=== asac_ is now known as asac [08:32] asac: I sax the new fennec package; thanks for pointing it out! I need to try it out now [08:33] thanks fta too [08:53] lool: file bugs ;) [09:08] asac, we will :) [09:08] * ogra hugs fta in absence for packaging it [09:09] There's a fennec package? [09:09] since yesterday [09:09] i mean, there is one since wuite some time ... but it entered universe yesterday [09:10] I see that [09:10] i tried it on the n810 before and wasnt very thriled yet [09:10] *thrilled [09:10] Oooh, it's tiny [09:10] while its faster its still lacking a lot in speed and handling compared to the maemo browser [09:11] but if oyu tried the first fennec package you notice the improvements === asac_ is now known as asac [11:27] Agenda up for Team Meeting in 34 minutes at https://wiki.ubuntu.com/MobileTeam/Meeting/2009/20090115 : please add anything you'd like discussed. [11:39] ugh, i still havent finished the touchscreen spec (though started to work on the code :/ ) damned paperwork [11:40] It's all about paperwork! [11:41] its all about code ! [11:41] paperwork just documents it :) [11:42] I suppose. I rather like the waterfall approach to software: it helps get the solution right for the largest number of people the first time. === StevenK is now known as MootBot === MootBot is now known as StevenK === mcasadevall is now known as NCommander === mcasadevall is now known as NCommander [13:37] heya everybody [22:24] hello everyone, I am working on an OEM install for an Atom based netbook, is there any documentation on why I should use lpia vs i386? [22:25] i think there is some info at moblin.org [22:25] right now i386 seems to be a smoother experience for us in terms of installation and support from people like Skype and IBM (who don't know how to package stuff properly) [22:26] alanbell, I dont think the i386 has the same powersaving flags in the kernel [22:27] The kernels are different: it's worth comparing the trees to see what you need. [22:28] Vantrax: so battery life might be better with lpia. Anyone got a quantitative comparison? (I will probably do my own and publish the results) [22:28] right, there are some extra powermanagement switches in lpia, but the main issue is that i386 is compiled for m586 while lpia is 686 ... that will give you the most significant difference from a kernel POV [22:29] Except it's not. [22:29] ?? [22:29] The lpia kernel has CONFIG_M586 set. [22:29] oh, it didnt when i looked last time [22:29] which is quite a while ago i admit [22:29] Well it did a couple days ago :) [22:30] why was that dropped ? [22:30] afaik that makes a significant difference on ATOM and lpia wont run on non ATOM [22:31] Actually, it will, now. [22:31] oh ? [22:31] cool [22:31] persia ftw [22:31] Anyway, I don't know anything about why: I only read the config trying to sort out which kernel to install. [22:31] * ogra should read the UDS spec [22:31] * Vantrax should too [22:31] guessing there is one :) [22:32] Vantrax, Except it's not necessarily good that the lpia kernel works on everything, because it means fewer optimisations. [22:32] yeah [22:32] thats true [22:32] I thought the idea was to make it super efficient for atoms, and everything else was irrelevant [22:32] I see quite a few things that are not available for lpia and lots of scary looking advice about dpkg --force-architecture to get i386 debs installed [22:32] and i guess another for the ARM processors [22:33] in that light you should probably go for i386 [22:33] alanbell, yeah, that's not ideal, but nothing really to do about that. [22:33] Vantrax, Actually, there are about 4 kernels already in the archive for arm, and probably 2-3 more expected (e.g. OMAP, pax2xx, etc.) [22:33] even if you lose 20min battery life through it (which i doubt) if you *need* something like skype thats the cleanest you can do [22:34] thats good to hear, Im guessing thats the next big expansion area [22:34] There was a UDS session in the Mobile track about lpia vs. i386 where we basically determined that lpia was essentially experimental, to look at optimisations, and understand future direction. [22:34] ogra: trouble is from the OEM point of view it means that if we put lpia on we are going to get customers who can't install skype and can't figure out why [22:35] and flash and symphony and Lotu Notes etc. [22:35] right, thats what i mean [22:35] ... no offence but symphony should never be used, and lotus notes, even the 8.5 linux client is crap [22:35] ogra> in that light you should probably go for i386 [22:35] even if you lose 20min battery life through it (which i doubt) if you *need* something like skype thats the cleanest you can do [22:35] is wht i was saying :) [22:36] best way i've found to do lotus notes is 6.5.3 in wine [22:37] Vantrax: yes, I know that but I am a Notes developer(trying to quit) and I don't need the flak from my friends at IBM :-) [22:37] any thoughts on this script http://www.matt-helps.com/convert-deb-package-architecture-from-i386-to-lpia [22:38] I am thinking if lpia could automagically on the fly repackage i386 stuff it would be a bit smoother [22:38] alanbell, I pity you... its one of the largest gripes with most IT support groups, especially on linux. [22:39] the 8.5 beta for linux is incredibly slow, and the 8.0.1 gave java errors all over the place [22:39] * ogra lols about alanbell " I am a Notes developer(trying to quit)" [22:39] lol [22:39] * ogra is a smoker (not trying to quit) :) [22:39] It seems that domino is losing ground to exchange too... [22:39] * Vantrax thinks linux needs an answer to AD/NDS and Domino/Exchange [22:40] Domino is still the fastest app dev environment for a specific class of applications [22:40] I never saw it as an email system [22:40] just a document database that is good at moving documents between databases [22:40] most people do... tho were I work we run a large amount of other functions through domino too [22:40] email is a trivial app to write using that framework [22:41] yet they cant get it right >.< [22:41] no, because you can write a better email app if you start out trying to write one. [22:42] I remember ages ago asking IBM to just stop with the email thing and call it a collaboration application platform [22:42] but anyhow, enough of talking about non-free software :-) [22:43] I think I am going to have to do some i386 vs lpia benchmarks to figure out if we are going to gain anything much with lpia [22:45] one big problem is that the desktop CD for lpia simply doesn't install. It fails to find an installable kernel. [22:45] there are a load of instructions on how to nurse it though on the Aspire One [22:45] but that is a nasty to give to customers. [22:48] whats the customer base? [22:48] Vantrax: netbook, like the Aspire. Previously aimed at bulk consumer market [22:49] this model will be a linux build as an online sale [22:49] somewhat sophisticated customers (selecting Linux as a positive choice) [22:50] probably 2nd computer [22:50] also possibly corporate deals and education [22:50] then id stick with i386 [22:51] I think I am heading that way. I just need answers when people say "you are using Atom, but you didn't use the right build for the chip - you idiot!" [22:52] especially when they see the Mini9 has lpia [22:52] ive found the default 8.10 install rather good on my mini 9 [22:52] better battery life than windows [22:53] in fact, this is me with my first problem with lpia on mini 9 [22:53] http://www.dominux.co.uk/index.php/2009/01/09/notes-85-on-intel-atom-lpia-architecture/ [22:53] in my trying to quit life :-) [22:54] alanbell, You may find your script easier with dpkg -e and dpkg -x [22:56] alanbell, You may also find sed -i useful. Lastly note that blindly changing may work today, but it's not guaranteed to work as the architectures diverge. [22:57] persia: thanks [22:57] I kind of think that dpkg should be more sophisticated about architecture relationships [22:58] Well, architectures aren't supposed to be that similar :) [22:58] sort of knowing that you can install i386 on lpia but you can't install lpia on i386 [22:59] More than anything, it's that even the techs at Intel are still figuring out what to change to optimise for power, rather than optimising for performance. I expect it will be another year or two before we get distinct enough that it's not convertible. [22:59] Well, actually, with today's packages, you can install lpia on i386. [22:59] there are plenty of similar architectures if you look hard enough [23:00] there are a lot of Chinese MIPSel chips that implement the bits of MIPS that are out of copyright [23:00] so they end up as MIPS but without the maths co-processor [23:01] I expect the various ARM flavours are like that as it is a licensed core design [23:04] is there any documentation of the UDS discussions about this https://blueprints.launchpad.net/ubuntu/+spec/lpia-versus-i386 [23:10] alanbell, Well, actually, all the Ubuntu-supported ARM processors use the same userspace, which is "armel". [23:11] If the wiki spec isn't written yet, you can watch the video. [23:12] I think it is 00002.ogv from http://videos.ubuntu.com/uds/jaunty/Mobile/2008-12-09/afternoon/ [23:12] Mind you, these aren't very excellent videos, and it's live discussion, so lots of "umm..." and the like. [23:14] persia: perfect, thanks. [23:16] persia: yes, all the ARM are the same build today, I was just thinking that someone could do an ARM with some special optimisations available. [23:16] but I know very little about processsor design [23:17] Well, maybe. Personally, I'm not convinced that it's worth having a separate architecture for optimisations. [23:18] architecture differences imply instruction set differences. As I understand things, Intel is expecting this to happen with lpia. [23:21] lpia-versus-i386 starts about 28:37 in 00002.ogv