[00:52] <Sarvatt> robclark: so you're hitting utouch issues, it's a deep rabbit hole. things are dependent on the implementation in 1.10 in ubuntu at the moment but it should be fixed in precise very soon if it isn't already (not sure if you're using oneiric or precise)
[00:52] <Sarvatt> oh heck he's gone already
[00:54] <RAOF> :)
[00:58] <Sarvatt> multimonitor fix for unity making each display use a different fbo was mark fix released so im assuming chases fix is in precise now, im still on oneiric all around
[00:59] <Sarvatt> guess i better upgrade before going to budapest
[01:00] <Sarvatt> we just got done enabling one oem's ivybridge systems on oneiric and have a new one to do now so still gotta care about oneiric
[01:00] <Sarvatt> time to upgrade just in time to hit upstart breakage in precise, yay
[01:01] <RAOF> Oh, yu mean that I get to undo the gnome-settings-daemon thingy now? :)
[01:02] <bryce> Sarvatt, do you plan to do work on precise?  If not, may as well just hold off until after you get back.  Although, so far precise has been pretty stable.
[01:03] <Sarvatt> "work on precise"?
[01:03] <Sarvatt> i've been updating crap every now and then
[01:04] <Sarvatt> well just libdrm and intel updates so far
[01:05] <RAOF> There'll be the opportunity for mesa 7.12/8.0 packaging in Budapest, too.
[01:05] <bryce> RAOF, speaking of which how's xserver?
[01:05] <RAOF> xserver is bonza.
[01:06] <RAOF> The remaining blocker is synaptics being a pile of infuriating.
[01:06] <Sarvatt> budapest is all gonna be precise, im planning on ditching the hwe room to be in the desktop room to get crap done
[01:06] <bryce> Sarvatt, \o/
[01:06] <Sarvatt> they'll probably put hwe in the left side of the place with 4 rooms instead of the right side of the conference center with 15+ rooms again
[01:07] <Sarvatt> no cert going this time we got put with so who  knows
[01:07] <bryce> Sarvatt, yeah to sprints I usually bring two systems, one with devel so I can test stuff, one with stable so I can get work done.  This time both will have precise loaded
[01:07] <RAOF> All the touchpads I've got access to teleport the cursor to (0,0) at not-apparently-deterministinc intertvals, and it's not clear to me what level of the stack is blowing up (but my guess is input backport).  Chase is aware of it, and is doing synaptics work, so I'm waiting for that.
[01:07] <Sarvatt> hmm thats a good point
[01:07] <Sarvatt> i was just planning on bringing the macbook air
[01:08] <Sarvatt> better off bringing another
[01:08]  * RAOF seems to be running out of hardware that works enough to bother to bring to sprints.
[01:09] <Sarvatt> all my discrete gpu systems are too huge to bring though, hmm
[01:10] <Sarvatt> 8+lbs each
[01:10] <Sarvatt> asus g73, dell m4600
[01:10] <RAOF> This ati/intel hybrid will be making an appearance, as the only non-netbook I have that contains entirely functional hardware :)
[01:10] <bryce> yeah, I'm going with the thin arrandale and a little netbook.
[01:11] <bryce> the arrandale's wireless is borked at the moment though
[01:11] <Sarvatt> cert/oem guys arent going this time so there wont be a plethora of unreleased systems we could nab :(
[01:11] <RAOF> Oh, balls.
[01:11] <RAOF> Yes.
[01:11] <bryce> fortunately I'll probably just be doing libxrandrutils hacking, don't need any special hardware for that
[01:12] <Sarvatt> my goals are mesa 8.0 in precise, vmwgfx packaging, xserver 1.12 in edgers
[01:12] <Sarvatt> and of course the xserver 1.11 stuff
[01:13] <Sarvatt> frankenserver is going to be most of the week
[01:13] <RAOF> As Sarvatt suggested, we might also want to consider plain xserver 1.12 while we're all in the neighbourhood.
[01:13] <Sarvatt> i only suggested that to you in PM
[01:13] <Sarvatt> it has lots of drawbacks
[01:14] <RAOF> Sarvatt: What other work is necessary on the frankenserver.
[01:14] <RAOF> ?
[01:14] <Sarvatt> like, we've distributed to OEMS that we're shipping 1.11, and they are asking intel for 1.11 support for cedarview
[01:14] <Sarvatt> intel is only shipping xserver 1.9 support in the initial cdv driver
[01:14] <Sarvatt> xserver 1.9 and mesa 7.9
[01:14] <Sarvatt> and kernel 3.0
[01:14] <Sarvatt> such a nightmare
[01:14] <bryce> yeah we can talk about versions of things.  I've been procrastinating sending out that email about what versions of things we'll be shipping, since it's felt a bit uncertain
[01:15] <Sarvatt> basically whatever chromeos ships
[01:15] <RAOF> Heh.  Ok.  Or, we could have a simpler conversation here; it sounds like our previous statements have a larger inertia than I thought :)
[01:16] <Sarvatt> intel has no plans to support 1.11 for cdv, its all pie in the sky things so we can flip flop still if 1.12 is better :)
[01:16] <Sarvatt> this was literally last week
[01:16] <Sarvatt> intel UMG is a nightmare to deal with, poulsbo..
[01:17] <bryce> Just Say No
[01:17]  * bryce cracks an egg on a hot frying pan
[01:17] <Sarvatt> i have been for over a year
[01:17] <bryce> that's your brain on poulsbo
[01:17] <Sarvatt> they finally got it last week after detailing in a huge doc about what we've experienced wdealing with them about this
[01:17] <Sarvatt> opengl support? not happening
[01:18] <RAOF> That sounds like a *particularly* appealing netbook :)
[01:18] <Sarvatt> recompile clutter using gles instead, which will require a new architecture like lpia basically
[01:18] <RAOF> Oh.  I thought linaro already had basically the same problems, and solved them?
[01:18] <bryce> Sarvatt, you mentioned the oem's were spun up on 1.11; aside from cedarview is there anything else that would impact them if a different version was shipped?
[01:19] <Sarvatt> "yeah we experienced problems with GL apps too, what we did in meego was recompiled clutter using gles instead of gl (compile time option that requires everything recompiled against the different clutter) for mutter to work in meego"
[01:19] <Sarvatt> anything using GL hangs the system within 10 seconds
[01:19] <Sarvatt> nah its only cedarview thats a hot ticket item
[01:20] <Sarvatt> i'm more worried about breaking fglrx for DX
[01:20] <RAOF> I *think* linaro have the pluggable clutter gl/gles stuff licked.
[01:20] <Sarvatt> RAOF: the problem is its a compile time option, clutter using gles is incompatible with clutter compiled using gl
[01:20] <RAOF> But it occurs to me that you may well be in a better position to know that :)
[01:20]  * RAOF was pretty sure there was work to make that a run-time option, and that work had landed.
[01:21] <RAOF> Maybe I'm misremembering my blog postings :)
[01:21] <Sarvatt> arm doesn't care because they can target gles on that architecture because all the drivers ship gles and gl is an afterthought
[01:21] <Sarvatt> nope, look at debian/rules in clutter source
[01:21] <RAOF> Eh.
[01:22] <Sarvatt> are you thinking of cairo?
[01:22] <Sarvatt> thats what alf was doing
[01:22] <Sarvatt> for linaro
[01:22] <RAOF> That could be it.
[01:23] <Sarvatt> unity will have gles support in 12.04, but then someone is going to buy a netbook and try to install a game like openarena
[01:23] <RAOF> Hm.  I might *also* be thinking of run-time switchable wayland/X backends.
[01:23] <Sarvatt> and the system is going to lock up within 10 seconds
[01:23] <RAOF> No; they'll get terrible performance with llvmpipe, but we'll just disable GL.
[01:23] <RAOF> BY HOOK OR BY CROOK!
[01:24] <Sarvatt> yeah as long as we disable any kind of accelerated 3D support which is perfectly doable
[01:24] <RAOF> We can't selectively disable OpenGL?
[01:24] <Sarvatt> then they cat a 10k /var/log/Xorg.0.log in gnome-terminal and it takes 45 seconds to show in the terminal because the 2D "acceleration" is so slow
[01:26] <Sarvatt> its really bad, thats what i've been dealing with for the past month, sorry to complain :)
[01:26] <RAOF> What is IRC for if not for venting at hardware madness?
[01:27]  * RAOF wonders who hardware engineers vent at.
[01:27] <broder> their fabs
[01:28] <Sarvatt> the bongs they hit 24/7 for getting clogged up
[01:28] <bryce> heh
[01:33] <Sarvatt> i think i'm pulling a ROAF this time
[01:33] <Sarvatt> havent even packed for budapest yet
[01:34] <Sarvatt> tseliot isn't coming again btw
[01:34] <Sarvatt> yet another X team dinner missing someone :)
[01:34] <bryce> shucks
[01:36] <Sarvatt> RAOF: what synaptics problem were you hitting?
[01:36]  * bryce eyes bug 903859
[01:36] <ubot4> Launchpad bug 903859 in xorg (Ubuntu) "wireless connection is not hooking up after i upgraded to version 12 (affects: 1) (dups: 1) (heat: 11)" [Undecided,New] https://launchpad.net/bugs/903859
[01:36] <Sarvatt> i just realized, i have no machines that use synaptics
[01:36] <Sarvatt> dell has been all alps for over a year
[01:36] <RAOF> Sarvatt: The DDX?  Really?
[01:37] <Sarvatt> hmm i can use synaptics on the mba and lose all gestures if i really wanted to try it out apparently
[01:37] <RAOF> Sarvatt: I thought basically all touchpad-like hardware went through xserver-xorg-input-synaptics; my Apple Wireless Trackpad certainly does.
[01:38] <Sarvatt> RAOF: thats a udev rule change to make it happen
[01:38] <Sarvatt> evdev should be used
[01:38] <RAOF> That was true for... 11.04?  But synaptics definitely grew gesture support.
[01:38] <Sarvatt> hmm
[01:39] <RAOF> We were having a discussion with Chase for that extent - to what extent are gestures worth a really terrible touchpad experience.
[01:39] <RAOF> Because evdev makes for a rubbish touchpad DDX.
[01:39] <Sarvatt> you lose tap to click doing that dont you
[01:39] <Sarvatt> which is pretty major
[01:39] <Sarvatt> oh wait thats when you use evdev instead of synaptics
[01:39] <Sarvatt> and probably why it defaults to synaptics now
[01:40] <RAOF> Tap to click, two and 3 finger tap, two finger scroll, a bunch of filtering, and proper acceleration.
[01:40] <RAOF> You can *really* tell when you're using evdev rather than synaptics - your touchpad will feel like it's been posessed by the devil.
[01:41] <RAOF> Or, with the frankenserver, you can tell when you *are* using synaptics, because your pointer will teleport to (0,0) all the time for no obvious reason.
[01:41] <RAOF> :)
[01:43] <Sarvatt> hey someone said that happened even on 1.12
[01:43] <Sarvatt> i think it was ricotz
[01:44] <RAOF> That could well be the case.
[01:47] <RAOF> In all other respects the frankenserver seems to be a perfectly functional backport of the new input stack, so it wouldn't surprise me to find that bugs in the frankenserver are backported from 1.12.
[01:56] <bryce> we should bring lag bolts to tape on the side of our laptops
[01:56] <RAOF> What is a lag bolt?
[01:57] <bryce> http://www.boltdepot.com/Lag_bolts.aspx
[01:57] <bryce> well obviously any old bolt would do.  "lag bolt" sounds more frankensteiny though.
[02:09] <bryce> RAOF, hey did you get your OLCP box up and running with ubuntu?
[02:10] <bryce> I just received mine right before xmas, still haven't even opened it up to install the serial port wires
[02:10] <RAOF> I've got all the prerequisites, but I haven't actually spend the hour or so required to make it bootable.
[02:12] <Sarvatt> dont those use marvell?
[02:12] <Sarvatt> aka dont work post jaunty in ubuntu
[02:12] <Sarvatt> armv5
[02:12] <bryce> yeah arm hw
[02:12] <Sarvatt> we went armv6+ only in karmic
[02:13] <bryce> dunno if it's armv5 or 6
[02:13] <Sarvatt> debian armel will work at least
[02:14] <Sarvatt> i might be wrong but i thought they used the same hardware as sheevaplugs and stuff, marvell cpus that were armv5
[02:15] <bryce> CPU: Marvell Sheeva ARM (PJ4); 
[02:15] <bryce> ARMv7 architecture compatible; 
[02:15] <Sarvatt> oh cool
[02:15] <Sarvatt> not sure if theres a compatible kernel though
[02:16] <bryce> Linux Kernel: recent Linux 2.6.xx; Fedora base environment.
[02:16] <bryce>     The OLPC specific bits of the kernel are pulled from the olpc-2.6 GIT tree on dev.laptop.org: (http://dev.laptop.org/git/olpc-2.6) 
[02:16] <Sarvatt> but at least userspace supports it if its v6+
[02:17] <bryce> anyway something to tinker with when I'm bored
[02:17] <bryce> might have to enlist Dutch in doing some QA work on it ;-)
[02:17] <Sarvatt> lol
[02:19] <Sarvatt> i have the perfect system for dutch, if only it was pinetrail and not cedartrail
[02:23] <bryce> he loved playing with the babysitter's iphone.  wish I still had that xt2 touchscreen laptop
[02:23] <Sarvatt> well its for elementary schoolers, probably a bit too much for him still :)
[02:23] <Sarvatt> i keep forgetting how young he really is, he's a big kid :)
[02:28] <Sarvatt> http://www.cceinfo.com.br/produtos/governo/proj-secre-educ-pe/TabletPC%20-%20EC10IS2
[02:28] <Sarvatt> one of those, its got a handle and made to take a beating from kids
[02:29] <bryce> yeah, tho he's coming along.  He can recite the entire alphabet accurately now, names the letters correctly about 50/50, and can count to 20 only missing a few numbers along the way
[02:29] <bryce> he also has USB plugging in skillz now
[02:30] <Sarvatt> bryce: so it takes him 3 times to plug it in right, the same as us?
[02:30] <bryce> Sarvatt, that's right
[02:30] <Sarvatt> awesome :)
[19:54] <Prf_Jakob> Sarvatt, bryce, RAOF: How nervous do you guys get if we wait untill middle of next week to cut the branch?
[19:55] <Prf_Jakob> We are so close of getting GL3 support into softpipe.
[19:55] <Sarvatt> Prf_Jakob: feb 16th is just the date that makes us nervous :)
[19:55] <Prf_Jakob> Ok
[19:56] <Prf_Jakob> Next week it is, lets see if I can pause this release in time.
[19:57] <Sarvatt> at most it'll take a week to prepare and stage and put out a call for testing before putting it in the archive, can probably skip that last part but its a big change and mesa updates always break something for someone (hi KDE)
[19:58] <Sarvatt> i wonder if kwin still relies on hardcoded gl version strings that need updating every mesa release still
[19:59] <bryce> Prf_Jakob, next week would be fine with me
[19:59] <bryce> I suspect we're going to be too busy with sprint stuff to worry too much about much else
[20:00] <Prf_Jakob> okay
[20:00] <Prf_Jakob> Doing it now doesn't give us anything else?
[20:03] <bryce> Prf_Jakob, it would enable us to potentially do the packaging/testing work at the sprint, which would mean we could get this out to users pretty quick
[20:04] <bryce> but to be honest there's so many frequent interruptions during sprints that it's hard to do intensive packaging work, at least not without making a lot of errors
[20:16] <Prf_Jakob> bryce: ok, I think the release will be better of with better GL3 support.
[20:16] <Prf_Jakob> Is the sprint all week?
[20:16] <bryce> yes
[20:18] <bryce> Prf_Jakob, would there be much chance if adding the GL3 support might add regressions?
[20:18] <Sarvatt> we most likely wont have GL3 support in the archive for precise though, it'd need to go through legal and almost guaranteed that can't happen in this time frame. floating point textures and all
[20:18] <Prf_Jakob> Sarvatt: oh that crap...
[20:19] <bryce> (knowing how sometimes many bugs in a software release come from whatever feature got added right before release ;-) )
[20:20] <Prf_Jakob> bryce: we are getting a lot better at regression testing with piglit, and we are really really close so there isn't much to do.
[20:20] <Prf_Jakob> I think most of the common code is already there.
[20:22] <bryce> ok cool
[20:32] <Prf_Jakob> bryce, Sarvatt: http://lists.freedesktop.org/archives/mesa-dev/2012-January/017026.html
[20:36] <bryce> *nod*
[22:10] <FernandoMiguel> evening