/srv/irclogs.ubuntu.com/2012/12/21/#ubuntu-arm.txt

robruhttps://bugs.launchpad.net/ubuntu-nexus7/+bug/1092734 am I the only person experiencing this bug? seems odd that nobody would have reported this until now01:43
ubot2`Launchpad bug 1092734 in ubuntu-nexus7 "Software updater crashes, cannot update software graphically." [High,New]01:43
=== zz_chihchun is now known as chihchun
=== asiekierka_ is now known as asiekierka
dholbachgood morning07:59
=== yofel_ is now known as yofel
=== chihchun is now known as zz_chihchun
=== zz_chihchun is now known as chihchun
=== rsalveti_ is now known as rsalveti
=== Ursinha is now known as Ursinha-afk
=== Ursinha-afk is now known as Ursinha
=== chihchun is now known as zz_chihchun
tempinhi11:06
tempinhow can i search for / browse the availbable packages for ubuntu for arm? like 'apt-cache search' but online?11:07
tempinI want to search a bit before I replace my Angstrom with Ubuntu11:08
persiatempin: packages.ubuntu.com11:10
infinitytempin: 99% of what you find in apt-cache on x86 will also be there on arm.11:10
infinitypersia: packages.ubuntu.com doesn't list ports.11:10
persiasomeone should kill the very idea of ports :/11:10
infinitytempin: rmadison will work, if you're curious if ports are in sync for something specific.11:10
infinitypersia: This isn't actually about the "ports" split at all, it's just that Rhonda hasn't gotten around to making it do all arches (I've asked).11:11
infinitypersia: packages.ubuntu.com isn't an "official" service in any meaningful way.11:11
infinitytempin: Generally, unless you happen to have a deep love of haskell bits, you won't find anything missing on ARM that you care about, I imagine.11:12
infinitytempin: "Ubuntu for ARM" is about as meaningless as "Ubuntu for i386", it's all just Ubuntu, built for many arches.11:12
tempininfinity: thanks, didn't know that! I got a beaglebone as a gift and this is all completely new to me :)11:17
infinitytempin: It's one of the nicer things we inherited from Debian.  The idea that you can bounce from arch to arch and, other than some just inherently unportable or broken software, the rest is all there, in the same versions, and works exactly the same.11:18
infinitySo, once you get past bootloaders, installers, and kernels being slightly different (sometimes), and you've logged in, you'll find it's indistinguishable from your Intel box.  Except slow. :P11:19
* persia has intel machines that are slower than some arm machines with Ubuntu installed11:21
infinityI don't have any PPros that old (nor any A15s)11:21
persiaMine's an A10011:22
infinityI used to have slow Intel kit running Ubuntu (P200MMX), but we dropped support. :/11:22
persiaI think at 450MHz, but it might be 600.  Used to run lpia, back when folk thought that was a good idea.11:22
infinityAhh, yeah, a decent A9 would outperform that.11:22
infinityThat said, the current crop of A15s will outperform a fair chunk of Intel hardware we support.11:23
infinityThe gap is closing faster than I thought it would.11:23
infinityI'm actually getting pretty excited to see functional aardvarch64 silicon.11:24
persiaYou underestimate the volume of underclocked ia32 chips with limited featuresets used in the embedded world.11:24
infinityDefine "limited"... We don't support anything << 686 anymore.11:24
persiaIt's doesn't require that much power.  An 800MHz i.MX51 handily outperformed 600MHz A100.11:25
infinityBut yeah, even in the 686 world, there's a ton of variety out there, I know.11:25
persiaMuch of "lpia" claimed i686, and could handle enough of it that we've only lost a few processors (including the A100: one needs at least an Atom to run Ubuntu)11:25
infinityReally?  What instruction is missing?11:26
persiaIIRC CMOV11:26
infinityOh, we may have gone cmov by default, I forgot about that.11:27
persiaRight, we did with karmic, and lots of folk with that big tablet that had the split keyboard complained it wouldn't boot.11:27
infinityThat's not "686", mind you, that's a conscious decision to be 686+cmov and screw a few CPUs.11:27
persiaIndeed, which is why I say your claim we need 686 isn'5 sufficient, but we're well off topic.11:27
persiaIn summary, don't claim ARM is slow because it's ARM: that's not the case (except in certain very special circumstances).11:28
infinityOh, I was doing no such thing.11:28
persiaClaim that a *given specific* chip is slow because it has a low clock rate, or because it has some design feature that requires lots of overhead.11:28
infinityARM is slow at the top end because of its historical target market.  That's the gap I was talking about closing faster than I thought it would.11:29
persiaExcept slow. :P!" :p11:29
infinitypersia: A beaglebone is slower than most people's desktops.  Just sayin'.11:29
persiaThat's not really fair, but yeah :)11:29
persia(even I don't have any Intel kit which works and is that slow)11:29
tempinthanks all for the quick info/help! byebye11:30
infinityI think we bored him off IRC.11:30
infinityDo we win a prize?11:30
persiaDunce hats: it's our turn to sit in the corner and feel guilty.11:31
infinityI might just have a dunce nap instead.11:32
persia:)11:33
lilstevielol11:35
=== doko__ is now known as doko
=== Quintasan_ is now known as Quintasan
hrwis not cmov part of pentiumpro core (first i686)?13:37
lilsteviehrw: I think he was saying it is missing on some lpia cpus13:41
hrwx86 is a mess13:45
lilstevie:p13:45
lilsteviealthough arm is too13:45
hrwi486 was last sensible name. i586 had too many options already13:46
hrwlilstevie: not so much.13:46
lilsteviehrw: heh13:46
hrwbut also messy13:46
hrwarmv4 is one. armv4t had few options (ep93xx vfp for example)13:47
hrwarmv5t could have iwmmxt if was intel/marvell13:47
hrwarmv6 had vfp optional13:47
persiaISA designers seem to take 4 generations to discover that there are multiple audiences for their products...13:47
hrwyes13:47
persiaVersions up to 4 are usually OK, and after tend to be all over the map.13:48
persiaIIRC, the split between POWER and PPC was with POWER5 as well.13:48
lilstevietbh arm problems aren't really related to the ISA though, more the inconsistencies between vendors13:48
lilstevienot that embedded x86 is any better13:48
persialilstevie: Such inconsistency is partially driven by the presence of optional features in the ISA.  With a consistent ISA, software doesn't care as much about implementation differences.13:49
lilsteviepersia: except with the boot flow, that is just each vendor doing it their own way13:49
persiaYeah, ARM is special that way.  Most architectures agree on a common pre-boot interface and implement solutions to reach a common starting point.13:50
* persia blames a big gap in available general purpose consumer devices in the ARM world13:50
lilstevieyeah13:51
persia(the same thing is starting to be more common for ppc too, because there's a lack of general devices, so everyone "optimises" differently playing embedded games)13:51
hrwpersia: even on x86 you have several booting methods. but much less popular then on arm13:51
persiahrw: Right, but there are (four) standards that are mostly used outside of the embedded market.  For true embedded devices, it's *much better* to specialise.13:52
* persia remains triumphantly glad that nobody every made any serious effort to make Ubuntu embedded-suitable13:53
lilstevielol13:53
persiaNo, really.  There was talk about it with the "UME" effort, which was quelled enough that "Ubuntu Mobile" was a normal OS, if tuned a bit for lower power consumption and touch.13:54
persiaAnd every once in a while someone brings it up again, like the folk that wanted to rebuild as much as possible against dietlibc13:54
persiaBut most folk really want to be able to program in a general purpose environment (and only a few actually understand how to work without automated memory management, or even without automatic garbage collection), so it doesn't happen.13:55
hrwdietlibc...14:07
hrwit's for those which have cpus with 4K of ram?14:07
mhall119mfisch: ping14:47
mhall119is the Raring Nexus 7 image using the Ubuntu kernel now, or still a modified version of the Android kernel?14:48
lilsteviemhall119: it will be a modified version of the android kernel else you would not be able to get acceleration14:49
mhall119thanks lilstevie14:54
mfischmhall119: yeah still android and I believe it still also has some binary drivers14:57
=== ndec_ is now known as ndec|vacations
kulvewhat makes it "android" kernel? It is the same yes, but configured differently. It's not missing anything by being an "android kernel". A bit older than what it could be otherwise though15:30
kulvealso it will always have binary libraries for hw acceleration15:31
vanhoofogra_: we're screwed! time to go shopping :|15:43
=== steev__ is now known as steev
marvin24_janimo: I pushed a linux-ac100-3.8 to my gitorious tree17:33
marvin24_what's the best way to provide patches?17:33
astra05hey ubuntu arm18:11
astra05i was wondering if anyone know about support for the Avatar AVRA-138A118:11
achiangmarvin24_: janimo is on holidays until 2013... email might be better than irc18:14
marvin24_achiang: ok, thanks18:15
* marvin24_ should have guessed that ...18:15
astra05or anyone have a resource on determing bootability?18:19
=== Hexxeh_ is now known as Hexxeh
=== awafaa is now known as FunkyPenguin
=== awafaa_ is now known as awafaa

Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!