**** BEGIN LOGGING AT Sun Jul 11 02:59:57 2010 Jul 11 05:10:09 MeeGo 1.0 Netbook Update "a lot of translation update" ... Jul 11 05:22:05 is meego's source available fully? Jul 11 05:22:34 I mean, is it absolutely and completely open? (unlike android, which is only partially open) Jul 11 05:23:08 There is a separate repo for non-oss so i suppose it is Jul 11 05:24:03 comradekingu: what does non-oss contain? Jul 11 05:24:18 non-open source software, presumably Jul 11 05:24:27 can I have the link to such repo? Jul 11 05:25:17 well, even debian has such a thing, question is how much does the actual installation pull in Jul 11 05:26:24 I just want a truly open platform for my phone. Android doesn't cut it, it's too closed for my taste! Jul 11 05:26:49 so much that I have to rely on rumors for what comes in the next version. Jul 11 05:27:53 fengshaun: My suggestion would be debian Jul 11 05:28:10 put debian on my phone? Jul 11 05:29:45 fengshaun: make of it what you will Jul 11 05:30:13 hmmm I haven't thought about it, but I'll look into that for sure. Thanks. Jul 11 05:35:09 fengshaun: http://www.gnewsense.org/index.php?n=FAQ.FAQ Jul 11 05:35:37 gNewSense 3.0 will be based on Debian, so thats that Jul 11 05:36:33 comradekingu: cool! thanks a lot! Although that kind of 'free' is hardcore! Jul 11 05:38:53 fengshaun: Short of running GNU/hurd I suppose it is taking it to the extreme Jul 11 05:50:31 so, what parts of meego are still closed? Jul 11 06:30:06 the pieces that vendors choose not to open? Jul 11 08:38:11 hi all Jul 11 09:09:19 Stskeeps, in relation to your recent thread meego-dev, part of the points raised (by others) was over the "shared infrastructure" i.e. that used under the and in the name of meego.com branding Jul 11 09:10:35 the point made by intel representatives was that there is a cost to adding an architecture to the shared infrastructure and that intel/nokia shouldn't be expected to fund unlimited expansion of meego in this way Jul 11 09:12:01 this is understandable, but it is not clear how a body/corporation/group-of-people would get on equal footing under the meego branding if there is to be a situation that the shared infrastructure can only be used to further the product lines Intel/Nokia directly support Jul 11 09:14:15 for example as soon as "shared infrastructure" it comes online people might wish to add i686 support (i.e. for systems without SSSE3) and also n8x0 (and even n900) support Jul 11 09:14:56 while this support is not directly supported by intel/nokia it does further the cause of MeeGo brand in a good way and intel/nokia is not expected to directly put any man-hours into it Jul 11 09:15:32 however there is a cost to hosting/building/distributing it, so what is not clear to me is how that works Jul 11 09:16:33 i'm going to need more coffee before discussing :) Jul 11 09:17:27 I hope soon for clarification, since a misunderstaning could result in a fork and a separate fund-raising initiative to obtain "independent shared infrastructure" that would truly be the communities, along the lines of debian/centos Jul 11 09:18:35 other archs even on debian need donations of boards, i don't see why intel would have to donate MIPS boards.. as long as there's a possibility to get extra infrastructure bought, i don't see a problem Jul 11 09:18:44 such an effort might not carry the MeeGo branding (due to unknown legal issues) but a more generic "open distribution for linux on mobile/netbook/embedded devices", which would not be a good thing, as it would further fragment efforts Jul 11 09:18:50 sure on the more coffee! Jul 11 09:19:20 and keep in mind that it isn't only for furthering intel and nokia efforts, it is exclusively about the platforms supported! Jul 11 09:19:23 :P Jul 11 09:19:32 meego infrastructure exists to support the roadmap Jul 11 09:19:44 well this point is the most important for my own continued interest in "MeeGo" Jul 11 09:20:28 there is a big cost to build infrastructure and the LF umbrella exists for others to be able to put infra in. i think that makes sense. Jul 11 09:20:42 but I don't see anyone asking anyone to manage a MIPS platform, does that exist already in the spare MeeGo is in (portable and emebded devices) Jul 11 09:21:09 s/apre/space/ Jul 11 09:21:43 i think you're worrying too much :) Jul 11 09:22:11 yes maybe I am, but this is the first time I have seem intel representatives hold a view that is problematic to me Jul 11 09:22:31 in the narrow space of build power, they have a pretty good point Jul 11 09:22:53 especially since setting up your own OBS or maybe even donating more power to the cluster is possible Jul 11 09:23:49 even if i686 and n8x0/n900 is not supported, I do expect to see it up on the shared infrastructure, maybe not with equal footing, since I would not have a problem if intel/nokia added extra resources dedicated to their goals (so long as some amount of shared/pooled resources exist for community lead goals) at the moment there are no resources (other than communication) Jul 11 09:24:23 If this is the first time Intel has worried you, you must not have been watching what they have been doing. Jul 11 09:24:36 ok, you're writing too long sentences again - what should be run on the main meego builder is stuff that exists on the meego roadmap. Jul 11 09:24:56 Termana, what are your concerns then ? you mean Intel+Android ? or ??? Jul 11 09:24:59 the most common way to get things implemented on roadmap is saying you'll take the lead on it Jul 11 09:25:11 ie, resources, people, contributions Jul 11 09:25:35 odin_, no I have concerns over things Intel have done regarding MeeGo. I have no concern that they are also supporting Android, the more the merrier. Jul 11 09:25:44 and the community OBS is right around the corner, as well Jul 11 09:26:46 Termana, feel free to explain further, I have concerns with any non-desktop compatible choices, where backward compatibility has been lost with Linux desktop/workstation/server/whatever, i.e. optimization is not king, interoperability is king Jul 11 09:26:53 a product needs to be put out, resources are limited and additional architectures -does- slow down general process of project Jul 11 09:27:03 if additional resources are not contributed Jul 11 09:29:50 which, frankly, if you don't have tunnel vision and a general overview of the project, makes a lot of sense :) Jul 11 09:30:33 yes understood on the main goals of intel/nokia interest, to get a end user product out the door, but also, they want a community and eco-system to continue to sell into for their efforts Jul 11 09:31:31 actually, they just want a platform product out the door, like all of us - community and eco-system will appear using the community OBS and those efforts might be recognised and included into real roadmap Jul 11 09:31:53 if we can't keep deadlines or get a product out, we might as well go home and call ourselves openmoko :P Jul 11 09:32:38 odin_: let's step back a bit Jul 11 09:32:43 I personally dislike the fact that Intel has decided i386 is "not their platform". What's worse is they are setting it up so that i386 support is going to be a pain in the arse. Eg. apps will be compiled for SSSE3 platforms only. Jul 11 09:32:58 well its not even clear to me why MeeGo 1.1 will be a step up from Meamo5 PR1.2 or whatever, I'd be very happy with a straight conversion, its only other forces pushing for other changes Jul 11 09:34:20 Termana, yes I agree with the general sentiment there, if they used their own dynamic link path or ELF capabilities to achieve the goal then that would seem ok Jul 11 09:34:20 Termana: if they want to have a chance with power saving, they need to target atom somehow :P Jul 11 09:34:26 Luckily, I don't concern myself with the x86 side of MeeGo, and only care about the ARM side. :P Jul 11 09:35:33 odin_: i think the best way to see meego is as a company. the 'owner' is linux foundation, investors are intel and nokia and others. a project structure has been set up, company plans (roadmap), budgets and such. the exception is that the employees of the company can be anyone and there's a free-flow HR process where people can walk into the building and watch how people work, perhaps give a hand. these people work on roadmap and use ... Jul 11 09:35:35 I don't myself see the connection between power saving and atom and how this connects with enforcing SSSE3, the matters seem unrelated to me Jul 11 09:35:39 ... company resources for that - when there's additional projects that needs resources it needs to be approved on budget, if sufficiently large, an investor might have to be found. Jul 11 09:36:17 Stskeeps, now who's writing too long sentences now? :P hehe Jul 11 09:36:33 Termana: odin was going to preempt me in the middle of describing otherwise Jul 11 09:36:33 :P Jul 11 09:38:00 does anyone have an article on how SSSE3 optimizations apply to general compiler usage, I've looked over the insn set and can't see much that can be used for general purpose code Jul 11 09:39:13 sure there are alignment and instruction order optimizations but those are backward/binary compatible with older generations of chips Jul 11 09:40:29 SSSE3 implies a bunch of other SIMD schemes, so maybe the optimization for general purpose code generation relates to instructions in those older schemes (SSE, SSE2, SSE3, SSE3.1 or whatever they are called) Jul 11 09:40:40 i think ssse3 makes things faster in general and makes things more convincing :P Jul 11 09:40:46 especially rendering Jul 11 09:40:58 you're not "convincing" me like that! Jul 11 09:41:33 i'm not here to convince you about ssse3 Jul 11 09:41:33 :P Jul 11 09:41:53 rendering is specialized, as in Mesa/OpenGL, I'm trying to understand the ABI changes, this is the breakage I have objections about Jul 11 09:41:54 the above was so you understand why some things are like they are, or opinions Jul 11 09:42:11 (the company thing) Jul 11 09:42:52 there's general practical concerns behind things Jul 11 09:43:12 using newer CPU instructions support for optimizing specific libraries is not the problem, breaking the ABI linkage is Jul 11 09:43:30 when are they breaking abi linkage? :P Jul 11 09:43:45 having code that crashes with SIGILL Jul 11 09:43:58 you can still use i386 binaries on the meego atom platform? Jul 11 09:44:15 not sure on the answer to that Jul 11 09:44:16 i think you might be misunderstanding the problem a bit :) Jul 11 09:44:45 the SIGILL comes from people who are running SSSE3 binaries on non-SSSE3 machines Jul 11 09:44:54 normal i386 binaries work fine on meego atom platform Jul 11 09:44:56 no program should crash with SIGILL due to use of CPU specific instruction not supported by the Linux IA32 ABI Jul 11 09:45:14 ^^^ that project should not even run Jul 11 09:45:30 [citation needed] Jul 11 09:45:57 i think SIGILL is a perfectly natural response to well, an illegal instruction, being asked to run a SSSE3 instruction on a non-SSSE3 system Jul 11 09:46:01 let me see what the wording of the Linux IA32 ABI reads Jul 11 09:46:17 yes when the kernel is forced into trying to execute it Jul 11 09:46:28 just like ARMv7 binaries doesn't run on ARMv5 Jul 11 09:46:37 it is the correct response Jul 11 09:46:39 (and they share same EABI) Jul 11 09:47:16 the problem is that there is no recovery mechanism for SIGILL (in most programs) Jul 11 09:47:25 so that means potential data lossage Jul 11 09:47:42 why should there be? it's a binary optimized for a certain instruction set/processor Jul 11 09:47:53 which normally means this doesn't work on anything lower Jul 11 09:48:33 but ABIs dictate the instructions you can use Jul 11 09:48:38 not really Jul 11 09:48:50 "really" ? Jul 11 09:48:52 the moment that you got the SIGILL, you're lost Jul 11 09:48:57 [citation needed], still Jul 11 09:49:23 odin_: on arm there's stuff like thumb interworking and such Jul 11 09:49:39 but a decided ban on 'processor specific' instructions, i highly doubt Jul 11 09:49:57 no the spec is written the other way Jul 11 09:49:59 since you side stepped this topic a bit, do you now see why they wouldn't automatically add another architecture, regarding my view of meego as a company needing to make something happen? Jul 11 09:50:11 odin_: well, then find proof - i'll go make coffee Jul 11 09:50:29 as in the older CPU with fewer instructions was around the time the ABI is created Jul 11 09:50:52 when new CPU comes and new INSNs then ABI is updated or a new one created about how to support them Jul 11 09:51:01 are we trying to make atom binaries run on non-atom, or are we trying to get non-atom binaries running on atom? .. Jul 11 09:51:28 because i'm astonished you're surprised that binaries build for a newer processor doesn't run on old ones Jul 11 09:52:41 we are trying to make a non-atom not run atom binaries in the first place, where non-atom means lowest supported CPU for the Linux IA32 ABI and atom means running code that directly uses SSSE3 without guarding it Jul 11 09:53:16 ok, find citations that this is required, thank you Jul 11 09:53:17 you can't run unguarded instructions Jul 11 09:53:35 well, you could with kernel help emulating those instructions Jul 11 09:53:42 so in the case of MeeGo, a non-atom CPU installed MeeGo netbook edition, he boots up and init fails to run Jul 11 09:54:01 right, at least we're on same page now Jul 11 09:54:04 expected result, yes Jul 11 09:54:23 guarding comes with a large performance cost and that's why i doubt it's in ABI Jul 11 09:54:28 plenty of application correctly guard optimizations Jul 11 09:54:53 huh what performance cost ? Jul 11 09:54:58 in source, right? Jul 11 09:55:18 keep in mind this is the compiler spitting out ssse3 instructions, not assembly in the source :) Jul 11 09:55:27 you do CPU detection, you then modify your DSO entrypoints to point functions/symbols at the versions that work for your CPU, this means one DSO can support an unlimited number of optimizations Jul 11 09:55:34 yes, it's possible Jul 11 09:55:51 odin_: yes, but does gcc do this? :P Jul 11 09:56:01 I don't know if ld.so does it either Jul 11 09:56:14 which also brings the point, what general purpose code is spitting out SSSE3, I'd like to understand that, which instructions are we talking about and what is the i386 way of doing the same? Jul 11 09:56:24 odin_: gcc is spitting it out and any vector like things Jul 11 09:56:28 odin_: gcc does that Jul 11 09:57:02 disassemble one of the binaries and find any SSSE3 instructions Jul 11 09:58:35 "vector like things" somewhat vague, since vector work used in OpenGL is domain specific, i.e. there isn't any C language expression to ask the compiler to do something for general purpose code Jul 11 09:59:24 i don't know specifics, but gcc generates those things Jul 11 09:59:25 :P Jul 11 09:59:52 yes I'm not even sure which binaries have SSSE3 instructions in them, do you know any to run, "objdump -d ./a.out" on / Jul 11 10:00:44 we're drifting though - it will generate the SSSE3 instructions, and why is this a problem? Jul 11 10:00:47 odin_: http://lists.meego.com/pipermail/meego-dev/2010-July/003816.html Jul 11 10:01:33 not that anyone replied in a meaningful way ... Jul 11 10:01:36 all binaries potentially contain SSSE3 Jul 11 10:01:51 which ones actually do depends on the optimisation flags you used, the compiler version and the source code Jul 11 10:02:01 and a change in any of them can add or remove instructions Jul 11 10:02:33 lbt, yes have read that reply, there is no problem with motivation to provide SSSE3 optimized binaries, the issue is "the method" Jul 11 10:03:13 what method? Jul 11 10:03:37 Intel are using MeeGo.com to make "the Intel MeeGo" not "the community MeeGo" Jul 11 10:03:53 thiago_home, and the issue there is that such binaries just should not start/load/run on systems that do not support them (it does not matter how they are supported, i.e. native or emulated by kernel traps) Jul 11 10:04:22 odin_: so you're saying that this should be encoded in the ELF header and the kernel should refuse to load? Jul 11 10:04:39 lbt: Intel contributed the hardware and they want it optimised for their platform Jul 11 10:04:44 someone else can build for other platforms Jul 11 10:04:59 not a good argument Jul 11 10:05:13 it is a good argument Jul 11 10:05:19 the LF can build for a lower platform Jul 11 10:05:23 Intel won't Jul 11 10:05:26 that's not sponsorship; that's a vendor specific build platform Jul 11 10:05:34 thiago, yes (if supported by the existing ABIs), or a change in dynamic linker path (if support is not compatible) Jul 11 10:05:39 that's not sponsorship, it's their hardware Jul 11 10:05:44 ok.. then it should be build.meego.intel.com Jul 11 10:06:02 like I said, LF can set up the other builds, with money sponsored Jul 11 10:06:56 I actually thought MeeGo.com was the LF platform Jul 11 10:07:19 * thiago_home couldn't find any SSSE3 instruction in his SSE4-enabled QtCore, QtGui and QtWebKit Jul 11 10:07:55 thiago_home: that's the point ... it's "random" - or at least not humanly determinable if a given application will choke Jul 11 10:08:01 yes, it is Jul 11 10:08:08 ^^^ and that also is part of the problem, you don't know which do and which don't and most don't, so is the breaking backward compatibility cost worth this much to intel ? Jul 11 10:08:10 in the ICC build I did find pshufb Jul 11 10:08:16 now palignr Jul 11 10:08:21 and given the entirety of Qt doesn't use it then I challeng the "performance" it offers Jul 11 10:08:38 lbt, here here me too Jul 11 10:08:59 is there an ICC OBS platform ? Jul 11 10:09:07 a total of 4 SSSE3 instructions on QtGui Jul 11 10:09:13 do intel intend to use ICC to distribute "Intel MeeGo on a device" Jul 11 10:09:13 lbt: on the other hand 'optimized for atom' sells Jul 11 10:09:20 no, forget ICC Jul 11 10:09:25 Stskeeps: and that's my point... Jul 11 10:09:28 I'm just looking at it because I do have it Jul 11 10:09:29 (as per the mail) Jul 11 10:09:45 lbt: and attracts developers to the platform and set of development hardware, etc.. Jul 11 10:09:48 :P Jul 11 10:09:52 it can be alignment and instruction order optimized for atom and still be called "optimized for atom" Jul 11 10:10:12 A little offtopic: could somebody tell me how to configure ssh client to avoid dropping connections? Jul 11 10:10:26 RST38h, dropping in what way ? you mean when idle over time ? Jul 11 10:10:28 that's -mtune=atom Jul 11 10:10:44 I still don't get what the problem is Jul 11 10:10:49 odin: I think so Jul 11 10:10:53 why hasn't anyone built for generic i586 yet? Jul 11 10:11:01 thiago_home: how? Jul 11 10:11:11 aren't the sources available? Jul 11 10:11:21 thiago: Maybe nobody needs it, really? Jul 11 10:11:26 and isn't that the job of MeeGo.com? Jul 11 10:11:34 I mean, there is Ubuntu Linux =) Jul 11 10:11:41 RST38h, -2 -o TCPKeepAlive=yes -o ServerAliveInterval=900 Jul 11 10:11:57 odin: ACK. Jul 11 10:12:04 RST38h, man ssh_config and sshd_config (for "Alive") Jul 11 10:12:09 lbt: maybe it is. Have you contacted the LF people about this? Jul 11 10:12:13 RST38h: and that's the point "oh, I need that HW to run it. Ubuntu runs on my laptop. Bye MeeGo" Jul 11 10:12:30 thiago_home: not directly Jul 11 10:12:35 lbt: has anyone? Jul 11 10:12:41 although dl9pf is LF iirc Jul 11 10:13:02 lbt: Believe me, when reaching that point, I do not even consider the hardware I need Jul 11 10:13:09 thiago_home: should we need to thiago_home? Jul 11 10:13:19 isn't meego infrastructure for roadmapped activities? :P Jul 11 10:13:22 lbt: yeah Jul 11 10:13:26 odin; That is the weird part: ssh_config does not appear to have these options Jul 11 10:13:31 don't assume people thought of those issues Jul 11 10:13:34 bring it up Jul 11 10:14:09 make the case as to why this would be necessary, and work together with them to bring up the solution Jul 11 10:14:17 an OBS to build for generic i586 Jul 11 10:14:27 I started writing 15 mins ago :) Jul 11 10:14:28 RST38h, what distro ? they have not always been there but have been over the past 3 years of OpenSSH releases Jul 11 10:14:34 LF might have to search for a sponsor for the machines and storage Jul 11 10:14:50 thiago_home: although obviously the community OBS is a potential place for this Jul 11 10:14:58 and it's on my todo list Jul 11 10:15:08 and isn't that something that LF can/will host? Jul 11 10:15:24 we've been promised some HW for mid-june Jul 11 10:15:30 thiago_home no generic i586 as there is no comminity OBS and distribution point, but the recent meego-dev mailing list topics have indicated that such things might not be supported by Intel/Nokia community platform anyway Jul 11 10:15:32 * lbt checks his watch Jul 11 10:15:51 odin_: what do you mean? Jul 11 10:15:51 odin: Ubuntu Jul 11 10:15:52 odin_: there is a community OBS Jul 11 10:15:59 Anyway, added them, let me reconnect... Jul 11 10:16:13 RST38h: stop trolling; you got ssh support... be nice Jul 11 10:16:21 thiago_home, which raises the question, should the community setup a desperate disto (maybe not called MeeGo) and raise its own funds (like debian/CentOS) for its own goals on that Jul 11 10:16:28 lbt: But, the thing is, I am not trolling Jul 11 10:16:55 odin_: I'd rather MeeGo.com became the community distro Jul 11 10:16:58 lbt: I am pointing out a legitimate reason why Meego is not very popular on laptops (netbooks included) Jul 11 10:17:05 odin_: you lost me at the previous line Jul 11 10:17:05 thiago_home, the point has been of the "cost" of the shared infrastructure and Intel employees having said that Intel should not be burdened with the cost of supporting other arch/platforms/builds Jul 11 10:17:08 odin_: what do you mean? Jul 11 10:17:23 lbt: There is well established competition that has everything set up and working Jul 11 10:17:31 RST38h: once is enough... I know it. I *work* for Nokia and AFAIK can't run MeeGo. I've actually never seen it running :) Jul 11 10:18:00 lbt: Oh, I see it running daily on platforms that have not see silicon yet, but still ... Jul 11 10:18:02 thiago_home, worst case scenario I paint is that these resources can not be used for i386/i686 builds not for n8x0/n900 builds Jul 11 10:19:20 lbt, yes I'd rather MeeGo was a community distro, but I am just trying to understand the commercial aspects of the "shared resources" and who gets to dictate what is and is not on the agenda, since I think that my idea of "community" is not the same as "Intel's idea of community" Jul 11 10:19:57 the model is that Nokia (as a vendor) has an internal OBS to optimise for a commercial product Jul 11 10:20:00 it never is... Jul 11 10:20:14 Why does Intel get to optimise on the MeeGo.com ? Jul 11 10:20:28 lbt, so Intel should have the same, a private OBS for their own R&D and optimizations and stuff Jul 11 10:20:37 odin: The idea is that there is a distro and a bunch of manufacturers who take it and customize it to their needs Jul 11 10:20:52 lbt: and why can't the externals get the results of Nokia's build? Jul 11 10:20:53 odin: The distro is largely x86-oriented Jul 11 10:20:58 lbt, I do not have an issue with intel optimizing on MeeGo.com, I have an issue with them blocking the potential for an i386/i686 (non -SSSE3) distro from also being put on there Jul 11 10:21:19 odin: Or, Atom-oriented to be exact. Although this may change in the future to embrace other x86 platforms Jul 11 10:21:23 thiago_home: they can Jul 11 10:21:35 they buy Nokia product and get access Jul 11 10:21:53 so I can't put it on my other device? Jul 11 10:22:02 license permitting Jul 11 10:22:11 odin: As far as your problem goes, I have heard that you only need to recompile the kernel without SSSE3 quirks Jul 11 10:22:16 MeeGo is designed to allow commercialisation Jul 11 10:22:19 we know that Jul 11 10:22:38 but MeeGo.com is supposed to be open and to minimise barrier to entry Jul 11 10:22:47 lbt: But it does not prevent anyone from making their own fully totally community distro based on Meego Jul 11 10:22:48 from my point of view, the Atom-optimised build is Intel's OBS that is just in the public Jul 11 10:22:59 RST38h: correct Jul 11 10:23:03 should there be a non-Atom optimised build? Maybe, who would use it? Jul 11 10:23:07 RST38h, I don't think the kernel is incompatible, I hear its a userspace problem, this is because more than a handful of people have reported putting it on their non-SSSE3 notebook and it appears to work, but something keeps crashing Jul 11 10:23:15 thiago_home: anyone with a non-Atom device Jul 11 10:23:24 eg my AMD desktop Jul 11 10:23:32 RST38h, something, as in just this one thing, and they want support on why Jul 11 10:23:37 why are you running MeeGo on that? Jul 11 10:23:39 is it a netbook? Jul 11 10:23:52 thiago_home: I can't run MeeGo on it Jul 11 10:23:56 that's the point Jul 11 10:23:59 what is a netbook, is that a marketing term? be more scientific in the description Jul 11 10:24:06 or, to be fair, I don't think I can Jul 11 10:24:19 why are you trying to? Is it a netbook? Jul 11 10:24:28 odin: I have heard from some OTC people here (correct me if I am wrong) that SSSE3 gives you +20% performance on certain tasks Jul 11 10:24:29 I completely agree that the SDK must run on your workstation Jul 11 10:24:42 thiago_home: As Arjan says ... we should be OK running MeeGo as our desktop for daily life Jul 11 10:24:46 eat dogfood Jul 11 10:24:55 odin: And because they target Atom based devices anyway, they can say that SSSE3 will be there Jul 11 10:25:04 we are making a product for which we are not the target audience Jul 11 10:25:11 thiago_home: again, AFAIK, the SDK won't run on the desktop... Jul 11 10:25:16 yes, I should be able to run MeeGo on my netbook Jul 11 10:25:18 some will but the imager fails Jul 11 10:25:21 but I don't want to run it on my workstation Jul 11 10:25:28 thiago: As long as THERE IS a target audience for it, and you are being paid, I see no problem Jul 11 10:25:28 thiago_home: want != should be able to Jul 11 10:25:29 lbt: it should. The SDK should run everywhere. Jul 11 10:25:32 RST38h, yes, sure no one is claiming performance of "certain tasks" is not boosted by it, but I don't think those tasks include "general purpose compiler code generation" I think they only cover domain specific uses, OpenGL, encryption, codecs, etc... Jul 11 10:25:37 thiago_home: I know it should Jul 11 10:25:37 thiago: Actually, even #1 is optional :) Jul 11 10:25:57 I ran workshops in Nokia and devs with old laptops couldn't make images for netbooks Jul 11 10:26:11 lbt: you should be able to run the image on a netbook. Jul 11 10:26:15 odin: Sorry, but what do you want to use Meego for, anyway? Jul 11 10:26:17 so our build/image/test cycle failed Jul 11 10:26:33 thiago_home: couldn't *make* image ... SDK failed Jul 11 10:27:10 how about convincing Nokia to use that existing OBS infrastructure then? Jul 11 10:27:22 mmm ? Jul 11 10:27:25 they do Jul 11 10:27:27 RST38h, I want to use MeeGo on my portable and embeded devices in the future, I am looking for a platform that is similar to desktop/server linux but which allows me to roll forward any "capital" I create/invest forward into the next device, since the portable device is a somewhat disposable market with new devices every few years Jul 11 10:27:31 for building generic i586 Jul 11 10:27:37 ah... Jul 11 10:27:43 anyway, I'm not seeing anyone block a generic build Jul 11 10:27:49 but I'm not seeing anyone step up to do it either Jul 11 10:28:15 I guess the point is that MeeGo should be doing it Jul 11 10:28:16 odin: Ok. Get an Ubuntu desktop. Get Qt. Develop apps with it. Jul 11 10:28:20 not Intel Jul 11 10:28:22 MeeGo Jul 11 10:28:27 I agree Jul 11 10:28:33 :) Jul 11 10:28:49 so let's talk to LF Jul 11 10:28:53 odin: Absolutely no reason to do it on Meego directly, at least not right now. Jul 11 10:29:21 thiago_home: I'm putting an email together Jul 11 10:29:24 odin: And yes, I also believe that the Meego Developer Edition images should run on any x86, not just on SSSE3-equipped ones. Jul 11 10:29:41 But the key is, you do not need Meego to do development. Jul 11 10:30:13 RST38h, but its not an "app" I am developing, its non-braindead operating system components, i.e. stuff my vendor did badly Jul 11 10:31:17 odin: Ok. So, let me clarify: you are developing drivers for your vendor who wants you to develop them in Meego but has not supplied you with a proper hardware? Jul 11 10:31:26 RST38h, so Ubuntu support for N900 is good and its usable as a phone still ? Jul 11 10:31:52 odin: No. But you are not talking about the N900, right? You are talking about your x86 laptop? Jul 11 10:32:15 Otherwise, the SSSE3 issue would not come up at all, no? Jul 11 10:32:59 RST38h, nope I am not working for any other vested interest but myself, I am not a subcontractor, just a user tied of previous mobile device eco-systems and have been waiting for Linux on mobile to arrive since my original Nokia communicator in 1999 Jul 11 10:33:21 RST38h, no I have no current use for MeeGo on Intel platform Jul 11 10:33:37 odin: Then why are you complaining about the SSSE3 reliance? Jul 11 10:35:30 because in future I know I will have a device when the chip/device arrives, because providing free support in this IRC channel over the same issue is not good for MeeGo, because the method by which SSSE3 is being supported will definitely affect me in the future, because there are better ways to support SSSE3 Jul 11 10:35:58 any other questions you seem to have a lot about my personal view (rather than about the topic) Jul 11 10:36:08 the best way to use SSSE3 is to have a build dedicated for it Jul 11 10:36:09 Mmmm Jul 11 10:36:23 odin: Are you sure you are going to get a device without SSSE3 on which you would like to run Meego? Jul 11 10:36:29 runtime dispatch is a performance overhead Jul 11 10:36:45 thiago; I think the point is that every upcoming device supposed to run Meego will have SSSE3 anyway Jul 11 10:37:12 thiago: And the backlash we see is from a bunch of people who would like to run Meego on their current, previous generation devices Jul 11 10:37:23 RST38h, I don't view you question a relevant to the topic, on the basis that if the matter was tackled better there would be no issue either way Jul 11 10:37:24 I don't see the distinction Jul 11 10:37:32 yes, all new devices are newer than old devices Jul 11 10:37:35 :-) Jul 11 10:37:53 thiago: Distinction is that the older devices are not the target platforms for running Meego Jul 11 10:38:24 Ok, DCC file selection works, at least Jul 11 10:38:27 RST38h, ahem, " the older devices are not _INTEL's_ target platforms for running Meego" - there fixed for you Jul 11 10:38:39 odin: Agreed Jul 11 10:38:54 odin: But given that Meego is a rebranded Moblin, I do not see a problem there Jul 11 10:39:10 no Meego is rebranded Maemo !!! Jul 11 10:39:18 odin: Yes, as a Linux distributive, it has a target audience Jul 11 10:39:23 odin: Not really, sorry. Jul 11 10:42:17 I'd be happy if MeeGo 1.1 was just Maemo5 packaged and rebuilt, new SDK, new packaging, a straight conversion, maybe one of the matters Nokia overlooks as to why their community is their tear up and change direction approach to the OS part Jul 11 10:42:58 and replacing some parts too with new technology Jul 11 10:42:59 since there is no invested capital to rollover into the next product since based on part performance that capital will not be relevant Jul 11 10:43:17 isn't this a matter of point of view? Jul 11 10:43:33 what does it matter if you start with A or B if the result is the same? Jul 11 10:43:50 well aren't there exactly 2 Symbian platforms to develop for ? Jul 11 10:44:16 within each of these there are "feature sets" and backwardly compatible and additive stuff Jul 11 10:44:57 exactly 2? Jul 11 10:45:14 try 20 Jul 11 10:45:16 or 60 Jul 11 10:45:18 the Linux offering has not been additive so far, so developers don't accumulate capital they accumulate headaches Jul 11 10:45:51 thiago: Well, to be exact, the result will not be the same Jul 11 10:45:57 S60 and S^3 Jul 11 10:46:10 thiago: But I do not see how starting with Moblin is worse than starting with Maemo, frankly Jul 11 10:46:17 S60 subdivides into a hundred different and incompatible versions Jul 11 10:46:22 Both have enough wrinkles Jul 11 10:46:41 Mmm...no (on S60) Jul 11 10:46:56 There is Symbian. Then, there are UI frameworks on top of Symbian. Jul 11 10:47:09 One of them is Avkon, also known as S60 to the customers. Jul 11 10:47:19 Another was Uikon, known to you as UIQ Jul 11 10:47:39 but there are still enough differences in the platform, like operator-specific API Jul 11 10:47:45 The "bare" Symbian also comes with its own UI framework, called Eikon and used to patch all the unimplemented holes in S60/UIQ Jul 11 10:48:00 thiago: yes, that too, but few people use operator specific crap Jul 11 10:48:26 Now, though, Symbian itself (not the UI) went through a few drastic changes Jul 11 10:48:54 For example, before Symbian9 (aka S60e3 in Nokia terms), you could not have executables, only DLLs that the OS ran in a specific manner Jul 11 10:49:19 So, no static data, and other "niceties" Jul 11 10:49:46 And Nokia also had two other UI frameworks (S80 and S90) running on top of Symbian Jul 11 10:50:25 are those S^1 and ^2 and ^3 ? Jul 11 10:50:32 And S60 itself gets constantly extended, hence S60e1 (Nokia 7650/3650), S60e2, S60e3 (E50/E70/E61), S60e5 (touch version) Jul 11 10:50:57 The powered version is a rebrand they have done after spinning Symbian into a separate non-profit entity and open sourcing it Jul 11 10:51:23 You can consider them a continuation of the S60eX labeling Jul 11 10:51:33 odin_: S^1 = S60 5.0 Jul 11 10:51:46 S^2 = 5.1, rebranded Jul 11 10:51:53 I have written for Symbian before (over 4 years ago now, something I would never go and do again), been waiting for a viable Linux based phone for 10 years now and the N900 was finally a product that ticked all the boxes, ah, Nokia finally gets it! Jul 11 10:51:55 Notice that UIQ (Uikon) has not survived the reorganization and died a quiet, unmourned death Jul 11 10:52:16 Which is kinda sad, given that Uikon is way better structured than Avkon Jul 11 10:52:22 S^3 is also the last edition with Avkon Jul 11 10:52:26 Just never debugged to a fully usable state Jul 11 10:53:03 thiago: So, they just implemnented Avkon in Qt :) Jul 11 10:53:19 thiago: ith the dreaded MVC model and all Jul 11 10:53:22 s/ith/with Jul 11 10:53:50 the new lib is nowhere that bad Jul 11 10:54:13 thiago: But do you still need to write 5 classes just to show Hello World? Jul 11 10:54:58 thiago: or can you just start with main() nowadays, like in all normal OSes? Jul 11 10:55:13 like normal a OS Jul 11 10:55:22 go take a look at the examples, you'll find out more Jul 11 10:57:23 thiago: sounds refershing Jul 11 10:59:01 Meego should die quickly as well as predecessors? Jul 11 10:59:16 what predecessors? Jul 11 10:59:29 mmm Jul 11 11:00:23 ok, looks like you will have to work on your opening lines a bit more :) Jul 11 11:02:10 Symbian will leave in the past, probably Jul 11 11:02:33 will leave what? Jul 11 11:03:37 <_wolf_> I think he means live. But I smell troll :P Jul 11 11:04:13 I quite like the idea of "will leave" "in the past" Jul 11 11:04:46 * lbt tries it for size : "meegodotby will leave in the past" Jul 11 11:05:55 He probably means "live" Jul 11 11:06:11 <_wolf_> I Jul 11 11:06:35 <_wolf_> I'd say everything leaves the past (behind) Jul 11 11:06:58 But yes, learning the language helps one's trolling skills tremendously Jul 11 11:13:25 yes, i badly know language, i study English, but i understand you, sorry Jul 11 11:14:21 meegodotby: :) OK .... we thought you meant something else. Jul 11 11:18:37 BTW, has anyone already complained about N900 causing capral tunnel syndrome on tmo? :) Jul 11 11:22:25 don't you have a blue tooth keyboard neural implant ? Jul 11 11:23:53 alas they are not supported in Maemo5, bug filed =) Jul 11 11:24:07 maybe lbt can help in #opensuse-buildservice :) Jul 11 11:25:07 I do have one of these bluetooth laser keyboard, nice and gimmicky Jul 11 11:26:54 odin_: ? Jul 11 12:41:11 mmmm http://www.theinquirer.net/inquirer/news/1721739/meego-flies-benchmark-test Jul 11 12:41:37 ARM vs atom..? Jul 11 12:42:02 when i read the comparatives, i stopped reading :P Jul 11 12:47:00 A moment, I think I have seen a clean test Jul 11 12:47:38 let's see... Jul 11 12:48:00 one benchmark run on an A4, one benchmark run on an OMAP3 or Snapdragon Jul 11 12:48:07 and one benchmark run on a dual-core Atom 1.5 GHz Jul 11 12:48:11 guess which one ran faster Jul 11 12:48:23 Whiever has got faster memory bus Jul 11 12:48:36 CPU does not even figure in the results ;) Jul 11 12:49:12 http://www.techradar.com/news/internet/tested-chrome-vs-ie8-vs-firefox-3-1-vs-safari-4-582159 Jul 11 12:49:19 No mirrors available for Meego? the download is REALLY SLOW! Jul 11 12:49:47 That was in March 2009 though Jul 11 12:50:06 Chrome is faster, different, and better Javascript engines Jul 11 12:51:22 http://www.xg-group.com/index.php/2010/06/23/firefox-counters-googles-browser-speed-test/ Jul 11 12:51:28 this is the latest Jul 11 12:52:37 Although I fail to understand people's obsession with interpreted (or JITted) languages Jul 11 12:53:16 They heroically go against the computational complexity laws and...well...fail every time :) Jul 11 12:57:16 Well, it seems like for some people *performance* is not the most important factor Jul 11 13:00:13 Then why have benchmarks? :) Jul 11 13:00:53 And why do people always complain about slow, unresponsive JavaScript or Java apps? :) Jul 11 13:01:12 Because they are *fed up* with them! Jul 11 13:01:25 In general users do that Jul 11 13:01:25 Nobody explained to them that performance has not been the most important factor? =) Jul 11 13:01:40 Programmers control users. Jul 11 13:04:32 lbt: you need testers ? Jul 11 13:06:56 For 3 days I'm trying to download the Meego Netbook 1.0 without success, everytime I get bad checksum or I just cancel the download being so *SLOW*, Meego SDK on the other hand works fine, grr Jul 11 13:09:05 damn it 1 hr and 30 minutes to download, this is insane Jul 11 13:14:37 dl9pf: yep :) Jul 11 13:15:33 i can give it a whirl Jul 11 13:29:32 dl9pf: on the community OBS I was thinking to have :current Jul 11 13:29:43 this would be replaced each week Jul 11 13:29:51 yes, from the snapshot - right Jul 11 13:30:09 we'd keep 2 or 3 dated versions for regression tests Jul 11 13:30:47 X-Fade: ping... if you're about Jul 11 13:31:06 for cross@armv7, we need the -accel packages there and the -x86-arm packages all in the armv7/:full Jul 11 13:31:08 I'm not sure what X-Fade's setup in terms of syncing in Jul 11 13:34:09 right... looks wrongly setup... :full -> /backend_storage/obs/meego1.0/i586_full/armel_full Jul 11 13:44:09 yes, all of these - looks complete Jul 11 13:44:48 including cross-*-accel*armv7l and *-x86*.armv7l Jul 11 13:48:14 do you want to sync now ? Jul 11 13:48:41 yep... looking for the rpms Jul 11 13:48:50 http://repo.meego.com/MeeGo/builds/1.0.80/ Jul 11 13:49:33 http://repo.meego.com/MeeGo/builds/1.0.80/1.0.80.9.20100706.1/ thats gcc4.4.2 Jul 11 13:49:48 yes Jul 11 13:51:13 ok - lets continue when synced. Jul 11 13:51:40 yep.... http://wiki.maemo.org/OpenSuse_Build_Service/MeeGo_Setup Jul 11 13:51:47 working on that Jul 11 13:52:30 what size is a full meego system now Jul 11 13:54:46 define full Jul 11 13:55:41 meego-netbook-chromium-ia32-1.0-20100524.1.img 25-May-2010 13:46 776M Jul 11 13:56:38 meego-n900-open-armv7l-1.0.0.20100525.1-sda.raw.bz2 26-May-2010 02:15 137M Jul 11 13:57:19 ScottishDuck: default netbook installation takes ~1,9 GB Jul 11 13:57:22 so that really depends on what "full" is - ok, comparing core with a full-blown netbook image is unfair. Jul 11 13:58:02 dl9pf: I don't see that n900 image as accurate Jul 11 13:58:06 oh, it's 1.0 Jul 11 14:27:22 Hi, is lbt or X-Fade online? Jul 11 14:34:59 yes Jul 11 15:02:30 ahma: ping Jul 11 15:41:16 hi where can i download meego 1.0.1 iso? Jul 11 15:44:42 i tried http://meego.com/downloads/releases/updates/meego-v1.0.1-netbook-update there is no download link Jul 11 15:47:32 anybody knows Jul 11 15:47:42 rjb: click MeeGo v1.0 for Netbooks at the page Jul 11 15:50:14 kimitake_ that link is for Meego v1.0 i want Meego v1.0.1 update Jul 11 15:51:47 rjb: oh, but you can update on the system, actually I have updated my system Jul 11 15:52:15 i wanted to try it live first Jul 11 15:55:26 please can you update http://repo.meego.com/MeeGo/releases/1.0/netbook/images/meego-netbook-chromium-ia32/ Jul 11 15:55:37 please add Meego v1.0.1 Jul 11 15:58:27 rjb, you can create a liveUSB using Fedora tool, it can be updated so then you can upgrade to v1.0.1.. Jul 11 16:03:19 kimitake_ thank you Jul 11 18:39:08 <_wolf_> oXwro77! Jul 11 18:39:22 <_wolf_> fsck! Jul 11 18:43:49 time for a new password :) Jul 11 20:01:53 * thiago_home discovers LinuxCon Brazil Jul 11 20:03:47 :o Jul 11 20:04:27 less than two months away from it Jul 11 20:05:04 and how many thousands of miles? Jul 11 20:05:19 nothing Jul 11 20:05:22 I'll be very close to it Jul 11 20:05:41 but will miss it by a few days Jul 11 20:07:56 I'm going to Brazil this Friday now Jul 11 20:24:53 is meego.com downloads slow? ~50Kb Jul 11 20:25:03 I heard it was slow, yeah Jul 11 20:25:36 I started trying to download the repo this afternoon from an ISP connected server... Jul 11 20:25:47 for the community OBS... Jul 11 20:26:06 its up to "evolution" Jul 11 20:26:40 * lbt wonders if he'll be able to keep up with the weekly builds at this rate ;) Jul 11 21:31:58 What's the link to the community OBS? Jul 11 21:34:12 lbt: X-Fade ^^ Jul 11 21:37:49 hey jebba \o Jul 11 21:38:31 * lcuk throws a ball @ w00t_ Jul 11 21:38:40 evening Jul 11 21:39:46 did you watch or have you been hacking Jul 11 22:17:29 hi Jul 11 22:17:35 is it possible to change UI fonts? Jul 11 22:17:48 font settings dont seem to apply Jul 11 23:58:43 hi jebba I don't think "community OBS" exists just yet, maybe big foot knows what the links is, ha ha Jul 12 00:26:16 odin_: well, there was a post to meego-dev and to the OBS bug in bugzilla saying it was in beta, so it's gotta be somewhere ;) Jul 12 00:26:41 they just didnt link to it so i cant quite figure out what is "community" OBS and which is the "meego" OBS Jul 12 00:26:56 [MeeGo-dev] Community OBS looking for beta testers   David Greaves http://lists.meego.com/pipermail/meego-dev/2010-July/003927.html Jul 12 00:27:24 "Bug 615 - Open the access of MeeGo OBS" http://bugs.meego.com/show_bug.cgi?id=615 Jul 12 00:33:03 why do people always believe open source means open development Jul 12 00:35:25 ScottishDuck: because if we have no input, why should we care? Jul 12 00:55:29 lbt: ping re: "Community OBS looking for beta testers" Jul 12 01:22:03 Hello Jul 12 01:22:06 anyone here Jul 12 01:22:13 I need some help with a little problem Jul 12 01:22:39 Well, you better wear a helmet around here.... Jul 12 01:23:27 ok Jul 12 01:23:41 I am trying to update my install and nothing happens. Jul 12 01:23:52 When i try to install a package Jul 12 01:24:03 I get a message about a policykit missing Jul 12 01:24:13 i am guessing they are related Jul 12 01:24:25 can i update via the command lie? Jul 12 01:24:27 line? Jul 12 01:24:32 terminal Jul 12 01:25:10 anyone seen that? Jul 12 01:26:16 davidvasta: "yum update" Jul 12 01:27:16 ok Jul 12 01:27:19 fair enough Jul 12 01:27:28 is there a command i should run Jul 12 01:27:32 not a YUM user Jul 12 01:28:14 yum Jul 12 01:30:30 got it Jul 12 01:30:35 thanks for the clue shaggy Jul 12 02:02:59 any nostagic diablo fans here **** ENDING LOGGING AT Mon Jul 12 02:59:57 2010