[00:47] <soreau> bryceh: cool
[07:19] <mlankhorst> morning
[07:20] <mlankhorst> RAOF: re-ping
[07:21] <RAOF> mlankhorst: When's your membership-board appointment?
[07:21] <mlankhorst> going for uds-r summit
[07:21] <RAOF> mlankhorst: I've just been attacking colord with a stick today, so I haven't got around to endorsing you :)
[07:21] <mlankhorst> I'm sure you used colorful language to fight it off!
[07:22] <tjaalton> mlankhorst: I uploaded an earlier version of xserver, didn't have the latest commit, oops .)
[07:22] <tjaalton> :)
[07:23] <mlankhorst> mvo: I attached logs to bug 1062503 yesterday
[08:24] <tjaalton> mlankhorst: is llvmpipe working for you on the panda? looks like I have just the normal swrast, and thus no compiz on quantal
[08:36] <mlankhorst> tjaalton: I use precise still, the ti extras do some patching so quantal won't work out of the box
[08:37] <mlankhorst> presumably for the experimental dri2 video extension
[08:38] <tjaalton> I dist-upgraded to quantal..
[08:39] <tjaalton> don't have the ppa
[08:40] <tjaalton> best to review the build-logs then
[08:40] <tjaalton>         llvm:            no
[08:40] <tjaalton> sigh
[08:41] <mlankhorst> armel or armhf?
[08:41] <tjaalton> armhf
[08:41] <tjaalton> same on armel
[08:42] <mlankhorst> yeah but on armhf i have the illusion you can still get some speed
[08:42] <tjaalton> --with-gallium-drivers=" nouveau r600 r300 svga swrast"
[08:43] <mlankhorst> I don't see llvmpipe in debian/rules otherwise
[08:44] <mlankhorst> unless it's that other rule, let me test..
[08:45] <tjaalton> it's enabled, ifeq (,$(filter $(DEB_HOST_ARCH_CPU), amd64 i386 arm))
[08:45] <tjaalton> but llvm is turned off somehow
[08:46] <tjaalton> oh heh
[08:46] <mlankhorst>   # LLVM is required for r300g and recommended for swrastg on x86:
[08:46] <mlankhorst> that one? :p
[08:46] <tjaalton> yeah that :)
[08:46] <tjaalton> sheesh
[08:46] <mlankhorst> might want to enable it for radeonsi too then
[08:48] <tjaalton> it's enabled on x86 already, so it should be good?
[08:48] <mlankhorst> on arm? :p
[08:48] <tjaalton> radeonsi for arm?
[08:49] <mlankhorst> we build the rest, so why not?
[08:49] <tjaalton> just makes the build longer :)
[08:50] <mlankhorst> throw more pandas at it!
[08:50] <mlankhorst> seem to be 11 that are building armhf now
[08:51] <tjaalton> I see 8
[08:51] <tjaalton> installed
[08:52] <tjaalton> only one of them makes sense for arm :)
[08:52] <mlankhorst> don't some arm come with a pci-e port?
[08:52] <tjaalton> can you buy it?
[08:53] <mlankhorst> https://www.globalscaletechnologies.com/t-openrdudetails.aspx
[08:54] <tjaalton> are there any pcie x1 cards available?-)
[08:54] <mlankhorst> don't make me look harder, but presumably :P
[08:55] <tjaalton> bah
[08:55] <tjaalton> but not newer radeon ones anyway :)
[08:55] <tjaalton> so I'll just try to make llvmpipe work
[08:57] <mlankhorst> booooooooooo
[08:58] <mlankhorst> just because there aren't any today, doesn't mean that amd has no secret plans to release them tomorrow
[09:03] <tjaalton> when such a miracle happens, I'll gladly flick the switch :)
[09:03] <tjaalton> even as an sru
[09:03] <tjaalton> but now that we're frozen..
[09:06] <tjaalton> should it be possible to crosscompile for armhf on amd64?
[09:06] <tjaalton> with a chroot
[09:06] <jcristau> if you have a cross-compiler..
[09:06] <mlankhorst> nah you just add it as foreign arch and use multiarch >:X
[09:07] <mlankhorst> should work
[09:07] <mlankhorst> still going to need a cross compiler though
[09:07] <tjaalton> hum, ok
[09:09] <mlankhorst> although I don't think it would be a really supported use of multiarch at this point
[09:09] <tjaalton> meh, I'll just build on the damn box
[09:10] <mlankhorst> it's what I've been doing so far :)
[09:10] <tjaalton> is eth0 gigabit or not?
[09:10] <mlankhorst> you wish
[09:10] <tjaalton> wonder if it's faster to build on nfs
[09:10] <mlankhorst> but it's still faster than a sd card
[09:10] <tjaalton> right
[09:10] <mlankhorst> I'm using a ssd now on the usb port
[09:10] <mlankhorst> 22 megabyte/second
[09:11] <tjaalton> blimey
[09:11] <mlankhorst> it's great for complete silence though :)
[09:11] <tjaalton> that's about it
[09:12] <mlankhorst> still a great use!
[09:22] <mvo> mlankhorst: thanks for the logs, sorry, the near release is giving me a hard time to find a free time to properly look at it, it looks like a bug in apt, one nice thing would be to know if its also happening with the quantal version of apt
[09:23] <mlankhorst> can I just upgrade apt on precise?
[09:24] <mvo> mlankhorst: not sure, it will probably bring a bunch of stuff like a new libstdc++ with it, but if you don't mind (or have a VM) then that should be ok
[09:27] <mlankhorst> seems to work, lets see..
[09:29] <mlankhorst> mvo: likely failing in the same way
[09:30] <mlankhorst> using newer libapt + apt from quantal
[09:31] <mvo> mlankhorst: yeah, the ordering code has not much changed iirc, just wanted to double check that I/we are not hunting smething that is already taken care of
[09:31] <mlankhorst> I think the ordering is fine, it's just for some reason deciding to configure before it is even extracted
[09:56] <mlankhorst> tjaalton: hah, radeonsi builds fine
[10:02] <tjaalton> mlankhorst: bah :)
[10:02] <tjaalton> what, the panda is single-core?
[10:02] <mlankhorst> it's dual core
[10:03] <mlankhorst> I disabled one though since the ti ppa had some issue with smp iirc
[10:03] <tjaalton> oh right
[10:03] <tjaalton> misread /proc/cpuinfo
[10:05] <mlankhorst> however mine probably still finished faster because of the faster io
[10:05] <tjaalton> how long did it take?
[10:06] <mlankhorst> still going at the install phase :P
[10:07] <mlankhorst> judging from timestamps, total is approximately 1 hour 20 minutes
[10:07] <tjaalton> great
[10:08] <tjaalton> and someone wants these in servers..
[10:08] <mlankhorst> there's 2 ways to measure performance per watt, and intel and arm can both claim they're the best at it
[10:10] <mlankhorst> one is keeping the total watts fixed, in that case arm wins since intel doesn't come close in total power consumption to things like phones, other is simply measuring flops/watt in which case intel would win :p
[10:15] <tjaalton> they're good for something, this box isn't :)
[10:25] <mlankhorst> it's good for testing that type of box!
[10:25] <tjaalton> yeah, I'm just venting
[10:26] <mlankhorst> my n9 might have git on it, but it's not a real buildbox
[10:27] <tjaalton> yeah it's io-capped here
[10:28] <tjaalton> panda that is
[10:50] <mlankhorst> do we have blueprints for R yet?
[10:53] <tjaalton> sure
[10:53] <tjaalton> ah
[10:53] <tjaalton> the team doesn't
[10:53] <mlankhorst> yeah looks like I should probably write the one for lts backports
[10:57] <tjaalton> meh, still no llvm
[10:57] <mlankhorst> how long did your build take?
[10:58] <tjaalton> running lintian
[10:58] <tjaalton> should finish soon
[10:59] <tjaalton> duh
[10:59] <tjaalton> DEB_HOST_ARCH != DEB_HOST_ARCH_CPU
[11:00] <mlankhorst> haha
[11:00] <mlankhorst> mine built :P
[11:00] <mlankhorst>    # LLVM is required for r300g and recommended for swrastg on x86:
[11:00] <mlankhorst> -  ifneq (,$(filter $(DEB_HOST_ARCH),amd64 i386 kfreebsd-amd64 kfreebsd-i386))
[11:00] <mlankhorst> +  ifneq (,$(filter $(DEB_HOST_ARCH),amd64 i386 kfreebsd-amd64 kfreebsd-i386 armel armhf))
[11:00] <tjaalton> yup
[11:01] <tjaalton> debuild -b  4189,16s user 543,60s system 139% cpu 56:37,54 total
[11:02] <mlankhorst> aw you won because of multithreading
[11:02]  * mlankhorst shakes fist
[11:02] <tjaalton> bwahaha..
[11:02] <tjaalton> if only it could fit in the ramdisk..
[11:09] <tjaalton> need to fix the build-deps too
[11:15] <mlankhorst> oh right
[11:15] <tjaalton> cheated, and dropped all the other dri drivers for this build..
[11:15] <mlankhorst> OMG :o
[11:15] <tjaalton> should be a bit faster to build :)
[11:17] <mlankhorst> can I cheat and update https://wiki.ubuntu.com/X/Blueprints/LtsPointUpdatesForXorg for R?
[11:17] <tjaalton> go ahead :)
[11:24]  * mlankhorst didn't just add a question mark to the end of the url just to get it in launchpad as specification a second time *whistles innocently*
[12:05] <tjaalton> doesn't work
[12:45] <mlankhorst> :/
[12:46] <mlankhorst> it builds llvmpipe though
[13:05] <tjaalton> uploaded 9.0 anyway
[13:06] <tjaalton> without the change
[13:53] <mvo> mlankhorst: just to confirm , that quantal version of apt-get had the same error #1062503 ?
[13:54] <mlankhorst> yeah
[13:54] <mlankhorst> which reminds me have to downgrade it
[14:21] <mlankhorst> oh 3.6.1-rt1 is a fun read :-)
[20:30] <bryceh> lots of xorg-server crash bugs this week
[20:36] <mlankhorst> yeah not happy
[20:37] <mlankhorst> especially since the backtraces are useless
[22:02] <RAOF> That's annoying.
[22:02] <RAOF> What's generating rubbish backtraces?
[22:03] <mlankhorst> my guess is xorg-server trying to do a clean shutdown after something already went wrong
[22:04] <RAOF> This is quantal, right?
[22:04] <mlankhorst> yeah
[23:13] <Prf_Jakob> mlankhorst: have you asked Thomas btw?