=== jderose1 is now known as jderose [05:51] good morning [06:53] good morning === rodrigo__ is now known as rodrigo_ === API is now known as apinheiro [08:37] Saviq: hi, how can you tell metacity to layout 9 workspaces in a 3x3 grid? I always get 2 rows, with 5 and 4 workspaces [08:37] mardy: I can't ;) [08:38] mardy: I hardcoded 3x3 in Workspaces.qml [08:38] Saviq: ahhh :-) OK, I'll try that as well [08:38] that's something we need to address, too - when _NET_DESKTOP_LAYOUT isn't set [08:38] we should have sqrt(workspaces) rows [10:49] Hi all, anyone here can answer specific question regarding unity-2d code? [10:53] eugenesan, Kaleo or mardy should be able to help [10:54] Thanks for tip, I suppose they are not here. [10:54] eugenesan: I'm here :-) [10:55] Oh, nice. [10:56] mardy: I am trying to revive homebutton and fighting next issue: HomeButtonApplet::enterEvent@homebuttonapplet.cpp called twice and I can't figure why. Any ideas why? [10:58] eugenesan: is it the home button applet for the top panel? [10:58] `yes, the one removed in recent versions === MacSlow is now known as MacSlow|lunch [11:03] mardy: the jumping background is a regression, I don't observe that in currently released version [11:04] I've to relogin to check something, brb [11:06] Saviq: I'm trying with the latest trunk, and it's there [11:07] mardy: yes [11:07] that's what I meant, a regression between current release and trunk, sorry [11:07] Saviq: ah, you mean that the current release works fine? [11:07] mardy: if you spread from a workspace with a maximized window, does your launcher still contain the underlying window? [11:08] mardy: yes [11:08] Saviq: yes [11:09] Saviq: might be because of that geometry change I made to ScreenInfo [11:09] do you remember if it wasn't like that before? [11:09] mardy: but that's also there in current release [11:09] can't remember :-( [11:10] bbiab === Saviq is now known as Saviq|lunch [11:23] eugenesan: I remember it was functional, but since it was removed I never tried to use it. Try to run the panel in gdb, and see who calls enterEvent [11:27] mardy: Thanks for the tip === Saviq|lunch is now known as Saviq [12:04] hey everyone! i cannot build unity, getting lots of "error: ‘nux::State’ has not been declared" (i'm on nux trunk). is this known? [12:05] htorque_: did you use the nux 2.0 branch? [12:05] oh, probably not. === MacSlow|lunch is now known as MacSlow [12:39] Trevinho: 'trunk' is actually the same as '2.0', but i guess recent nux commits cause the problem: http://paste.ubuntu.com/714142/ and http://bazaar.launchpad.net/~unity-team/nux/2.0/revision/504 [12:49] smspillaz: Have you ever had any issues with override-redirect windows painting? It seems all my menus and notify-osd windows aren't rendering correctly, almost as if they aren't getting damaged [12:50] Amaranth: nope. What graphics hardware ? [12:50] smspillaz: It's intel with my gles changes :P [12:51] In an hour or two I'll know if it happens on my panda too but I'm expecting it will [12:51] I did make a change to have it so that decorations would paint on windows that didn't receive a damage event yet [12:51] (so if an application mapped a window, never drew on it then locked up you'd be able to close it [12:51] I get a flash of them drawing then the only part that draws is the part I have highlighted, presumably because the color change on highlight triggers a damage [12:51] and notify-osd windows only show up when fading in or out [12:52] It's gotta be something I'm doing in unity, I'm afraid it's the FBO stuff [12:57] Amaranth: ah [12:57] Amaranth: yeah it would be [12:57] Amaranth: check if the fbo is still bound after the nux stuff draws [12:57] Amaranth: I remember I hit problems with this [12:57] Amaranth: you know, its funny, I read over the blur plugin code the other day and finally had the "so that's how it works! moment" [12:57] nuxPrologue is handling that, iirc [12:58] God blur would be so much simpler if it was using Nux APIs :P [12:58] Amaranth: check the viewports are reset correctly [12:58] Amaranth: it does use the nux api's. Kind of [12:58] Oh, right, I think I don't do that anymore, oops [12:59] I still have some pending SRU changes to that fbo code, but I don't think its relevant to what we're doing [12:59] I think that the best approach is when the blur plugin itself is ported, we should make compiz specify a screen region to blur [12:59] using the same shader [13:00] Amaranth: so I finally figured out that compiz does it by doing a horizontal blur on the backbuffer first, then doing a backbuffer-to-texture copy (eg glCopyTexSubImage2D) and then painting that texture into an fbo with a vertical blur shader [13:00] (took a bit of jumping around in the code to figure that out) [13:01] sounds...fun === m_conley_away is now known as m_conley [13:11] Amaranth: I dunno why it does it that way, reading off the backbuffer was hella expensive back in the days of not having a proper memory model [13:12] Amaranth: but then again, gaussian blur required fbos which required a memory model, and memory models made it about as expensive to do glCopyTexSubImage2D and glBindBuffer anyways (since you have to flush the pipeline in both cases) [13:16] hello im experiencing strange behavior of xfce4-panel and indicator-plugin > the indicators are in reverse mode , how can i change the diplaying? [13:16] (if displaying exists as word ...) [13:17] http://i.imgur.com/szLEx.png > top right check the system indicator(at left side of indicator [13:18] should i report a bug , against wich package ? [13:44] smspillaz: But nvidia was clearly doing FBOs without a proper memory manager [13:44] So that argument isn't entirely valid [13:44] Although it probably was for this particular case [13:44] Actually I think onestone wrote that while using fglrx so... [13:46] Amaranth: david wrote the gaussian blur code [13:48] Are you sure? [13:48] I thought blur was originally a beryl feature that required core changes and then onestone rewrote it for compiz in a way that didn't require those changes [13:49] also, ugh, unity takes forever to compile [13:51] smspillaz: hehe, resetting the viewport in nuxEpilogue actually makes things worse, now menus don't show up at all === zyga is now known as zyga-afk [14:07] Amaranth: the blur plugin for compiz was originally written by david. blurfx was dropped === API is now known as apinheiro [14:11] njpatel, how can i tell if "All" is active in a CheckOptionFilter? [14:11] or didrocks ^^ [14:12] 4menot4U [14:12] d'oh [14:12] there goes that password :( [14:13] greyback, hahaha [14:13] hate it when that happens :) [14:15] greyback: rotfl [14:15] kenvandine: at least it wasn't an important one *looks around suspiciously* [14:42] smspillaz: So if I just wanted to disable UnityFBO usage entirely what would I need to do? [14:44] Amaranth: replace the hook points ? [14:44] unityfbo is not that complicated [14:45] smspillaz: I must be missing something because when I disable using it nothing draws anymore [14:45] Oh, actually... [14:45] I just forced allowWindowPaint to false all the time, seems to have done the trick [15:00] Amaranth: that will force unity to go on top of everything [15:00] you probably dont want that [15:00] (unless you like it appearing on top of the lock screen too) [15:01] smspillaz: Actually other than showing the name of the app I'm running that's a rather nice effect [15:02] I can't interact with the top bar but I can see the time, date, wifi status, battery status, etc [15:02] Kind of like the gnome-shell lock screen, no? [15:06] Amaranth: that's great, except when the app name happens to be "the big world domination red button" or whatever [15:06] oh, and the launcher showing $not_child_friendly_app pinned to the launcher [15:07] the launcher doesn't seem to be showing at all for me [15:07] oh and then the subsequent political crisis and emergency that occurrs claiming that window stacking is broken again and OH MY GOD THE WORLD IS GOING TO END [15:07] on the lock screen, I mean [15:07] Amaranth: make it not autohide [15:07] I wish I could figure out how [15:07] move windows out of the way [15:07] smspillaz: There are no windows in the way [15:08] Except the screensaver window [15:08] oh that would do it [15:09] there we go, now it listen to my "autohide must die" preference [15:09] and now it shows on the screensaver [15:10] smspillaz: Got a better suggestion? I was supposed to have a release out already [15:10] Amaranth: can I see the code ? [15:10] smspillaz: lp:~linaro-graphics-wg/unity/unity-gles2 [15:10] thx [15:11] The unity plugin itself is running in to the same problem I had with the rest of compiz together [15:11] hm? [15:11] Too much state that you have to know to know what a given codepath is going to do [15:11] yeah, its [15:11] hacky [15:12] Doesn't help that I would normally be in bed a couple hours ago :P [15:12] * cwillu_at_work pokes lazy positioning with an smspillaz [15:12] mostly because there is no such support for inserting arbitrary obejcts into the paint stack [15:12] cwillu_at_work: I'm not even meant to be working this week :( [15:12] there's a pill for that [15:12] heh, I wish [15:13] hehe, you're not working, you're helping me ;) [15:13] well its more like [15:13] I don't know why Unity's FBO usage gives me so much trouble [15:13] people can still harrass me [15:13] I just can't wrap my head around what it does and where [15:13] but then I have a legitimate reason to tell them to go away [15:14] its like, the only difference is that being paid revokes that right to tell people to go away [15:14] you should log on with an appropriate nick when you're being paid === smspillaz is now known as smslave [15:14] very good. [15:14] hi [15:15] * cwillu_at_work nearly avoids making various snarky remarks about that nick [15:15] Amaranth: you should just be able to disable it by making bind () and unbind () return false [15:15] and paint () return [15:16] smslave: That should be the same as commenting out the lines those are called on [15:16] smslave: Which makes the screen stop painting [15:16] That's why I ported this stuff to nux rather than disable it and leave it for someone else to port later :P [15:17] ... so, why do we need to deactivate it again ? [15:18] smslave: Menus (and presumably all override-redirect windows) aren't drawing at all [15:18] smslave, the lazy positioning fixes aren't in -proposed/don't exist yet, correct? [15:18] * cwillu_at_work just wants to make sure it's not considered fixed, and will avoid pestering further if that's the case :p [15:20] Amaranth: http://bazaar.launchpad.net/~linaro-graphics-wg/unity/unity-gles2/view/head:/plugins/unityshell/src/unityshell.cpp#L462 [15:21] cwillu_at_work: can you show me the bug # ? [15:21] sec [15:21] smslave: You change it to mActiveFbo, not me :P [15:21] err, changed* [15:21] Amaranth: I added that becauset there was some nuxPrologue/nuxEpilogue wrappers around paintPanelShadow [15:21] so what unityfbo did was it set the "active framebuffer" in unityscreen to the currently bound framebuffer [15:22] smslave, bug #874514 is what I reported [15:22] Launchpad bug 874514 in compiz (Ubuntu) "Nothing on screen updates other than cursor when moving window" [Undecided,Confirmed] https://launchpad.net/bugs/874514 [15:22] so that when nuxepilogue gets called it will be bound back to the correct one or 0 when there is no buffer bound [15:23] Amaranth: since you don't update mActiveFbo anymore when binding the nux framebuffers, that line can be nuked [15:23] hmm [15:24] well, no, I can nuke mActiveFbo and set that to 0 though [15:24] cwillu_at_work: that should be fixed with the *next* proposed update (some event matching fixes came in there) [15:24] cwillu_at_work: the reason is because lazy positioning in its current form is really unsafe when we have made configure requests on that window [15:25] incidentally, I still get the configure notify missed message thingy occasionally under the current -proposed [15:25] right, those should be fixed with the *next* proposed (dunno when that was uploaded) [15:26] matching configure notify events is surprisingly hard [15:32] smslave: http://bazaar.launchpad.net/~unity-team/unity/4.0/view/head:/plugins/unityshell/src/unityshell.cpp#L954 [15:33] you set it to true then test if its true? :) [15:33] I sure hope the compiler optimizes that out [15:33] ret = gScreen->glPaintOutput(attrib, transform, region, output, mask); [15:33] stuff happens [15:33] namely [15:33] http://bazaar.launchpad.net/~unity-team/unity/4.0/view/head:/plugins/unityshell/src/unityshell.cpp#L649 [15:34] you didn't link to glPaintOutput code you linked to paintDisplay [15:34] which is what that if is guarding [15:35] Amaranth: http://bazaar.launchpad.net/~unity-team/unity/4.0/view/head:/plugins/unityshell/src/unityshell.cpp#L1642 [15:35] I have no idea what you're linking me and why [15:35] glPaintOutput -> base glPaintOutput -> UnityWindow::glDraw -> paintDisplay (if condition met) -> doShellRepaint = false [15:36] You set it to true, draw something if its true, set it to false, then another part of code checks if its true or false [15:36] so the idea is [15:36] set it to true [15:36] wait for a window to paint [15:36] if we need to paint directly below that window [15:36] paint [15:36] set it to false [15:36] I'm pretty sure this bit is the cause of my problem when turning off FBO usage [15:37] if it is still true, then we didn't paint [15:37] so paint on top of everything [15:37] mardy: regarding duplicated HomeButtonApplet::enterEven, looks like appearance of launcher causes additional event. [15:37] Amaranth: the cause of your problem was that [15:37] Amaranth: an invalid fbo was bound [15:37] so nothing was drawn [15:37] FBO 0 isn't invalid [15:38] invalid in the logistical sense [15:38] and then nothing would get drawn until the next paint pass when that state was released [15:42] smslave: hello from metacity [15:42] smslave: that wasn't the problem :( [15:43] afaik binding FBO 0 means frontbuffer [15:43] so that couldn't have been the problem [15:46] Amaranth: you should remove that line anyways [15:47] smslave: That line has existed since before your UnityFBO stuff [15:47] It just used to explicitly set to 0 [15:47] because nux paints stuff, it was used to reset state iirc [15:47] Because while blur and unityshell aren't doing something 0 is the wrong answer [15:47] err, 0 is the right answer [15:48] theoretically that line isn't needed because for every bind you do an unbind [15:48] but it doesn't hurt to have it and it protects against accidents [16:00] I think I'm just going to get some sleep and work on this again when I'm more alert [16:01] sounds like a plan [16:01] night! === MacSlow is now known as MacSlow|afk === zyga-afk is now known as zyga [16:22] is there a keypress to make the launcher bar go away ? [16:22] when someone says my name in irc, and i'm working, it pops up (which is nice) but sticks around too long. [16:22] as the xchat icon wiggles === MacSlow|afk is now known as MacSlow [16:49] kenvandine, filtering == false [16:58] there goes njpatel with a drive by answer, love that guy :) [17:08] although his answer is either not correct or i found a bug [19:19] the other day, I found a page with a description of how to use the launcher API. I can't find it again. Anyone? === yofel_ is now known as yofel [20:24] jo-erlend, https://wiki.ubuntu.com/Unity/LauncherAPI [20:40] someone is getting two launchers at the same time. Is that a known bug? [20:46] haha, he had used a screenshot as a desktop background :> [20:47] * cwillu_at_work hands jo-erlend a knife, to be used in whatever way comes naturally [20:49] actually, it was quite nice to be able to disprove a bug for once :> [23:03] is there a known bug about drag&dropping files form Nautilus to Launcher? I cannot drop a pdf :\ the çauncher keeps staying transparent [23:24] Andy80, yep; I believe there's a fix in -proposed === m_conley is now known as m_conley_away