[05:25] -GitHub[m]:#mir-server- **[MirServer/mir]** RAOF closed [pull request #2451](https://github.com/MirServer/mir/pull/2451): [autopkgtest] Add file missing from initial autopkgtest commit. [07:16] Hm. I'm *pretty* sure this `std::vector` shouldn't have a `capacity` of `-43`. [07:26] Hm. Why are you using the `DumbDisplayProvider`, code? [08:21] Good morning o/ [08:36] Saviq: do you think the same hack we used for mir-test-tools on Nvidia would work for Frame?(Re: https://forum.snapcraft.io/t/nvidia-drivers-in-a-mir-ubuntu-frame-snap/30364) [08:37] alan_g[m]: On Classic it would [08:40] It does require kms to be enabled on the kernel cmdline, that's it really. [08:40] Whether we want to open that can of worms with Frame I'm not sure [08:41] You did make an experimental branch including the eglstream platform, right? Just it didn't look for glvnd bits in the right places? [08:41] s/right/snap/ [08:42] Saviq: I did, updating that draft [08:43] I suppose we could make a `beta/nvidia` channel [08:44] With a YMMV caveat [08:47] Saviq: We could just try it. To prove the theory. (But I don't have any output connection to my Nvidia card for Mir to use before hybrid support lands) [08:56] I can do the lab dance [08:58] videos or it didn't happen ° [08:59] It's uploaded (amd64) as `edge/pr43` (https://github.com/MirServer/ubuntu-frame/pull/43) [10:45] alan_g where was your trick to avoid trailing colons? [10:46] Ah `${LD_LIBRARY_PATH:+:$LD_LIBRARY_PATH}` [10:46] Not my trick, just a :+ bash substitution [10:48] Oh! That was missing in the mir-test-tools hack? [10:48] Not that, but yeah, commit incoming [10:52] alan_g https://github.com/MirServer/ubuntu-frame/pull/43/commits/09f6cc275bb11c98ff3e06d86885014bf4c15dd2 [10:54] `libnvidia-egl-wayland1`? [10:57] https://packages.ubuntu.com/focal/libnvidia-egl-wayland1 [10:59] I'm just surprised we need it here (and not in mir-test-tools). Shouldn't the nvidia stack come from the host? [10:59] We do need it in mir-test-tools [11:00] And we have it there [11:00] But apparently as a dependency, not an explicit stage [11:01] Hmm mir-platform-graphics-eglstream-kms* pulls it in, why wasn't it there… [11:01] Ah. it is a dependency of mir-platform-graphics-eglstream [11:02] Sorry! We're down the rabbit hole. (I really didn't mean to make this a distraction for us both.) [11:09] Right, but it wasn't there in `edge/pr43` for some reason [11:18] Odd. When I build it locally it is there. [11:25] Also when I download (on amd64) [11:27] unsquashfs -d try-me ubuntu-frame_2494.snap... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/d54a3b1bb82214c3b8abfc201edb14f3b62ce7ad) [11:32] Yes, but not the JSON… https://packages.ubuntu.com/focal/amd64/libnvidia-egl-wayland1/filelist [11:32] But now I can't find it in any of the ubuntu-frame snaps I built o.O [11:33] You're looking at the wrong thing: [11:33] platform:... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/d884f7fdb298064da22ef3f35c61c6928c689132) [11:33] Note the prime doesn't select it [11:35] Yup. So that's what's missing, rather than the stage [11:36] Yup [11:36] Fixed [11:38] Not quite - you need to point __EGL_EXTERNAL_PLATFORM_CONFIG_DIRS at it too [11:39] Oh, you did [11:57] Except I didn't `export` that one. Just did. [11:58] The others were in the environment already, so didn't need exporting… [12:00] You should have just added it to the snapcraft.yaml [12:02] Could've, yes. OTOH the others could get that same treatment. [12:04] Though you have those conditional on the snapd dir [12:11] OK done, that's more appropriate since that's coming with the snap rather than from snapd [16:30] Yeah Kodi drops to llvmpipe, would need more digging [16:39] Nested mir-test-tools on Nvidia isn't working either:... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/d3a38167ea0e7baf70d414580b3040b3807309cd) [16:39] wl_display@1: error 1: invalid arguments for wl_eglstream_display@6.create_stream [16:39] Mir fatal error: Critical error in Wayland platform: /build/mir-7O9ED4/mir-2.8.0+dev210-g8baffa9609/src/platforms/wayland/display.cpp(275): Throw in function void mir::graphics::wayland::Display::run() [16:40] * Saviq sent a code block: https://libera.ems.host/_matrix/media/r0/download/libera.chat/3ebbf5e68377c09d6be09d00d9af5cf35f0aa1a3 [16:40] OK, at least we know there's a world of pain on that road [16:40] :this: [16:46] Time to make notes for the next traveller and get back to what you were doing before I distracted you [16:55] -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq opened [issue #2462](https://github.com/MirServer/mir/issues/2462): Snapped, nested on Nvidia/eglstream doesn't work... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/f74c1a83355123637d2a0ec66cb822ccccbdb7d8) [17:11] Good night o/ [18:47] -GitHub[m]:#mir-server- **[MirServer/mir]** graysonguarino pushed something to [pull request #2299](https://github.com/MirServer/mir/pull/2299): Filter 2 distinct bogus touch event scenarios [19:27] -GitHub[m]:#mir-server- **[MirServer/wlcs]** vberger opened [issue #234](https://github.com/MirServer/wlcs/issues/234): WLCS will happily bind a global version higher than the one defined in the wayland.xml it was compiled for... (full message at https://libera.ems.host/_matrix/media/r0/download/libera.chat/72eaea2bd003eccd3a83c5272ec244c91efde293) [20:46] -GitHub[m]:#mir-server- **[MirServer/mir]** graysonguarino closed [issue #2346](https://github.com/MirServer/mir/issues/2346): Support idle inhibit protocol [21:40] -GitHub[m]:#mir-server- **[MirServer/wlcs]** wmww opened [pull request #235](https://github.com/MirServer/wlcs/pull/235): Do not bind to interface versions newer than supported [21:40] -GitHub[m]:#mir-server- [21:40] -GitHub[m]:#mir-server- > Fixes #234 [21:48] -GitHub[m]:#mir-server- **[MirServer/mir]** wmww closed [issue #2312](https://github.com/MirServer/mir/issues/2312): Memroy leak on X11 platform window resize