 21:48:05 the application SDK has to run pretty much everywhere. May 05 21:48:14 I've been telling so to everyone who wants to listen. May 05 21:48:39 thiago: both are development tools; why I'd need to create platform images from the target device? May 05 21:48:44 platform SDK also needs to run everywhere. May 05 21:48:44 the platform SDK is a nother matter. Especially for bootstrapping and creating images, and make world. May 05 21:48:46 (evening, thiago_home + lbt and others) May 05 21:49:06 :) May 05 21:49:28 thiago_home: one of the biggest headaches in moblin, as a community member, was the almost daily questions of "can I run moblin on this ?....Why not, it's intel .....That sucks" May 05 21:49:35 anyway, we should optimise the most. And I don't think we should compromise there because some people can't run the software natively *and* there's a clear technical solution (improve qemu) May 05 21:49:40 * javispedro is confused and attacks himself. May 05 21:50:15 otherwise someone is going to come and say we can't have atomic instructions on x86 because someone is still using a 386 May 05 21:50:34 (moot point, of course, Qt doesn't run on 386) May 05 21:50:46 thiago_home: it doesn't? we must fix this! May 05 21:50:50 so therefore AMD users should upgrade? May 05 21:51:02 w00t_: it requires at least one 486 instructin May 05 21:51:11 microlith: no, they should fix qemu May 05 21:51:15 I'm just kidding, don't let me sidetrack the conversation May 05 21:52:06 thiago: nobody is going to emulate its compiler; we've been through this already (scratchbox) May 05 21:52:18 I'm not talking about emulating compilers May 05 21:52:34 compilers have to be compiled for all major archs May 05 21:52:37 then why you mention platform SDK vs application SDK? both should be native. May 05 21:52:41 they have to be compiled for Windows and Mac too May 05 21:52:44 both should generate non-native binaries. May 05 21:53:00 application SDK = cross-compiler only May 05 21:53:13 platform SDK = cross-compiler and a lot more helper tools May 05 21:53:17 * javispedro fails to see it any difference between this and the classic decade old cross compiling environment May 05 21:53:25 thiago: ah well. May 05 21:53:41 the image-creation tool doesn't need to be in the app SDK May 05 21:54:49 "You want to create a MeeGo-based device and launch it in the market? You can certainly invest in a couple of good machines running Linux" May 05 21:55:02 (I'm not saying good == Intel) May 05 21:55:30 as opposed to "You want to make some apps for MeeGo? You can use your existing laptop, running whatever environment you have already" May 05 21:56:04 you just stopped practically every single bedroom coder in one step May 05 21:56:16 yet another barrier for serious development. May 05 21:56:31 I don't follow May 05 21:56:34 me neither May 05 21:56:36 thats *worse* than the entry level required for maemo May 05 21:56:45 having to change machines May 05 21:56:48 thiago's point is that a cross compiler lowers the barrier to entry May 05 21:56:55 entry level required for maemo = Debian-based 32-bit Linux machine May 05 21:57:01 entry level I am requesting = machine May 05 21:57:03 no its windows May 05 21:57:09 actually, machine capable of virtualization :) May 05 21:57:20 I didn't say it had to virtualise May 05 21:57:25 that's still a lot more hassle than most people are willing to go through May 05 21:57:47 thats the level but if they currently code apps for a platform now May 05 21:57:49 if it can run a cross-compiler and you have a USB port, it should be enough May 05 21:57:53 for them to start coding for meego May 05 21:57:56 heck, the SDK works without USB ports May 05 21:58:02 thiago_home: for "app" development,r giht? May 05 21:58:06 right* May 05 21:58:06 you can deploy via ssh over wifi and do remote debugging May 05 21:58:10 microlith: yes, app SDK May 05 21:58:19 now who said that -I- want to only do app development? May 05 21:58:32 if you don't want to do app development, don't use the app SDK May 05 21:58:38 then use the platform SDK May 05 21:58:38 nobody is forcing you to May 05 21:58:44 above I was showing extremes May 05 21:58:54 i have to say that visual studio is still winner when it comes to remote debugging... May 05 21:58:54 I'm not saying that a simpler more portable "madde" is worse; I am saying that this is no excuse for adding more complexity to the platform SDK (scratchbox, etc.) (requiring SSSE3 -- yuck :) ) May 05 21:59:00 different alternatives for different people May 05 21:59:01 right *now* you have to have SSSE3 May 05 21:59:16 what I don't want is for the devices to have to settle for less because *you* don't have SSSE3 May 05 21:59:28 there are other solutions, including improving qemu May 05 21:59:44 I'm in two minds really May 05 21:59:59 improvements to qemu should be done anyway May 05 22:00:07 /are being done May 05 22:00:12 why should ssse3 matter for the core os though, that's what I don't get May 05 22:00:16 hardware (especially in some places) is not readily available, and it's one thing to say "fix qemu" and it's another to expect (potentially hobbyist) developers to do it May 05 22:00:30 w00t_, OSS apps get tidied up May 05 22:00:32 slaine: suppose there is a 5% gain. Should we do it? May 05 22:00:38 lcuk: i'm not denying they don't May 05 22:00:49 thiago_home: quantify that with some tests please May 05 22:00:56 lcuk: but you're barking in the wrong place, really May 05 22:00:57 slaine: not when I say "suppose" May 05 22:00:58 supposeeee May 05 22:01:09 I'm seeing figures banded about that people pull out of their you know whats, no data May 05 22:01:18 I'm not saying there is May 05 22:01:20 lcuk: if a new developer comes along and can't use this, and you say to him "fix qemu and then you can", there is a serious risk they'll wander off and find something else to spend his time on May 05 22:01:23 I'm asking a hypothetical question May 05 22:01:25 s/his/their/ May 05 22:01:25 w00t_ meant: lcuk: if a new developer comes along and can't use ttheir, and you say to him "fix qemu and then you can", there is a serious risk they'll wander off and find something else to spend his time on May 05 22:01:32 suppose someone did the benchmarks and calculated there's a 5% gain May 05 22:01:34 should we do it? May 05 22:01:36 sorry, issed the suppose May 05 22:01:39 missed May 05 22:01:47 * lbt notes that *right now* you can't develop without an ssse3 desktop. May 05 22:01:47 yeah w00t_ May 05 22:02:02 and *right now* there are no shipping devices May 05 22:02:08 btw, when it comes to simd instruction usage and such, any proper app dynamically selects generic or optimized code variant based on cpu capabilities May 05 22:02:17 so *right now* there's no performance benefit to sss3 May 05 22:02:26 and a huge barrier to engagement May 05 22:02:28 thiago_home, currently i am aware of a quantifiable build of libjpeg that has been improved by use of NEON - but that does not mean that libjpeg was useless before May 05 22:02:29 MiskaX: not very liked on a mobile distro. May 05 22:02:33 MiskaX: multimedia apps do May 05 22:02:42 since *right now* there is no app development for meego May 05 22:02:48 JPEG is a case where the GPU should be used May 05 22:02:53 thiago_home: yep, like my code May 05 22:02:59 and is where practical May 05 22:03:05 javispedro: i don't see any reason to avoid that on mobile either May 05 22:03:18 but that doesnt mean you should rebuild the entire OS because of it May 05 22:03:22 I'd love to base our products on a MeeGo foundation, I can't if there's an x86 ssse3 requirement, and I wouldn't be using the UX as it stands. May 05 22:03:33 all those code paths should be handwritten assembly anyway and pretty tiny amount of the total code May 05 22:03:49 with ifdefs to take care of the cases where its not May 05 22:03:50 slaine: that's a complete different take on the subject, though May 05 22:04:07 that's always been my take May 05 22:04:09 slaine: you're saying "I want to ship products on something that is not Atom or ARM OMAP3" May 05 22:04:15 yes May 05 22:04:40 right now, and until the direction changes, Atom and OMAP3 are the base configs May 05 22:04:42 * javispedro envisions MeeGo SDK with MEEGO_X86, MEEGO_X86_SSE3 and MEEGO_X86_ARMEL targets, and much like the current maemo sdk, with the shipped qemu incapable of emulating the latter two ones. May 05 22:04:54 I'll still try once 1.0 is out and I get all the source rpms May 05 22:04:54 *SSSE3 May 05 22:04:57 javispedro: that's probably easy to do May 05 22:05:04 unfortunately arm is a bit lacking on cpu capability detection front, my atomic ops detection code is a bit ugly atm May 05 22:05:08 hopefully the packaging issues have been expunged May 05 22:05:15 hand-written assembly should be selectable. The rest is a flag to the compiler. May 05 22:05:24 you don't detect the CPU in ARM May 05 22:05:31 does anybody know, would the "normal" middle of the road Pentium M without ssse3 be generally faster than the high spec atoms? May 05 22:05:33 you *know* what processor you put in your device. May 05 22:05:41 thiago++ May 05 22:05:42 lcuk: HELL YES May 05 22:05:56 so isnt the speedup point entirely moot anyway? May 05 22:06:04 completely May 05 22:06:10 lcuk: speed? yes May 05 22:06:16 lcuk: power consumption, not by a long shot May 05 22:06:18 well, if the atom is slow, you try to get the best out of it. May 05 22:06:25 thiago_home: that's not portable code... ;) May 05 22:06:29 so I understand the need for ssse3 if it does anything tangible. May 05 22:06:30 the calculation is speed by watt consumed May 05 22:06:40 MiskaX: ARM doesn't care about portable code. Or hasn't until now. May 05 22:06:52 thiago_home, my intel Pentium M based 12" touchscreen tablet works really well May 05 22:07:08 MiskaX: we're not necessarily talking about a single hotspot written in ASM. Think entire programs build for different FP architectures. May 05 22:07:10 i even run it at 600mhz powersave mode :D May 05 22:07:13 thiago_home: just as an example, try to build binary which runs optimally on OMAP2 _and_ OMAP3 May 05 22:07:35 MiskaX, i barely noticed the difference May 05 22:07:36 MiskaX: what's your point, that it's very hard? May 05 22:07:42 lcuk, We've the same cpu's in our gear May 05 22:07:59 i just build with -O2 and didnt upgrade scratchbox from diablo for ages May 05 22:08:27 lcuk: then don't upgrade cause the fremantle compiler is slight slower :) May 05 22:08:31 *slightly May 05 22:08:32 lcuk: the difference is quite huge in certain cases May 05 22:08:35 too late i needed to May 05 22:08:58 javispedro: vtables for high level functions built for two fp archs May 05 22:09:11 MiskaX: WHAT? May 05 22:09:22 and call that optimized? May 05 22:09:29 i'm doing that already on my dsp library for x86 May 05 22:09:41 you expect to do that manually for every application in existence? May 05 22:09:46 sse2, sse3, e3dnow, etc. May 05 22:09:46 are most of us in agreement then that for most cases theres no real technical reason to require ssse3 ? May 05 22:09:58 in most cases there isn't May 05 22:10:06 MiskaX, is special May 05 22:10:14 at least, looking at the SSSE3 instructions, I can't understand why May 05 22:10:15 but he doesnt have to build entire OS May 05 22:10:15 javispedro: no, i use c++ capabilities for that ;) May 05 22:10:19 nothing groundbreaking there May 05 22:10:29 and currently the only thing ssse3 does is stop people from developing for meego? May 05 22:10:37 I've seen gcc use SSE on its own, but no clue what level it was May 05 22:11:02 the one ABI issue I've seen is the FP one May 05 22:11:11 but that doesn't require SSSE3 AFAIK May 05 22:11:14 and thats entirely reasonable thiago_home May 05 22:11:18 well, supposedly, ssse3 offers a large performance improvement in FP code over sse2 May 05 22:11:37 its been proven since early days that FP improvements work well May 05 22:11:52 especially in graphics May 05 22:11:52 but the majority of the system isn't being held back by FP performance May 05 22:11:55 slaine, but we arent saying FP is going to be emulated etc May 05 22:12:15 all compilers are still bad on optimizing things like iir biquad code May 05 22:12:23 a few % makes a the difference between 30fps and 60fps May 05 22:12:31 of course May 05 22:12:37 that's 100% surely May 05 22:12:39 if you miss the 60fps mark, you have to settle for 30 to avoid tearing May 05 22:12:48 surely not! May 05 22:13:05 how does that affect my mke2fs program though ? May 05 22:13:11 if you can transfer the data in <1/60th of a second May 05 22:13:17 tearing is a completely different problem May 05 22:13:25 COMPLETELY May 05 22:13:35 lcuk: if you can't calculate and draw your frame in 16 ms, you've lost May 05 22:13:44 then you might as well take 33 May 05 22:13:52 look, we're not talking about writing games May 05 22:13:57 nor am I May 05 22:13:58 what? May 05 22:14:00 I'm talking about UIs May 05 22:14:04 scrolling, for example May 05 22:14:06 all is about games! May 05 22:14:12 what does the iphone has -- games. May 05 22:14:16 thiago_home, i know all about fluid uis May 05 22:14:18 I'm pretty sure you do scrolling May 05 22:14:18 scrolling is *important*. May 05 22:14:20 AH AH Ah dead leg May 05 22:14:23 just separate class implementations into different source files, use virtual base class and use a runtime-detection factory method to construct the correct specialiced instance May 05 22:14:28 what do people want to do with their underpowered netbooks -- games. May 05 22:14:47 it makes the difference between a beautiful UI experience and a trainwreck May 05 22:15:00 if tearing is a framerate issue, 60fps would be MORE susceptable than 30 May 05 22:15:13 ...different source files can be built with different options... May 05 22:15:21 lcuk: my point is that 58 fps has tearing May 05 22:15:22 tearing is refreshing the screen itself with some of the old and some of the new May 05 22:15:35 yes, 60fps is more susceptible May 05 22:15:40 tearing is because you havent transfered all the data across May 05 22:15:41 this is all theory though of course. the fact is that clutter et al work incredibly well on pretty old and underpowered systems so long as the hardware accelerated rendering works May 05 22:15:43 all the more reason to gain even 1% everywhere May 05 22:15:56 the tearing is 100% system transfer bus speed May 05 22:16:06 and also whether theres buffers on the lcd side May 05 22:16:18 you can have tearing on a 1fps system May 05 22:16:28 but no one wants a 1fps system May 05 22:16:43 I think the Nexus One feels horrible at 15 fps May 05 22:16:47 tearing eink-based ebook readers!! ohnoes!! May 05 22:16:52 but you asserted if you cannot have 60fps you have to drop to 30 May 05 22:16:54 thats silly May 05 22:17:11 lcuk: yes. But the point is that we want 60 fps. May 05 22:17:15 javispedro, shush i actually considered how to render to those once :p May 05 22:17:19 of course thiago_home May 05 22:17:23 and to have 60fps, we need to get all the optimisations we can get May 05 22:17:31 in graphics, were FP performance does matter. May 05 22:17:38 but even the highest spec graphics cards pumping out the highest visuals drop under 60fps sometimes May 05 22:17:54 yes they do May 05 22:17:58 and clearly, the ability to "take two registers, concatenate their values, and pull out a register-length section from an offset given by an immediate value encoded in the instruction" is going to help us reach the 60fps mark. May 05 22:18:02 but it shouldn't be because your graphics backend is slow May 05 22:18:09 it should be because the CPU is busy doing something important May 05 22:18:10 how freeedrich| quently depends on how optimized the graphics drivers are May 05 22:18:46 if ssse3 makes such a difference to those then they will be built with it anyway - we are expecting blobs to be from there May 05 22:18:48 anyway May 05 22:18:54 lcuk: i would compare this with audio playback. you can make audio software which never has dropouts. if you cannot keep up with audio frame rate, you have broken audio. same goes for certain graphics too May 05 22:19:35 realtime systems - my turf... May 05 22:19:45 MiskaX, sure i see graphical glitches every single day May 05 22:19:46 hey, at least PSHUFB is useful. May 05 22:19:59 * javispedro has at least once wished for it. May 05 22:20:04 anyway, to summarise what I am saying: there are solutions to allowing people to develop without having a Core2 that don't imply reducing the optimisation on devices May 05 22:20:06 if you car's abs breaks decided to hang for a second you can have a fatal crash May 05 22:20:23 lcuk: that's called bad software May 05 22:20:38 MiskaX, sometimes caused by hardware May 05 22:20:49 that's bad hadrware design then May 05 22:20:55 and long historical pipelines May 05 22:21:10 running deterministic software on deterministic hardware is deterministic system May 05 22:21:53 ? May 05 22:22:19 it's pretty rare that even a cheap DVD player would not keep up with DVD framerate for the entire length of a movies May 05 22:22:55 sure - because they purchased a single use chip that was designed to do that May 05 22:23:21 that's usually a CPU+GPU+software. i used to work on set-top-boxes... May 05 22:23:24 but if i buy a dvd drive and plug it into my computer with software it can only do the best it can May 05 22:23:34 (recording DVB-boxes) May 05 22:23:47 and all the media players include support for dropping frames and other things May 05 22:23:59 its like gaming: you have options to control quality settings May 05 22:24:23 that's why pc gaming is failing :) May 05 22:24:27 * javispedro ponders what happened to the "why are development tools built for ssse3" story. May 05 22:24:33 i can happily run the same games on a low spec graphics card as a high spec one May 05 22:24:40 javispedro: welcome to IRC, enjoy your stay May 05 22:24:42 i just turn the resolution down and tweak the polygon count May 05 22:25:12 or you buy xbox360 and have exactly pre-defined experience without any adjustments May 05 22:25:48 actually, nobody can do a thing about it so let's just forget about it hoping that amd grabs ssse3 before the meego judgement day May 05 22:26:16 since i've been doing audio software for ages, i find it so funny that graphics people find it hard to keep up with performance spec. May 05 22:26:21 or someone sets up a build server to drag in and redo things as the moblin remixes have shown is possible May 05 22:26:39 i can do several hour long recording over and over again without glitches :) May 05 22:26:47 audio consists of maximum 5 pixels May 05 22:26:48 ;) May 05 22:27:00 lcuk: indeed. hopefully one of the benefits of meego -- that you can redo it easily. May 05 22:27:09 well, 64 channels, 192 khz, 32-bit May 05 22:27:46 jesus audio is 32bit now? May 05 22:27:47 that's 46.8 MB/s + all the heavy DSP processing in realtime for that May 05 22:28:24 just for practical reasons, 24-bit samples are carried in 32-bit words (dma alignment on bus) May 05 22:28:40 can people tell the difference between 16 and 24/32 bit? May 05 22:28:59 i can pump up the cpu load to constant 80% and no glitches May 05 22:29:18 leaving 20% for the graphics \@/ May 05 22:29:22 you bastard :p May 05 22:29:28 sure, especially if you put some 40 dB digital gain May 05 22:29:51 the difference between 16- and 24-bit samples is the noise level May 05 22:30:09 if you're doing audio work, each operation adds to the noise May 05 22:30:22 the extra 8 bits mean the end result has less noise than otherwise May 05 22:31:12 nod, that does sound reasonable May 05 22:31:13 normalized mixing of two channels ((c1 + c2) >> 1) - 1 bit lost already May 05 22:31:29 or 3dB increase in noise May 05 22:33:47 anyway, i fully agree about optimization and fps, but i still tend to think that for most part code can autoselect hardware dependent optimization paths, either by autoselecting .so on-fly or by doing similar inline May 05 22:34:22 and helloworld doesn't have to require simd ;) May 05 22:34:29 MiskaX: it does May 05 22:35:06 for rest of the performance, it's about correctly utilizing kernel scheduler May 05 22:35:41 EDF scheduling for graphics and audio, anyone? ;) May 05 22:36:16 http://qt.gitorious.org/qt/qt/blobs/master/src/corelib/tools/qstring.cpp#line3722 May 05 22:36:25 this has nothing to do with graphics or audio May 05 22:36:28 but uses SIMD anyway May 05 22:36:55 sure, but specifically those can autoselect May 05 22:37:22 thiago_home, i *do* like optimisations like that May 05 22:37:32 the extra indirection would probably cost more performance that the benefit of doing it with sse2 May 05 22:37:38 using hardware available when its built makes total sense May 05 22:37:39 have two qstring class builds, simd and non-simd and autoconstruct correct one May 05 22:37:39 MiskaX: those aren't auto-selected May 05 22:37:54 if you compile with SSE2, you get those May 05 22:37:58 otherwise, you don't May 05 22:38:12 same for the NEON version of the inverse operation (http://qt.gitorious.org/qt/qt/blobs/master/src/corelib/tools/qstring.cpp#line3494) May 05 22:38:42 btw, gcc uses the same intrinsincs on its own code if you let it May 05 22:38:42 are there any similar optimisations built in for the ssse bit May 05 22:38:45 thiago_home: it could be... May 05 22:39:34 * lcuk tried inlining asm into the blitter on liqbase but always failed to get the compiler to accept them May 05 22:39:34 thiago: you mean gcc's neon autovectorizer works with that loop? May 05 22:39:40 in the past i used to have similar thing, but then there was the trouble that x86 linux distros wanted to build for lowest supported hardware May 05 22:40:03 that's why things are autodetected these days May 05 22:40:10 javispedro: I haven't seen SIMD in ARM code by gcc yet, but I have in x86 May 05 22:40:17 MiskaX: not completely true. May 05 22:40:23 let's say there are three ways to do it: May 05 22:40:26 mostly due to loop unrolling May 05 22:40:34 1. common denominator (386) May 05 22:40:35 -ftree-vector May 05 22:40:57 2. common denominator for most apps, ship all specifics versions for certain apps (kernel, glibc, mplayer...) May 05 22:41:07 3. specific version for all apps (gentoo, ... ) May 05 22:41:37 the graphics code does detect SSE2 and disables itself if not present. In fact, it auto-selects other variants. May 05 22:41:49 im going anyway, its been an informative evening, gnite thiago_home javispedro lbt #meego \o May 05 22:41:50 all modern distros are doing (2) May 05 22:41:51 it was judged to be to complex for QString though May 05 22:41:56 gnite lcuk May 05 22:42:03 w00t_, slaine MiskaX \o May 05 22:42:26 nite lcuk May 05 22:42:29 and both maemo and meego do (3) May 05 22:42:43 cause they know the target hardware. May 05 22:42:57 thiago_home: for example, look at http://subversion.jackaudio.org/jack/trunk/jack/libjack/port.c:jack_port_set_funcs() May 05 22:43:26 MiskaX: no doubt May 05 22:43:32 I've seen vlc do some evil things too May 05 22:43:38 CPU detection via SIGILL :-) May 05 22:44:29 thiago_home: this is what i wrote for jack and i don't consider it evil: http://subversion.jackaudio.org/jack/trunk/jack/libjack/simd.c May 05 22:45:37 there's the detection code at the beginning May 05 22:46:22 unfortunately for ARM there's no such nice way available May 05 22:46:54 30 vs 60 fps.. do you mean that at 58 you'd get the occasional duplicate frame, which would make scrollong look jerky? Tearing is when screen displays parts of several different frames simultaneously, and can be avoided as long as cpu can memcpy() a frame fast enough, or, you know, double buffering.. May 05 22:47:01 MiskaX: http://qt.gitorious.org/qt/qt/blobs/master/src/corelib/qsimd.cpp May 05 22:47:18 I meant http://qt.gitorious.org/qt/qt/blobs/master/src/corelib/tools/qsimd.cpp May 05 22:47:59 thiago_home: yep, seems to be same thing May 05 22:48:29 thiago_home: so for qstring you would construct internal private class instance out of various specialiced ones based on cached cpu detection instead of the funny ifdef May 05 22:48:30 there aren't many ways to flip bits in the same registers May 05 22:48:47 not really possible with C++ May 05 22:48:56 QString isn't a virtual class May 05 22:49:22 it could utilize vtables initialized at load time May 05 22:49:35 no vtables. It's not a virtual class. May 05 22:50:22 thiago_home: doesn't matter... May 05 22:50:32 I need to go to bed now May 05 22:50:53 same here, almost 2am.. May 05 22:52:07 really good example of funky vtable is the arm kernel-provided atomic ops May 05 22:53:09 automagically hardware specific without app knowing anything about it May 05 22:55:17 MPlayer's fast scaler on N8x0 constructs the code in ram at scaler init time, in order to avoid indirection and extra branches ;p May 05 22:55:49 and I'm spent May 05 22:55:57 catch you all tomorrow May 05 22:56:05 nite slaine May 05 22:56:15 nite May 06 02:10:40 good morning **** ENDING LOGGING AT Thu May 06 02:59:56 2010