[08:14] tsdgeos: Did you guys need to adjust U8 pointer acceleration for Mir 0.21.0 or not at all? [08:15] duflu: honestly i don't know, not a big user of u8 in the desktop myself [08:15] tsdgeos: Do you know which package to check? [08:15] I couldn't see any changes [08:17] I'm guessing there's a sprint on or coming up if certain people are away right now [08:17] Who else to ask... [08:18] duflu: mzanetti or ltinkl i guess [08:18] Ah yes [08:18] duflu: there's a management sprint in SLC this week i think yes [08:18] mzanetti: Did you guys need to change the pointer acceleration for Mir 0.21.0? [08:19] duflu, can I test mit 0.21 somehow easily? [08:19] duflu, only slightly [08:19] ltinkl, ah thanks [08:19] mzanetti: Yes, on a xenial desktop [08:19] it's been already released, right? [08:19] Yep [08:19] Last weekend [08:19] on the phone I mean... because there's a difference between the phone and the pc in how u-s-c is started [08:20] so on the desktop our settings are not applied [08:20] mzanetti: That's OK. I just thought it weird the desktop is *fixed* but only Mir had to change [08:20] It's not over-accelerated as I expected [08:20] the desktop was broken before (as in sloooow pointer) [08:20] Yep === sil2100_ is now known as sil2100 [08:21] mzanetti: Fixed last weekend [08:21] we only adjust the pointer settings in u-s-c, which only works when u-s-c is on the system bus [08:21] => on the phone/pocket desktop devices it might overaccelerate indeed now [08:21] Oh, I haven't checked pocket desktop since Mir 0.21 [08:22] * duflu is reminded to herd more libinput bugs [08:24] duflu: playing a bit with MIR_CLIENT_PERF_REPORT=log i get extremely low values like 3.06 FPS when doing very small drags on the view, is that expected because on how it works? [08:24] tsdgeos: Yes, that's right. A very small drag might only require 3 frames rendered [08:24] We idle when we can [08:25] In click-button scenarios you aim for a mouse click and app change to only require one frame :) [08:26] Then the system sleeps again [08:27] Although I endorse animations because they're pretty. So multiple frames then [08:27] i guess it makes sense from the "we only painted 3 frames so 3FPS" but it doesn't seem very useful for what i'm trying to do since otherwise we'd think it's being extremely slow in rendering [08:28] tsdgeos: Well the perf report is non-invasive, meaning it does to affect the numbers being reported. That can be a challenge otherwise, especially if displaying performance on the screen [08:29] See Compiz Benchmark plugin for an example of solving that problem [08:29] * duflu assumes it's still there [08:29] meaning it does *not* affect... [08:33] tsdgeos: For your slow scenario try extrapolating from the render time -- FPS = 1000/render_time [08:33] Hmm, I might add that as standard [08:34] Although it's slightly less accurate [08:36] ltinkl: Where did you change acceleration? [08:37] duflu, in system settings I believe, are you saying it won't take effect on desktop? :) [08:38] ltinkl: Ah kay thanks. I was confused when the unity8 package was unchanged [08:39] ltinkl: Weird. I see two sliders and neither changes anything [08:40] duflu, right, indeed, must have been some placebo effect :) [08:40] ltinkl: Have you tested it on desktop? [08:40] With USB mice? [08:41] duflu, yeah, I just did, USB mouse, it indeed has no effect [08:41] Also why two sliders? :) [08:41] duflu, no idea :) [08:41] Left hemisphere placebo, right hemisphere placebo [08:41] :) [08:42] duflu, ah right, I know what happened - I fixed my pointer speed using the mouse's dedicated "speed" buttons [08:42] Oh fancy [08:45] duflu, it has +/- buttons, looks like they increase/decrease the DPI: http://download01.logitech.com/support/13135.1.0.pdf [08:46] As if guessing the DPI in software wasn't hard enough [09:43] tsdgeos: regarding bug #1563320, the backport from 5.6 applies to 5.5 fine, but there is no hasAtlasTexture function at all in 5.4. care to take a look? there are some isAtlasTexture calls there, but I'm not sure if eg line 140 of src/quick/items/qquickshadereffectnode.cpp or line 715 of src/quick/items/qquickimage.cpp would benefit from having the extra check for "t" or not [09:43] bug 1563320 in qtdeclarative-opensource-src (Ubuntu) "[PATCH] Fix crash in hasAtlasTexture" [Undecided,Triaged] https://launchpad.net/bugs/1563320 [09:44] Mirv: sure [09:59] Mirv: seems not needed in 5.4 [10:30] tsdgeos: so hmm did you get the crashes in 5.5/xenial only then? I mean, xenial phone hasn't even been bootable lately, but maybe you developed on xenial x86 then when you got the crashes? [10:31] Mirv: yes, i was using my desktop and it crashed [10:31] tsdgeos: oh, well that makes things easier then. [10:32] tsdgeos: can you test silo 061 in that case on your desktop? [10:32] Mirv: sure [11:15] greyback, do you know the status of this? https://code.launchpad.net/~unity-team/qtmir/inline-gles-surfaces/+merge/291156 [11:16] greyback, it's been in 13 already, I guess it's ok to land it. mind giving it a quick look and approving? [11:17] mzanetti: I don't know the exact status, but the idea is sound. I never tested the gles packages though [11:17] mzanetti: so yeah, let's add it, but we should check the emulator while testing [11:17] greyback, it seems odd to me that we have to keep a copy of debian/rules though [11:17] like, we could place bets when it runs out of sync [11:18] it's not copying, it's patching the existing debian/rules [11:18] right... didn't see that it's a diff in the diff :D [11:18] still, it kinda has the complete file [11:18] yeah, ++ hard to see [11:19] mzanetti: well, if rules changes, the patch will fail, so we'll be forced to fix it [11:19] fair enough [11:19] but yeah, the general idea really fixes an issue. I like it [11:20] but it's marked WIP still [11:20] should I drop it from the silo for now? [11:20] or adding it as-is? [11:20] mzanetti: let's add it [11:21] kk [11:21] I've just marked it approved :) [11:21] building the silo now. I think I've added all prereqs [11:21] we can remove it if we find something wrong [11:21] cool === alan_g is now known as alan_g|lunch [12:18] Mirv: yes, confirmed it fixes the crash [12:39] tsdgeos: great. do you think it should be pushed to xenial archives (ie Kubuntu etc users too) or is overlay & xenial+1 enough? [12:39] Mirv: it is definitely a crash that can happen anywhere, not sure how common it is but it won't hurt to have it in xenial [12:40] otoh if you have to fight zillions of burrocracy just put it in overlay [12:41] tsdgeos: ok. no I don't need to fight bureacracy but it will be up to release team whether to accept it or not from their freeze queue. let's see. thanks. [13:15] hi, i'm trying to debug bug 1540400 [13:15] bug 1540400 in owncloud-client (Ubuntu) "non icon in the unity system tray" [Medium,Confirmed] https://launchpad.net/bugs/1540400 [13:15] update to latest upstream didn't help === alan_g|lunch is now known as alan_g === dandrader is now known as dandrader|afk === dandrader|afk is now known as dandrader === dandrader is now known as dandrader|lunch [15:23] anyone? [15:23] it's essentially useless right now, because there's no way to edit settings or anything [15:39] bregma: can you help tjaalton at all? === shuduo is now known as shuduo-afk [15:51] tjaalton, Unity does not have a system tray, if an app is designed to work only with a system tray, it's not compatible with Unity [15:52] it used to work [15:52] in vivid at least [15:53] and still does [15:53] bregma, this should work, at least it does with e.g. telegram [15:53] (just tried) [15:54] I'll try the newer version on vivid.. [16:03] works === dandrader|lunch is now known as dandrader [16:14] it doesn't need to be configured to trigger the issue. "the icon area" should show a cloud shortly after launching it, when the config dialog pops up === alan_g is now known as alan_g|EOD === dandrader_ is now known as dandrader === dandrader is now known as dandrader|afk [19:31] Saviq: did design ever get anything to us for rotated/convergent greeter? Last I saw was that it was coming mid-last-week === dandrader|afk is now known as dandrader [20:13] mterry, I think Saviq is on holidays this week