[01:00] You know what would be really awesome? [01:01] Documentation of the ownership semantics of pointers we pass into mirclient. === duflu_ is now known as duflu [03:27] RAOF: What is the ID in "X -mir ID" ? [03:28] It's the name that XMir uses to identify its connection. [03:28] duflu: It's a string that usc uses to distinguish between different xservers. [03:29] RAOF: Oh, my mistake for assuming it was always an int [03:30] It's passed verbatim through to mir_connect [03:30] usc just happens to pass integer strings for it :) [03:31] RAOF: Ah, kay. I know the connect string is harmless. Didn't know that's what -mir ID was [03:31] * RAOF should possibly document it better [03:35] RAOF: sometimes is great when a team has a techical writer for some of these items that need documentation. They don't get destracted by needing to code. [03:37] oiaohm: That's true, you do get better results with dedicated technical writers. However in this case we're talking about a very small section of code, not part of any docs [03:38] * duflu is impressed that --display-config sidebyside figures out where the monitors are in space and gets them *backwards* every time ;) [03:39] RAOF: Is there a reason why XMir changes the monitor layout immediately on startup, even with zero X clients? [03:42] duflu: Because that's X's default behaviour, and I haven't yet overridden it. [03:43] RAOF: Kind of?... X without Mir doesn't change the resolution or start mirroring. Whereas XMir does both (bug 1216224) [03:43] bug 1216224 in XMir "[multimonitor] XMir defaults to wrong resolution 1152x864 instead of 1920x1200." [Undecided,Confirmed] https://launchpad.net/bugs/1216224 [03:44] duflu: xf86-video-intel overrides X's default behaviour. [03:47] Whee, i915 hangcheck === duflu_ is now known as duflu [03:53] There are a lot of horible x11 features killed by drivers. [04:17] good morning :) [04:19] Aloha! [04:19] tvoss, hello:) [04:20] smartboyhw, hey there :) how is it going? [04:20] tvoss, quite good:) How's XMir? [04:20] smartboyhw, alive and we made feature freeze, now some issue to fix [04:20] tvoss, \o/ [04:46] * duflu goes to find lunch, and the store [05:24] tvoss: BTW if there are still radeon corruption issues without bugs, please log another bug. We only have one right now... bug 1218815 [05:24] bug 1218815 in xorg-server (Ubuntu) "[radeon] Graphic glitches and screen corruption since the latest update" [Undecided,Incomplete] https://launchpad.net/bugs/1218815 [06:15] Harumph. [06:16] Why am I receiving unpared focus/unfocus events? [06:17] Oh, of course! [06:17] mir_surface_release doesn't implicitly trigger a focus lost event. [06:20] RAOF: Wonder if it should.... [06:20] It would make my life easier. [06:21] And I don't think that my use case is *that* peculiar. [06:22] And given that I need to add at least *some* code to make this work... [06:25] RAOF: Though that's another reason why the even should be on the connection rather than on the surface... [06:26] Well, possibly multi-layered? [06:28] Don't know [06:28] RAOF: Is there a way to make X not ignore the LD_LIBRARY_PATH? [06:28] It doesn't ignore LD_LIBRARY_PATH [06:28] Hmm, mine is [06:28] Are you trying to pass LD_LIBRARY_PATH througgh sudo again? [06:28] RAOF: No [06:30] Hm. Should work. I've used it in the past. [06:36] RAOF: Sanity check: Does the frame ordering issue occur on nouveau/radeon? [06:37] duflu: I don't know; I don't have dual-head nouveau/radeon set up. === smartboyhw_ is now known as smartboyhw [06:51] RAOF: Sanity check (2): What branch/URL contains the latest XMir? [07:02] duflu: The Ubuntu archive. [07:02] good morning [07:02] Oh, I'm cloning that just now [07:02] Morning dholbach [07:02] hi duflu [07:04] Aha! unity-system-compositor needs to handle this. Maybe. [07:09] RAOF: So... maintaining in archive means maintained as a patch? That sounds painful [07:09] Possibly less painful than not being in archive though [07:10] duflu: Ah, no. I maintainish on vladmir-upstreaming branch of github, then dump the diff of that against vladmir-upstream-base to generate the patch. [07:11] duflu: If you want to do upstreamish hacking on xmir that's where you should hack, but if you want to do printf debugging you want to start with the archive. [07:11] Yep [07:12] We appear to have three subtly different focus mechanism classes in Shell :/ [07:13] Awesome. More the merrier [07:13] Or is that backwards? [07:17] Argh. How can building the archive version have missing symbols when my machine is up to date? [07:21] *Fails* with missing symbols? [07:21] (Also, what do you mean by missing symbols?) [07:25] RAOF: X: symbol lookup error: /usr/lib/xorg/modules/drivers/intel_drv.so: undefined symbol: xmir_window_to_windowptr [07:26] Hrm. Odd. [07:26] * duflu rebuilds the intel driver [07:26] Is xmir successfully loading/? [07:26] Because that woud happen in the past when xmir failed to find mir, and then would unload itself. [07:26] But I thought I made those errors fatal. [07:28] RAOF: Yes, now. But my packages are kind of screwed now. Trying to fix [07:34] RAOF: Oh. Someone has been dputting without updating the lp:ubuntu/.... [07:34] lp:ubuntu/ is automatically maintained. When it works. [07:35] Yeah it's quite behind right now [07:35] A few weeks [07:36] Maybe apt-get source will give me the actual correct code... [07:36] It shall. [07:51] * mlankhorst pokes tvoss with a glmark2 deb [07:51] mlankhorst, yeah :) [07:53] mlankhorst, sent [07:54] mlankhorst, sent you armhf, too [07:54] mlankhorst, not sure you need them, though [08:03] RAOF: Unexpectedly I appear to have fixed the artifact'd cursor [08:13] duflu, how so? :) [08:13] tvoss: Maybe I just hid it by going slower. Added more mir_wait_for in xmir [08:14] I'm still concerned by XMir's apparent lack of synchronization though [08:54] duflu: What were you mir_wait_foring? [08:57] RAOF: The swap buffers [08:57] Oh, yeah. That'd be adding significant wait. [08:58] RAOF: Cos there doesn't seem to be any guarantee otherwise that mir_surface_get_current_buffer is getting for the correct frame [08:58] Nothing should call that until has_free_buffer is true, which is set in the buffer_received callback. [08:59] And we should only ever have one swap_buffers in flight at any one time. [08:59] (per surface) [09:01] RAOF: OK, will have a closer look next time I'm in XMir [09:23] Ping alan_g [09:23] duflu: good morning [09:24] alan_g: Hi. Something that's tripped me up a few times... the display server test fixture hangs if you try to swap buffers more than once. Any idea why? [09:25] duflu: yes [09:25] Umm, good. [09:25] duflu: It uses a stubbed graphics component that never consumes buffers [09:26] alan_g: I keep getting lost in the whole "this test is now three processes" :( Do you think you can propose a way to consume them? [09:29] Why does C++11 not contain features in C99? :( [09:29] duflu: I can add it to the list. (I'll probably need a fuller test graphics implementation for nested mir at some point) [09:30] RAOF: because they're bad ideas? [09:30] RAOF: Because the ideas are *too good* [09:30] Does anyone implement all C99 anyway? [09:31] RAOF: Consider it a fork with no recent common ancestor :/ [09:35] alan_g: I don't think designated structure initialisers are a bad idea ;) [09:37] RAOF, now that's a good idea, indeed. Have stumbled over that multiple times, too [09:40] RAOF: as best as I can tell the only reason is that no-one proposed adding them [09:40] * duflu also votes for dynamically sized arrays in the local stack frame [09:41] I think people raised that as a security risk though [09:45] alan_g: Strangely, multiple swaps started working recently (with the switch branch??) and we added a test case that depends on it. But now I'm working on new buffering changes, the new test case hangs (expectedly) [09:46] duflu: in that case I have no idea [09:53] alan_g: OK, I *think* I found a fix. Will propose it soon [10:03] Hi folks, I've an 26" Apple display in front of me that I want to use as second monitor with my macbook pro. When I plug it in, I get black screens, except for on my laptop where the cursor remains. If I try to move the cursor, it takes 2-5 seconds for the display to update. Ideas? [10:09] greyback: radeon by any chance? [10:09] duflu: nope, intel sandybridge [10:09] greyback: Anything chewing CPU while the cursor is slow? [10:10] duflu: any way I can tell, without needing a second device to SSH in? [10:11] note unplugging made everything ok again [10:12] greyback: With Mir you will often need a second device for ssh :( [10:12] greyback: It's only OK _while_ unplugged? [10:14] duflu: yes. Plugged in, laptop display black screen with cursor, Apple display unchanged, as if not plugged in at all. If I unplug, laptop display returns to normal [10:15] greyback: Normal X session you mean? [10:16] greyback: X will auto-enable remembered multi-monitor layouts when new monitors become available AFAIK. Maybe you should start fresh by removing ~/.config/monitors.xml [10:16] duflu: yes, I'm using XMir. Ok will try that. [10:17] * greyback just pulled updates, will reboot for safety [10:17] greyback: You're not the only one reporting a black screen. So it would be helpful to get to the bottom of [10:18] duflu: ok. Will let you know once I reboot. Also have my tablet here with terminal app, so should be able to ssh into my laptop [10:21] greyback: Is that soon? I am off in a minute... === mpt_ is now known as mpt [10:22] duflu: I'll probably be 5 minutes. No worries if it's your EOD [10:23] greyback: OK, I can easily track the issue if you log a bug :) [10:23] duflu: ack [10:25] tvoss: hm sent me where? [10:25] maarten.lankhorst@canonical.com [10:27] odd, didn't arrive then [10:27] oh nm i see it now [10:28] derp [10:31] hikiko: have you time to review https://code.launchpad.net/~alan-griffiths/mir/tidy-up-nested-class-responsibilities/+merge/183233? [10:32] yes alan_g [10:32] in a sec to finish something [10:49] tvoss: hm seems most likely that nouveau is syncing glmark2 to vblank :P [10:50] mlankhorst, oops :) [10:50] mlankhorst, however, having something like 60fps would already be better than last week, both for X and XMir [10:51] try glmark2 --off-screen === chihchun is now known as chihchun_afk === hikiko is now known as hikiko|lunch [12:02] RAOF, ping === alan_g is now known as alan_g|lunch [12:41] hi, i build mir locally, i got an error, "undefined symbol: eglCreateImageKHR", how can i resolve this === hikiko|lunch is now known as hikiko === alan_g|lunch is now known as alan_g [13:34] penghuan: I can only guess that you've not got the right dependencies installed in your build environment. You've followed these instructions? http://unity.ubuntu.com/mir/building_source_for_pc.html [13:52] tvoss, (and the Mir team), I can't seem to load lightdm using XMir [13:53] penghuan, do you have libhybris or libhybris-dev installed? [13:53] smartboyhw, not sure what you mean? what are the symptoms you see? [13:54] tvoss, after Plymouth, it went black [13:54] Well, seems like that I have to report a bug:( [13:55] smartboyhw, that's fine :) please make sure to attach the relevant logs including dmesg [14:00] tvoss, for dmesg, how can I find PAST logs? [14:00] Since I've now booted in using Xorg [14:03] smartboyhw, you cannot really resurrect it, sorry [14:04] smartboyhw, what kind of setup are you testing? [14:04] tvoss, so unfortunately, I can't include it then [14:04] tvoss, Ubuntu 13.10 [14:04] smartboyhw, in terms of hw? [14:04] tvoss, laptop (amd) [14:04] I mean, amd card [14:21] tvoss, Bug 1219847 [14:21] bug 1219847 in mir (Ubuntu) "XMir can not start LightDM" [Undecided,New] https://launchpad.net/bugs/1219847 [14:21] smartboyhw, thanks [14:21] But, I can't attach anything (in Firefox or Chromium) weird [14:28] I get bug 1218456 with xserver-xorg-video-ati [14:28] bug 1218456 in xorg-server (Ubuntu) "Xorg assert failure: X: ../../../../include/privates.h:123: dixGetPrivateAddr: Assertion `key->initialized' failed." [Critical,Confirmed] https://launchpad.net/bugs/1218456 [14:29] jibel, I assume you upvoted it? [14:31] tvoss, yes [14:32] jibel, thanks for reporting :) [14:33] tvoss, yw, let me know if you need more info, this is my main work machine :) [14:33] jibel, ;) === alan_g is now known as alan_g|tea === alan_g|tea is now known as alan_g [14:57] kdub, ping [14:57] hello tvoss [14:57] kdub, unping, it's labor day, isn't it? [14:59] it is, but i thought I'd quickly update myself and check our status after my time off :) [15:01] kdub, ff is passed, we do see some performance issues on the galaxy nexus [15:01] kdub, first of all: welcome back :) [15:02] thanks :) === 16WAAZ40T is now known as tvoss_ === alan_g is now known as alan_g|EOD