**** BEGIN LOGGING AT Sun Aug 22 02:59:58 2010 Aug 22 07:53:11 Hi all. Aug 22 07:59:18 hi quet Aug 22 08:02:10 hi amjad, I'm keeping an eye on moblin and a litle on meego, more in mailing list than here Aug 22 08:03:27 It supuse I am doing a communication research about ways of doing free software Aug 22 08:03:47 ok,but intel is going meego way, so not too sure if moblin will be maintained Aug 22 08:06:15 I dont it will, but I am doing the analysis abot moblin Aug 22 08:06:27 sorry, Aug 22 08:06:44 I dont think moblin will be still be maintained Aug 22 08:07:07 ok Aug 22 08:07:25 but my research is about moblin Aug 22 08:08:05 so what you want to know from meego folks??, Aug 22 08:08:14 I don't think that's really correct Aug 22 08:08:28 maemo and moblin merged and got renamed meego Aug 22 08:08:37 moblin is not "dead" Aug 22 08:08:48 it is alive and well ... Aug 22 08:08:55 from an OSS point of view Aug 22 08:08:57 I have some questions about the relations between developers, vendors and users. Aug 22 08:09:32 (ie much of the code traces back to moblin) ... of course if you're talking marketing... whole different story Aug 22 08:10:11 a lot of us have questions about that trinity :) Aug 22 08:10:37 I think is both, marketing and code development Aug 22 08:10:49 you're wrong :) Aug 22 08:11:19 the name has changed... the code changelogs are continuous from moblin to meego Aug 22 08:11:30 and yes, I wondering wich kind of beneffits get each goup with the interaction with others Aug 22 08:11:36 no more releases will be made using the moblin name Aug 22 08:12:06 vendors get a very cost effective OS Aug 22 08:12:10 BTW sorry for my english Aug 22 08:12:16 developers get paid to work on it Aug 22 08:12:50 users get the benefits of lower costs since the work is co-operative and has no artificial "microsoft tax" Aug 22 08:13:40 "virtuous circle" Aug 22 08:16:17 and Intel and Nokia? Aug 22 08:16:37 just vendors with strategic vision Aug 22 08:16:37 how an FOSS project benefit them? Aug 22 08:16:49 it already does Aug 22 08:17:06 lets say 200 people work on MeeGo Aug 22 08:17:12 ok Aug 22 08:17:38 Nokia may pay for ~90 but gets the benefit of 200...that's cheap labour :) Aug 22 08:18:08 (a few are volunteers or from smaller partners) Aug 22 08:19:21 can we count there in those smaller partners to trolltech? Aug 22 08:19:30 They also know that MeeGo isn't designed to be obsolete and to force (licensed + costly) upgrade cycles like some OSes we know Aug 22 08:19:45 not really ... nokia bought trolltech :) Aug 22 08:20:02 but in many ways that was just another way of hiring more developers Aug 22 08:20:43 IVI has people working on MeeGo Aug 22 08:20:52 that's an automotive consortium Aug 22 08:21:15 look at the sponsorship/partners pages for more companies Aug 22 08:21:28 but that's the basic model Aug 22 08:23:28 vendors contribute both with financial sponsorship and people? code? Aug 22 08:23:48 yes Aug 22 08:24:09 and, importantly, by adding to the 'weight' of meego Aug 22 08:24:49 demonstrating viability increases the percieved value of MeeGo to other potential vendors Aug 22 08:25:03 something like increase the value of a product with users Aug 22 08:25:41 similar ... but that case is based on predicted revenue streams Aug 22 08:25:52 this is based on predicted value Aug 22 08:25:57 slightly different Aug 22 08:26:05 nice Aug 22 08:26:26 lbt, i think i am close to having bootstrapped non-ssse3 meego Aug 22 08:26:40 neat... are you on the maemo obs? Aug 22 08:26:45 yeah Aug 22 08:27:08 home:stskeeps:ssse3_bootstrap Aug 22 08:27:11 ali1234 was working on identifying it last night Aug 22 08:27:15 :nod: Aug 22 08:27:35 there's a bit of circular dependancies, so i've temporarily disabled gcc and glibc builds Aug 22 08:27:36 thanks lbt, I will think a little about this ;) Aug 22 08:27:39 I could do with some opinions on the "Trunk" project layout Aug 22 08:27:55 also, there's something wrong with your ssl tunneling Aug 22 08:28:01 it peers very often for larger loads Aug 22 08:28:01 quetzalzun: np :) .... posting something to the ml if you write it up would be cook Aug 22 08:28:16 shure Aug 22 08:28:20 couldn't do a client-side copy of vim or even commit it manually Aug 22 08:28:46 Stskeeps: for maemo obs Aug 22 08:29:00 X-Fade is the nw guy there Aug 22 08:29:02 k Aug 22 08:29:06 just noting it :) Aug 22 08:29:14 he's had some issues and iirc put varnish in to help Aug 22 08:29:33 the meego one is a bit more straightforward (for now) Aug 22 08:29:39 :nod: Aug 22 08:29:50 is it causing real problems or just niggles? Aug 22 08:30:17 can't upload vim, but not an important package for build (..) Aug 22 08:30:40 what error ? Aug 22 08:30:53 sec Aug 22 08:31:01 * Stskeeps tries to upload im again Aug 22 08:31:02 vim Aug 22 08:31:34 aside... I noticed the gcc chat yesterday... that looks evil Aug 22 08:32:10 the patch? Aug 22 08:32:20 I may have misunderstood Aug 22 08:32:32 it can be done in a better way, yes, but baselining is a normal practice Aug 22 08:32:33 but Meego core gcc forces ssse3? Aug 22 08:32:39 even on arm Aug 22 08:32:42 or defaults? Aug 22 08:32:48 that's a good question Aug 22 08:33:08 can you do -mno-ssse3? :P Aug 22 08:33:52 this stuff makes me really wonder about hidden agendas Aug 22 08:34:09 although I could just be paranoid Aug 22 08:34:47 i don't think so Aug 22 08:35:31 lbt: it's a fairly normal practice to baseline similar to this, but patch is formed in a way so you can't even do -mcpu=atom Aug 22 08:35:58 it's never been explained why they decided not to base meego off some other distro Aug 22 08:36:26 as it stands, meego is not even self-hosting Aug 22 08:36:39 um, sure it is Aug 22 08:36:50 maybe we should enhance the patch to allow for deselection... functionally the same but negatable Aug 22 08:37:04 you've compiled every single meego rpm on a real machine running meego? (not inside OBS) Aug 22 08:37:22 (since you'll note meego does not have OBS packages) Aug 22 08:37:45 ali1234: that might simply be cos OBS includes packages normal build dep doesn't.. Aug 22 08:37:50 scratchbox syndrome ;) Aug 22 08:37:53 that's not the point Aug 22 08:37:55 hows meego pulseaudio going Aug 22 08:38:17 ali1234: OBS only uses meego packages in the chroots Aug 22 08:38:26 it is, by definition, self hosting Aug 22 08:38:44 lbt: so when meego has an OBS package, meego will be self hosting Aug 22 08:38:48 a standard meego image may not be build-essential complete Aug 22 08:38:54 no Aug 22 08:39:01 OBS is not needed to build meego Aug 22 08:39:15 maybe that's true Aug 22 08:39:15 obs is nothing but a chroot hoster Aug 22 08:39:36 there is no current tested and documented method that anybody can use to build all of meego Aug 22 08:39:50 heh.... Aug 22 08:39:58 ...........................................................................SSL Error: (104, 'Connection reset by peer') Aug 22 08:40:03 hmm Aug 22 08:40:18 timeout...? Aug 22 08:40:28 most straightforward way to implement that is script modules maintained by the people who have built the current packages Aug 22 08:40:34 imo Aug 22 08:40:39 if they based on another distro this would not be a problem since that distro's tools could be used Aug 22 08:40:41 'here are the commands i used' Aug 22 08:41:02 lbt: dunno Aug 22 08:41:03 but if any tom dick and harry could do it, then it would be really easy to run it on competitor's hardware Aug 22 08:41:09 at least you can get compliance/participation when you make i simple enough Aug 22 08:41:24 pupnik_: and that's what we're aiming for Aug 22 08:41:56 X-Fade is offline atm Aug 22 08:42:25 so i have to ask, what do they actually gain from going alone? Aug 22 08:42:30 if you have to track meta information, you really need to demonstrate need. i hate people who have to architect everything Aug 22 08:42:38 ali1234: what distro does fedora base off? :P Aug 22 08:42:55 Stskeeps: uh... red hat? Aug 22 08:43:02 lol Aug 22 08:43:02 and redhat? Aug 22 08:43:07 ali1234: can you see where this is going... Aug 22 08:43:17 besides that, rebuilding meego is a matter of importing prjconf, binary packages, source packages Aug 22 08:43:20 red hat has been around for 17 years, and has a very mature toolset Aug 22 08:44:04 ali1234 stop saying "well, if *I* was solving this problem I wouldn't start from here" Aug 22 08:44:21 we are where we are Aug 22 08:44:37 heh, no, nokia and intel are where they are Aug 22 08:44:40 if you don't like it then invent a time-machine and fix it Aug 22 08:44:41 Besides, Fedora isnt Red Hat Aug 22 08:44:48 if you choose to join them, that's up to you Aug 22 08:45:02 Red Hat pulls from Fedora Aug 22 08:45:03 * lbt checks what chan ali1234 is in... oh, #meego Aug 22 08:45:17 and you can say "well if you;re not with us you're against us" if you want Aug 22 08:45:24 and given the work you've done on obs etc I assumed you were "we" Aug 22 08:45:41 but intel are saying the same thing, that is, when they are not talking about how great the meego community is going to be Aug 22 08:46:03 whereas you'd rather be a doomsayer... sure.. that's easy Aug 22 08:46:22 i'm not saying the project is doomed Aug 22 08:46:30 i think intel and nokia will do well with it Aug 22 08:46:39 http://www.quoteworld.org/quotes/7855 Aug 22 08:46:47 but i think they'll continue to shut out ... outsiders Aug 22 08:46:54 i think there's still room for meego to do something *more* developer friendly Aug 22 08:47:04 pupnik_: err... yeah! Aug 22 08:47:26 frankly I think intel are shooting meego in the foot Aug 22 08:47:33 it's not fatal but it bloody hurts Aug 22 08:47:39 if meego was serious about working with the community, they would work with it, instead of trying to build a new one Aug 22 08:47:40 and that could be the draw for some - if the subsystems are high-performance, developers can get better results on meego than android Aug 22 08:47:45 ali1234, off topic but weren't you discussing psx emulators and hildon-desktop two months ago in #maemo? Aug 22 08:47:53 at least games Aug 22 08:48:00 MohammadAG51: yeah Aug 22 08:48:36 i fear there may be something of a mhz war developing in the phone arena Aug 22 08:48:43 MohammadAG51: have they figured out how to make them run full speed yet? Aug 22 08:48:57 not that i know of, but that's off topic here ;) Aug 22 08:49:08 not if you port it to meego Aug 22 08:49:10 it should be ported to meego: not offtopic Aug 22 08:49:28 but first imho when the gpu portion is running Aug 22 08:49:31 The answer is simple - if you don't like what the community is doing with MeeGo, you have no obligation to be here or participate Aug 22 08:49:37 since we require the damn opengl anyway :| Aug 22 08:50:12 i wouldn't mind so much rendering via software if all transitions and compositing were user-disablable Aug 22 08:51:01 TermanaN900: which community are you talking about exactly? Aug 22 08:51:40 ali1234, the MeeGo community. This community. The people here contributing to MeeGo Aug 22 08:52:44 TermanaN900: so you mean intel and nokia then... Aug 22 08:52:52 the "official" community Aug 22 08:52:55 * MohammadAG51 facesmacks Aug 22 08:53:01 ali1234, no. I mean the community. Aug 22 08:53:16 Intel and Nokia employees are apart of this community Aug 22 08:53:39 so whom outside of intel and nokia has contributed to meego? Aug 22 08:53:45 <- Aug 22 08:53:46 :P Aug 22 08:53:53 Stskeeps: aren't you paid by nokia? Aug 22 08:53:57 LOL Aug 22 08:53:57 i thought they hired you? Aug 22 08:54:00 TermanaN900: "Intel and Nokia are apart from this community" or "Intel and Nokia are a part of this community"? ;-) Aug 22 08:54:06 was under my maemo.org work Aug 22 08:54:22 Jaffa, lol sorry the second one :P Aug 22 08:54:30 ali1234: as well as people sending patches, etc Aug 22 08:54:36 * lbt <---- Aug 22 08:54:37 things are picking up. Aug 22 08:54:47 I have paid and unpaid contributions to meego Aug 22 08:54:57 ok, so what about all the people who contributed directly to the kernel? are they part of the meego community? Aug 22 08:55:03 no Aug 22 08:55:07 what about the people who worked on Qt before meego even existed? Aug 22 08:55:25 are they part of the Ubuntu community? Aug 22 08:55:32 i don't know, ask canonical Aug 22 08:55:43 they decide who is part of their community Aug 22 08:55:57 are debian contributors to the kernel part of the pulseaudio community? Aug 22 08:56:19 lbt: it is an interesting problematique. to some extent, they're doing the right thing - if someone wants to take a shovel, create a non-SSSE3 they are able to do this. once there's a substansial reason to have non-SSSE3 SDK defaults (cos of an installable base), - they can't deny patches to fix 'blockers'.. but why would a manager from OSTC approve they spend money on something that would actually sell less atom processors? ... Aug 22 08:56:25 ... (non-SSSE3 build target on build.meego.com).. Aug 22 08:56:27 so in order to be in the meego community, you have to have patches submitted directly to meego? Aug 22 08:56:42 presumably they have to be accepted to Aug 22 08:56:52 Stskeeps: OTOH that's an Intel argument... not a MeeGo one Aug 22 08:56:55 which means that the maintainer of the software dictates who is in the community Aug 22 08:56:59 and guess who that is Aug 22 08:57:11 Stskeeps: and I think people get their hats confused Aug 22 08:57:18 i'm not saying there is anything wrong with this Aug 22 08:57:35 it just ... is Aug 22 08:57:36 lbt: yes and no - MeeGo seems to center around the feature being assigned to a person who then makes sure there's resources to get things done Aug 22 08:57:56 lbt: if nvidia wants to sponsor build capacity cos meego IT says they need that for non-SSSE3, that should be possible Aug 22 08:58:00 I agreee - but the success (and requirements) of meego depends on openness too Aug 22 08:58:33 if someone wants to donate extra build space for community/everyone to work on non-SSSE3, that should be possible too Aug 22 08:58:33 when you say "you're either with us or against us" you are saying "you're not welcome in this community unless you use meego the way we tell you" Aug 22 08:58:38 but nothing in life is free Aug 22 08:58:52 I think the problem I see is that when they submit a patch that prevents that being used elsewhere Aug 22 08:59:01 "all patches should be able to go upstream" Aug 22 08:59:09 "no meego specific patches" Aug 22 08:59:19 and ffs ... prjconf? Aug 22 08:59:22 Stskeeps: hey, could mxr.meego.com be added to the topic? :) Aug 22 08:59:51 lbt: no meego specific patches is impossible and i haven't seen that policy.. for kernel it makes sense Aug 22 09:00:30 timeless_mbp: too full already, get it linked on wiki or meego.com somehow Aug 22 09:00:30 I thought it was a goal Aug 22 09:00:35 not a constraint Aug 22 09:00:49 in principle we should have a minimal patch load Aug 22 09:01:27 one of the metrics I'll be looking at will be patch load Aug 22 09:02:04 :nod: Aug 22 09:03:24 I wonder if the Qt patches have been cleaned up Aug 22 09:03:35 there were some pretty odd ones in there when I looked shortly after the initial release Aug 22 09:07:11 lbt: perhaps it's time to ask if there's an actual market and vendors interested in non-SSSE3 and why Aug 22 09:08:01 my original arguement was that non-ssse3 was about ensuring minimum barrier to entry for community Aug 22 09:08:08 do AMD even make mobile chips? Aug 22 09:08:33 not about commercial viability Aug 22 09:08:54 lbt: right Aug 22 09:09:09 * Stskeeps ponders Aug 22 09:09:19 so, nvidia ion isn't a problem cos that's an atom processor Aug 22 09:09:44 is there list of supported chipsets ? Aug 22 09:10:04 bbiab Aug 22 09:10:36 via nano has SSSE3.. Aug 22 09:11:42 how much ram do meego OBS worked need each ? Aug 22 09:11:42 imo other issue is supported (intel) graphics, or are all supported. Aug 22 09:12:20 odin_: i ran 2 workers on 2GB of ram and 512MB swap partitions each Aug 22 09:12:26 s/worked/workers/ i.e. are there some packages needing gobs of ram ? Aug 22 09:12:39 i compiled gcc OK in that Aug 22 09:15:47 I have 3 x dual quad core HP DL360/380 in data centre to attack building of i686/n8x0, run stuff but 2 are mostly idle, if I can get workers down to 1Gb ram each I can prolly get a farm of 6 up somehow Aug 22 09:16:22 Stskeeps: anyway, the ssse3 issue is not one of selling devices without ssse3, it's the fact that my development machine can run ubuntu, fedora, windows, android, even meego arm - all at the same time. but not meego x86 Aug 22 09:16:55 ali1234: the question is however how generic the circumstances is - i mean, i don't really compile platform stuf locally Aug 22 09:17:04 you get used to working straight with obs Aug 22 09:17:37 OBS is no use if i can't run the binaries in gdb on a decent fast machine Aug 22 09:18:53 hmm, how long does meego usually take to boot up Aug 22 09:19:09 what do you see on screen? Aug 22 09:19:19 maybe intel can provide kernel assisted SSSE3 support ? there is no reason why the General Protection Fault can't be handed off the qemu-ssse3 to do the work and then control passed back to the application Aug 22 09:19:23 a _ Aug 22 09:19:33 which just scrolled down Aug 22 09:19:43 X is starting i'd say Aug 22 09:20:16 started, this takes time... Aug 22 09:20:23 to be honest with, if you know about Linux signal handling you can trap the SIGILL yourself in the application and emulate the SSSE3 instructions :) Aug 22 09:20:23 odin_: yes the instructions can be emulated in kernel Aug 22 09:20:36 or in a library Aug 22 09:20:54 Stskeeps, is xterm broken on meego atm? Aug 22 09:21:00 i don't really care as long as the x86 SDK includes such a library to make it work Aug 22 09:21:40 MohammadAG51: possible Aug 22 09:21:59 ali1234: i'm looking at sseplus by AMD Aug 22 09:22:30 so anyone know of an open source SSSE3 implementation ? I could probably explain how to whip it into a signal handler Aug 22 09:23:12 Stskeeps, and I'm assuming it has an ssh server installed? Aug 22 09:23:40 MohammadAG51: right Aug 22 09:24:06 odin_: http://developer.amd.com/cpu/Libraries/sseplus/Pages/default.aspx might be a start Aug 22 09:25:38 odin_: could it be done by compiler? http://gcc.gnu.org/onlinedocs/gcc/i386-and-x86_002d64-Options.html Aug 22 09:26:06 if someone wants to package up a bunch of functions implementing SSSE3, like hand optimized assembler, I will both confirm SIGILL works like SIGSEGV and do you a proof of concept which you can polish up Aug 22 09:26:28 there is already a kernel patch that can do this Aug 22 09:26:29 yes a mix of compiler and assembler, some *.S files and *.c files Aug 22 09:27:12 http://lkml.indiana.edu/hypermail/linux/kernel/0206.3/0631.html Aug 22 09:27:24 I don't know, only know that it is possible from both kernel and user-space Aug 22 09:28:25 can you look at GCC? see what code would get optimized into SSSE3 calls, then write it, and compile it with/without -mssse3 Aug 22 09:28:32 that link relates to linux 2.4 from 2002 ? Aug 22 09:28:47 odin_: yeah it emulates 486 instructions on 386 Aug 22 09:28:47 ali1234, yes yes ^^^ this I have put the question to gcc-list and intend to spend time on Aug 22 09:30:50 given the results of scanning for ssse3, it looks like gcc never generates any of it anyway Aug 22 09:30:59 ali1234, I need to research and understand the responses and also the background reading about how other platforms do it.. I noticed recently binaries have an ELF section .note.ABI-tag and also want to understand this purpose Aug 22 09:31:16 there is technically no reason why a kernel couldn't catch SSSE3 instructions. Aug 22 09:31:39 Stskeeps: the problem isn't catching them, it's what to do after you caught one Aug 22 09:31:44 yes yes this is my belief too and I think the intel developers on the linux are trying to defend -mtune=atom for performance and are getting confused with -march=ssse3 Aug 22 09:32:03 i don't think they are confused Aug 22 09:32:18 odin_: i don't think they're confused, there's a benefit to ssse3. Aug 22 09:32:22 nor do I, more like selective obfuscation of the problem Aug 22 09:32:25 i don't normally attribute to malice what can be explained by stupidity, but intel guys are smarter than that Aug 22 09:32:35 which unfortunately only leaves malice :( Aug 22 09:32:44 or choice Aug 22 09:33:08 what applications modify the SIGILL handler ? I bet almost none, so it maybe possible to effect a solution in userspace first Aug 22 09:33:09 linux needs good hw supporty. Aug 22 09:33:13 yes, another choice which cannot be explained given the existence of other options Aug 22 09:34:07 which can be setup by a modified dynamic linker for every executable, this would not quite be as invisible as a kernel based one but would allow rapid development and testing to then seek to move it to the kernel Aug 22 09:36:25 the objection here is simply to interoperability problems due to the quick and dirty method to enforce SSSE3, I still regard the desire for maximum optimization to be a worthy goal and Intel just need guiding in a better direction Aug 22 09:37:23 no, it's clear they have drawn a line in the sand and will not discuss the issue in a sensible way Aug 22 09:37:37 then we make it damage their pockets somewhat Aug 22 09:37:51 control the SDK Aug 22 09:38:08 um... what do you mean? Aug 22 09:38:13 control the media Aug 22 09:38:37 glwt Aug 22 09:38:45 now, who's the malice using ones here? Aug 22 09:38:45 :P Aug 22 09:38:59 you're crazier than me if you think a bunch of nerds have more control over the media than intel Aug 22 09:39:20 there is nothing stopping rolling a fixed SDK in a separate OBS, but this SDK will work for everyone and getting developers to use that and not intels one Aug 22 09:39:43 since it is the developers complaining they cant run non-SSSE3 to develop Aug 22 09:39:46 there is one thing: they can just say "you can't call it meego" and then nobody will ever hear about it Aug 22 09:40:15 great, run the media on that one Aug 22 09:40:41 i can't see it happening Aug 22 09:40:53 understood Aug 22 09:41:10 on the other hand, VIA nano benefits a lot from this Aug 22 09:41:19 they have ssse3 Aug 22 09:41:31 i want meego to use ssse3 on the actual device Aug 22 09:41:38 everybody does Aug 22 09:41:46 no point having that silicon doing nothing Aug 22 09:41:52 agreed, no one is saying don't have SSSE3 optimization capability Aug 22 09:41:54 can i ask what kind of developer you are? Aug 22 09:42:09 i'm the kind of developer who fixes bugs in other people's stuff Aug 22 09:42:13 morning... can someone help me figure out why certain keystrokes are not seen by X but work fine in console (after being dealt with setkeycodes)? Aug 22 09:42:59 maybe intel just have short them interest and in long run proprietary solutions are not so affordable. Aug 22 09:44:03 after the market has saturated Aug 22 09:44:07 anyhows we are getting closer to the answer to the question that is still not proven, "Can you measure SSSE3 optimization speed ups in General Purpose Code?" Aug 22 09:44:11 ali1234: ok, so, you'd actually be happy enough with a platform SDK hack that emulates SSSE3? Aug 22 09:44:19 Stskeeps: totally, yeah Aug 22 09:44:36 as long as it doesnt make meego slower on my 3.2ghz workstation, than it is on my netbook Aug 22 09:44:47 what happens when qemu-kvm encounters ssse3? Aug 22 09:45:20 afaik it raises a SIGILL inside the vm? Aug 22 09:45:31 does it technically have to? Aug 22 09:45:44 no idea Aug 22 09:45:51 i mean, virtualization gets SIGILL quite often, from some things Aug 22 09:46:03 qemu can hide host extensions... no reason it can't add them too i guess Aug 22 09:46:34 i'd rather the "fix" didn't use kvm though. i like to run virtualbox, and kvm is incompatible with it, afaict Aug 22 09:49:08 although virtualbox is probably dead now thanks to oracle... Aug 22 09:50:21 i wonder if you can do qemu-kvm -cpu host,+ssse3 Aug 22 09:50:54 only if host cpu supports it, currently Aug 22 09:51:04 SSEplus is Apache v2 license, I wonder if that is compatible with GPL v2 Aug 22 09:51:25 ali1234: sure? Aug 22 09:51:40 http://comments.gmane.org/gmane.comp.emulators.qemu/72324 Aug 22 09:52:04 seems to imply it will check the host cpu for that feature, and die if it is not found Aug 22 09:52:09 mm Aug 22 09:52:11 that would be funny an SSEplus library that AMD optimized (I presume for AMD devices) being used to provide developer happiness with Intel MeeGo Aug 22 09:52:14 or in that case, die even if it is found :) Aug 22 09:53:09 odin_: surely they made that lib exactly for this purpose? Aug 22 09:53:52 I have not read all the details its a userspace library to provide proof of concept and assist the development process AFAIKS Aug 22 09:55:34 this whole mess makes me want to move away from x86 entirely :( Aug 22 09:58:23 good luck with that one, I'm more for the binutils/gcc approach of marking executables/dsos with ISA requirements, this I see is future proof and provides me with the goal (of getting an immediate error back from the dynamic-linker when attempting to run a program that doesn't really work on my host system) Aug 22 09:58:30 compared to arm development x86 is a breeze Aug 22 09:59:10 sx0n|home: i haven't had that problem, what makes you say it? Aug 22 09:59:27 i don't like sb1 so much Aug 22 09:59:42 well that problem is caused by you using x86 Aug 22 09:59:48 not by arm Aug 22 10:00:02 if you were developing native you wouldn't need sb Aug 22 10:00:24 yep, i surfed arm machines but there was not really good ones in the market Aug 22 10:00:34 for compilation purposes. Aug 22 10:00:38 yeah, this is the problem Aug 22 10:01:56 multiprocessor ARM rack :P Aug 22 10:03:18 risc machines cost cows. Aug 22 10:03:23 I believe some vendor is looking at ARM for the server market now, since they are at the 1.7GHz barrier to entry now, so within 2 years I expect to see rack servers available (reminds me of MIPS based Cobalt Raq2 circa 2001) Aug 22 10:03:46 they dont even sell those for green money :) Aug 22 10:03:55 sx0n|home: ARM development is a breeze with OBS. Aug 22 10:04:18 anyone know how is cross-compile with sysroot looking ? Aug 22 10:04:32 Stskeeps: so i've been told. Aug 22 10:04:39 ali1234: so, in practice, you'd be happy with a SDK that was built for non-SSSE3? Aug 22 10:05:06 but that targetted ssse3 Aug 22 10:05:15 yes Aug 22 10:05:24 SDK+core (i.e. what goes into download images) which can target BOTH non-ssse3 and ssse3 Aug 22 10:06:00 ali1234: doesn't this pose a problem if you have applications that use SSSE3 and those binaries are used as intermediates? Aug 22 10:06:07 in build process Aug 22 10:06:17 Stskeeps: not my problem Aug 22 10:06:27 so the next question is i686 ok ? or does it need to go lower ? Aug 22 10:06:32 Stskeeps: it has to work on my development machine, or i won't use it Aug 22 10:07:09 Stskeeps: and it doesn't have to target ssse3... it has to target arm Aug 22 10:07:14 presumably the SDK can uses its own repo, which contains a full set of SDK+core RPMs Aug 22 10:07:34 ali1234: ah, i see the kicker now Aug 22 10:09:56 ali1234: ok, so, 'arm platform sdk' would be a chroot full of arm binaries.. Aug 22 10:10:01 not requiring atom Aug 22 10:10:13 that's when looking away from what OBS does Aug 22 10:10:45 plus an arm cross compiler not requiring atom, plus xephyr not requiring atom, ... Aug 22 10:11:21 ali1234: i think that's solve-able. Aug 22 10:11:31 Stskeeps: all this stuff is "solvable" Aug 22 10:11:38 as in, organisationally Aug 22 10:11:51 it's only solvable if intel actually wants to solve it Aug 22 10:11:59 in this particular case, no Aug 22 10:12:02 nobody else can solve it and still call the result meego Aug 22 10:12:41 we have a different set of glibc for cross compiler Aug 22 10:12:48 which is maintained by LF guys.. Aug 22 10:13:25 the question then becomes, if you can supply this stuff for arm, why not for x86 too? Aug 22 10:13:44 assuming you can come up with qemu that emulatees ssse3 Aug 22 10:13:46 x86 isn't a cross situation Aug 22 10:13:57 well, if you have qemu that emulates ssse3, a lot of things is easier Aug 22 10:14:14 this is why I asked about labelling at the TSG in the compliance space Aug 22 10:14:33 MeeGo (arm), MeeGo (x86), MeeGo (Atom) Aug 22 10:14:35 re what is "meego" and what is "not meego" ? Aug 22 10:14:46 odin_: see the compliance discussion Aug 22 10:19:16 is meego embedded or generic desktop project? Aug 22 10:19:29 yes Aug 22 10:19:48 can somebody help me to run a .NET application on meego? I'm getting a System.Windows.Forms.dll FileNotFoundException and i can't seem to find how to install it Aug 22 10:20:11 Ian--: check the wiki... it's on there Aug 22 10:21:39 How come there is no email address on the MeeGo website? I need to contact them because I was unable to register an account Aug 22 10:22:33 impje: that sounds like an oversight - and you probably can't file a bug without registering... Aug 22 10:23:02 lbt: yeah, my email address was added to the database, but I didn't receive a password or anything Aug 22 10:23:13 spam filtered? Aug 22 10:23:30 I checked but found nothing Aug 22 10:23:32 on the grounds that most people don't have a problem :) Aug 22 10:23:47 you can't register again or login at all ? Aug 22 10:23:50 try again from a gmail-type account? Aug 22 10:23:56 i'll file a bug if nobody else is going to :) Aug 22 10:24:07 about? Aug 22 10:24:22 there is no contact email on the website Aug 22 10:24:25 a Aug 22 10:24:27