=== RAOF_ is now known as RAOF [03:42] awesome, got the transition stuff hooked up with with the compiz launcher [03:44] :) [03:56] DBO: ping [03:57] DBO: I don't see this wall issue here (intel) [03:57] DBO: due to some driver bug I can't get unity running on this nvidia laptop (it's my fault, fubar'd) [03:57] DBO: could you pastebin the output of compiz in a terminal when it happens? maybe the plugins are loading a weird oorder [03:57] DBO: basically, the unityshell plugin should always come last, no matter what [03:59] erm well it should always come last in theory, but due to the way you've written it it needs to come before scale and expo (I'll need to fix that, you are initializing actiosn all wrong) [04:00] bbl 1h [05:40] smspillaz, you here? [06:01] DBO: pong [06:01] i figured out the problem [06:01] you are right [06:01] DBO: yeah sorry, I would have been here except that my internet went down [06:01] it has to do with transformed painting [06:01] DBO: yeah thought so [06:01] we must respect the clip mask [06:01] so we get multiple render passes now (annoying but okay) [06:02] which will... be interesting to try to make work [06:03] DBO: didn't just putting a if (doShellRepaint) paintDisplay (region); in glPaintTransformedOutput () work? [06:03] yes and no [06:03] it gets rid of the flicker [06:03] glPaintTransformedOutput can sometimes get called instead of glPaintOutput [06:03] however it comes at the cost of multiple rendering on top of each other [06:03] no [06:03] glPaintOutput gets called [06:04] ah right, yes you are right [06:04] my bad [06:04] then glPaintTransformedOutput gets called inside [06:04] I'm pretty sure there was a corner case where only one of the two would get called [06:04] DBO: in any case, I don't see the issue- try loading unityshell last [06:04] well third last [06:04] scale and expo last [06:05] why? [06:05] it puts it at the back of the paint queue [06:05] really... [06:05] yes [06:05] that's how WRAP/UNWRAP works [06:05] well, the compiz++ equavilent anyways [06:07] smspillaz, yeah no dice [06:07] still screws up [06:08] even with unityshell very last? [06:08] whoa [06:08] wait [06:08] you paint in transformed [06:08] and put unity last [06:08] and it works [06:08] fuck [06:09] wooo! [06:09] no wait [06:09] nope [06:09] broken [06:09] :( [06:09] you had my hopes up there DBO [06:09] sorry [06:10] it's weird [06:10] because it works fine her [06:10] *here [06:10] smspillaz, you have to open something that nux paints under [06:10] like a notification [06:11] if only notify send didn't segfault here [06:11] maybe an update fixed [06:11] * smspillaz updates [06:12] DBO: I got the settings migration working btw [06:12] DBO: what exactly is the issue when you switch with wall? [06:12] it paints twice? [06:13] well when I put it last [06:13] wall paints on top of unity [06:13] so you get a swipe effect [06:13] that should not happen [06:14] unityshell last = unityshell paints last [06:15] no [06:15] because the transformed paints [06:16] it should still paint last [06:16] since it's at the back of the glPaintTransformedOutput queue [06:16] unless [06:16] wait [06:16] unless some plugin is calling glPaintTransformedOutput inside of it's own glPaintOutput [06:16] and that would cause our glPaintTransformedOutput to get called [06:17] although plugins that do that are stupid [06:17] they should set PAINT_SCREEN_TRANSFORMED_OUTPUTS_MASK [06:18] smspillaz, thats exactly what is happening [06:20] ugh fuck me [06:20] I hate this stupid design [06:20] I'm going to send davidr a bill [06:42] DBO: what happens when you remove the paint calls in glPaintTransformedOutput ? [06:42] DBO: actually [06:42] it paints on top [06:43] DBO: in glPaintTransformedOutput make sure you check AFTER calling gScreen->glPaintTransformedOutput in that function [06:43] if we could tell when a paint transformed output was going to happen [06:43] that will paint on top [06:44] the problem is painting under a window (when the window test hits) [06:44] otherwise it works [06:53] morning [07:03] DBO: yeah [07:04] yeah? [07:04] DBO: we should catch transformed screen and not paint under windows when that happens [07:04] that's fairly easy to do [07:04] it will still double paint [07:04] or wait... [07:04] how come [07:04] i dont know [07:04] but okay [07:04] it only works if we load AFTER wall [07:05] DBO: screen -> transformed -> windows [07:05] I'll look into that for you [07:05] also [07:05] go to bed [07:05] ;-) [07:06] hold on [07:06] I almost have it working [07:06] hm? [07:06] I would be working on it, except that I'm waiting for this dist upgrade to finish [07:07] argh wtf is causing this [07:08] DBO: have you noticed a weird hang when you do metacity --replace && [07:09] yeah it happens [07:09] argh [07:10] since I checked with gdb [07:10] and I get this [07:10] ?? () [07:10] ?? () [07:10] ?? () [07:10] okay I have a fix [07:10] Corrupt Stack ? [07:10] awesome [07:10] it works now [07:10] cool [07:10] yes thats a corrupt stack [07:10] awesome, what's your fix? [07:11] basically we *dont* paint in transformed [07:11] but if we see a transformed paint call [07:12] we disable the window matching paint [07:12] and force a paint at the end of the cycle [07:13] sure, that would work [07:13] yeah [07:13] its what you suggested [07:13] transform is a weird situation [07:13] and transient [07:37] hmm I kinda think my disk is failing.. [07:37] my computer has been acting reallll wonky tonight [07:37] fans going nuts, db access being very slow [07:39] GOOD THINK IT'S ALMOST BLACK FRIDAY? [07:42] And the black friday super lenovo sales? :) [07:44] (too bad I spent all my monies on muse tickets) [07:45] [07:49] smspillaz: Have you been to one of their live shows before? They're pretty kickarse. [07:49] RAOF: I went in january to BDO [07:49] RAOF: mind = blown [07:49] RAOF: I was at the front [07:49] And that's a pretty half-arsed Muse show :) [07:49] I knoq [07:49] They can't do all their fancy lightshow stuff at the BDO :) [07:49] BUT I am going in december [07:50] My brother's going up to Sydney to catch them; I presume they're also playing in Perth? [07:50] at least they have some sense to not reject perth like 99% of all other bands [07:50] I don't know why bands would reject Perth - you've got an awesome music scene. [07:50] I know :( [07:50] I was going to go to BDO this year [07:51] and then crystal castles wasn't coming to perth [07:51] Heh. The Dandy Warhols came all the way down to Hobart this time :) [07:51] so I just thought "argghghghg!" [07:51] RAOF: :) [07:51] RAOF: yeah, poor hobart [07:51] Well, the Cat Empire have played here at least *twice*, so it's not all bad :) [07:51] :) [07:52] Now *there's* a band to see live! [07:54] good morning [07:56] ♪ You look good / Good morning / I'm up before the sun can breeeaaaaak ♫ [07:58] thanks for the musical morning RAOF :) [07:58] Everyone needs Good Morning by the Dandy Warhols as their alarm music :) [07:59] :) [08:00] * lamalex just saw them last week or so [08:00] well, my alarm clock was compiz plugin main FTBFS on the buildd. Of course, it built perfectly yesterday evening here… [08:14] didrocks: I got the settings migration working ^.^ [08:14] looking into this hang on --replace now [08:14] smspillaz: huhu, nice! [08:14] smspillaz: I didn't reproduced it with the glibmm branch now [08:15] reproduce* [08:15] ah ok [08:15] smspillaz: you still get it reliably? [08:15] it must be an ABI mismatch thing [08:15] didrocks: yes [08:15] didrocks: I [08:15] I'm rekompiling now [08:15] smspillaz: on, in that case :) [08:15] smspillaz: btw, the main plugin merger is broken [08:15] didrocks: hm? [08:15] smspillaz: the git is empty [08:15] didrocks: do git submodule init git submodule update [08:16] ahah :) [08:16] ok, will know for next time [08:16] right now, I carry those as a patch for the 3 submodules [08:16] but it FTBFS on the buildd and works there: http://launchpadlibrarian.net/59575447/buildlog_ubuntu-natty-i386.compiz-fusion-plugins-main_0.9.2.1-0ubuntu3_FAILEDTOBUILD.txt.gz [08:16] smspillaz: any idea? ^ [08:17] /build/buildd/compiz-fusion-plugins-main-0.9.2.1/wall/src/wall.cpp:591:15: error: 'class CompScreen' has no member named 'grabbed' [08:17] I don't think I've a more advanced version of compiz-core itself locally [08:18] didrocks: sounds like it's picking up 0.8 headers? [08:19] hum? do you think, let me check… [08:19] * DBO dances [08:19] didrocks: oh wait [08:19] didrocks: no your core is too old [08:19] Get:84 http://ftpmaster.internal/ubuntu/ natty/main compiz-core i386 1:0.9.2.1+glibmainloop-0ubuntu3 [377 kB] [08:19] or I forgot to push something [08:19] smspillaz: hum? why did it built there? [08:20] * didrocks puzzled [08:20] since I added ::grabbed () to CompScreen to make wall build [08:20] err [08:20] * DBO waves at didrocks [08:20] to make wall edges thing works [08:20] hey DBO! How are you? [08:20] I implemented awesome, again [08:20] smspillaz: ok, I should see a doctor then, I get it built locally with no issue… and I don't think I updated my local compiz-core [08:20] thats 2 nights in a row [08:21] DBO: ahah, the next release is going to get it? :) [08:21] DBO: win [08:21] didrocks, when is the next release? [08:21] oh crap :) [08:21] smspillaz: ok [08:21] smspillaz: I have the glibmm branch installed [08:21] smspillaz: so, I have the new method to CompScreen… [08:22] all is explained then, ok :) [08:22] DBO: depends on how many beers you can promess :) [08:22] DBO: more seriously, I think in 4/5 hours [08:22] uhm... [08:22] greetings [08:22] MacSlow, want to do a merge? [08:22] DBO: yeah [08:22] hey MacSlow [08:23] MacSlow, https://code.launchpad.net/~canonical-dx-team/unity/unity.autohide-like-a-boss/+merge/41813 [08:23] ok, getting the gsettings things finished now that njpatel merged is branch :) [08:23] his* [08:23] DBO, I've to finish something else first [08:23] alrighty [08:24] hum, bzr pull -> nothing new in trunk [08:24] I bet I saw a "merged" set manually by njpatel [08:24] he lied! [08:24] :) [08:26] yeah, it's not in trunk (https://code.launchpad.net/~unity-team/unity/favorite-store-update/+merge/41783) [08:27] didrocks, can you merge that then? [08:27] DBO: well, he fixed but didn't push his fix I think [08:27] ah [08:27] that liar [08:27] DBO: so the old branch seems to be the one you refused [08:27] he lies [08:27] likea rub [08:27] :) [08:27] like a rug [08:28] let me insult him on the merge :) [08:28] (badly of course) [08:29] smspillaz: btw, it seems you didn't store your canonical email adress to your launchpad account (there is no match on your name: https://code.launchpad.net/~unity-team/unity/trunk) [08:30] didrocks: weird [08:30] smspillaz: just add your canonical email address to your launchpad page: https://launchpad.net/~smspillaz [08:30] that will fix it :) [08:34] im so tired [08:35] DBO: go to bed, easy fix :) [08:35] didrocks, can you test my branch [08:35] please [08:35] didrocks, just confirm it works: https://code.launchpad.net/~canonical-dx-team/unity/unity.autohide-like-a-boss/+merge/41813 [08:37] DBO: sure, no requirements on nux/bamf? [08:37] didrocks, no [08:37] (still have last week version) [08:37] nice [08:37] just make sure to fresh cmake [08:37] sure [08:38] smspillaz: just to confirm, local extension (in ~/.compiz-1) are preferred on system ones, right? [08:40] didrocks: yep [08:40] smspillaz: thanks :) [08:42] i swear to god I am losing my sanity much faster than predicted [08:42] DBO: well, we noticed that for you long ago… :) [08:42] * didrocks runs [08:43] your branch look very nice! [08:43] thank you [08:43] did you try minimizing? [08:43] I like it :) [08:43] oh no [08:43] excellent! [08:43] its designed to work with the zoom animation [08:43] also make a window go urgent [08:43] that has a fun effect as well [08:44] oh nice as well! [08:44] didrocks: I fixed the hang btw, pushing [08:44] also, there is this "law-fitt" branch merged in it [08:44] smspillaz: waow, already? [08:44] \o/ [08:44] yeah turns out we were freeing shit while it was being executed [08:45] didrocks, no that was a trunk merge [08:45] DBO: it's nice in any case :) [08:45] :) [08:45] DBO: just a note: I was seetings it hilighted the launcher background for unstarted app [08:45] DBO: it doesn't make it anymore [08:45] what? [08:45] yeah, I get some kind of "light on and off" [08:46] probably just a bug, I'll file one when I can reproduce it reliably [08:46] oh what you are seeing is startup notification [08:46] so there is a "starting" animation [08:46] it slowly fades in and out [08:46] DBO: ok, maybe it was the startup notification then [08:47] close evolution then start it again (takes about 2 seconds) [08:47] you can see it [08:47] does the minimize action point to the launcher? [08:47] yes [08:47] DBO: the launcher is polished with those effects [08:47] I love this shit [08:47] this is great. [08:47] I am going to do intelihide tomorrow [08:48] DBO: it's much nicer than the spinning cursor alone :) [08:48] intelihide? [08:48] so in autohide right now [08:48] because you removed the timeout? [08:48] it is hidden unless you move the mouse over the BFB [08:48] in intellihide [08:48] it will unhide itself whenever there is no reason for it to hide [08:48] like when there are no open windows [08:49] or when there are no windows obstructing the launcher [08:49] hence [08:49] intellihide [08:49] hum, nice idea :) [08:49] yeah [08:49] i did it for docky [08:49] everyone loved it [08:50] nice (sorry, I used cairo-dock :)) [08:50] when I use some dock, which was… 30 minutes of my life? [08:50] didrocks, I understand, you dont want to choose between me and neil [08:50] well, 35 with awn and playing/spreading the icons :) [08:50] anyhow [08:50] i really feel loopy [08:50] so [08:50] goodnight [08:50] morning [08:50] whatever [08:50] DBO: you hadn't this cool and unuseful efffects with icons! [08:51] :) [08:51] DBO: good night dude [08:51] DBO: nice work! [08:51] cheers :) [08:51] oh [08:51] please poke johnlea about these changes if you see him [08:51] MacSlow: but I have a bad announce for you: still no tooltip with this branch [08:51] he'll be interested to hear about it [08:51] DBO: sure, will do [08:51] DBO: when did you merge trunk? [08:52] didrocks: What was the glibmm problem? Can I help? [08:52] smspillaz too [08:52] murrayc: wasn't related to glibmm, the issue was in the glibmm branch with compiz :) [08:52] ok, the 24 [08:52] so yesterday [08:52] MacSlow: so, with yesterday trunk, still no tooltip/ql [08:53] didrocks: compiz is using glibmm? compiz is C++? [08:53] murrayc: since 0.9, yes, smspillaz rocked this [08:54] s/me/onestone and me/ [08:54] Any particular reason? [08:54] And what does it use from glibmm? [08:54] because C++ is awesome [08:54] oh glibmm is a lot easier to integrate into existing classes [08:54] Has the compiz maintainer changed? Or was it always C++? [08:54] because you don't have all that gboolean/bool type conversion nightmare [08:55] gboolean/bool is not a type conversion nightmare. [08:55] (I am the glibmm maintainer) [08:55] murrayc: no, davidr stoped developing it after version 0.5, danny, dennis and (recently I) picked it up since then [08:55] murrayc: hi! [08:55] * hyperair still believes smspillaz debugged something wrongly and got the wrong impression. about gboolean and bool. [08:55] Ah that makes sense then. [08:56] Is that branch not here? http://cgit.compiz.org/ [08:56] murrayc: we were hitting stack overflow issues when doing foo boost::fucntion .callback (); return foo [08:56] murrayc: let me grab it for you [08:56] smspillaz: i still haven't seen that interesting use case where gboolean -> bool goes from false to true. [08:56] smspillaz: Sounds rather strange. [08:56] hyperair: murrayc: it was a weird race condition / stack overflow thing that only happened very rarely [08:56] smspillaz: So you are using boost::signals instead of libsigc++? [08:56] murrayc: yes [08:57] murrayc: boost::function rather [08:57] * hyperair wonders why [08:57] smspillaz: Doesn't sound like something that bool would help with much. [08:57] because boost is made of win? [08:57] Boost has an unstable API/ABI. [08:57] smspillaz: but boost's signals are slower than libsigc++'s [08:57] hyperair: That's not necessarily true. [08:57] And not very relevant. [08:57] murrayc: oh, this was a problem when moving from the C based glib (sizeof gboolean != sizeof bool) [08:57] murrayc: it was benchmarked to be true. [08:58] hyperair: we don't use boost::signals [08:58] smspillaz: but that really shouldn't matter. i don't see how false => nonzero int [08:58] hyperair: we used boost::function [08:58] gtkmm/glibmm would use boost signals if the API/ABI was stable. [08:58] hm that's interesting [08:58] murrayc: actually as soon as C++0x hits (hurry up!) you won't need boost [08:58] since C++0x is practically just libstdc++ + boost [08:59] smspillaz: No, signals will _not_ be in C++0x. [08:59] well, signals wont be there [08:59] but function objects will be [08:59] Is Unity still going to use compiz? [08:59] yes [09:00] problems? [09:00] I'm just surprised at the rather arbitrary changes in an important project. Nevermind. [09:00] murrayc: ah ok. it was for a number of reasons [09:01] And I think you'll have a world of pain with boost. [09:01] boost is quite nice [09:01] we don't use it in unity [09:01] but we do use it in compiz [09:01] API and ABI. [09:01] (only the very basic bits though) [09:01] like the bits they haven't changed for years (eg boost::function)_ [09:01] Or can you just copy in the files for boost::function? Maybe it has no library. [09:02] murrayc: how unstable is their A[PB]I anyway? [09:02] it's just templates [09:02] murrayc: and at least the header-only libraries are pretty safe to use imo [09:02] hyperair: Oficially: completely. [09:02]