| alan_g[m] | Saviq ubuntu-frame-20-edge hasn't picked up the latest ppa:mir-team/dev updates. Anything to do with disabling the USN rebuilds? | 08:15 |
|---|---|---|
| Saviq | Yes, it does. Let me kick it once. | 08:16 |
| Saviq | Conan Kudo hey, with https://github.com/MirServer/mir/pull/2822/files#diff-d8b32639e87ac44fb9220e5b3cc417ea7ac581e73c5b3994bce1539349d727a7 I'm still seeing gcc-12 being used, did you explicitly pick gcc-13 up somehow to encounter the missing include? Or did the upgrade to rawhide not work? | 11:34 |
| Saviq | How can I confirm things are coming from 38? I can't see no .fc38 packages installed | 11:34 |
| -GitHub[m]:#mir-server- **[MirServer/wlcs]** bors[bot] merged [pull request #267](https://github.com/MirServer/wlcs/pull/267): Update wayland.xml | 12:14 | |
| Eighth_Doctor | <Saviq> "Conan Kudo hey, with https://..." <- Sounds like the upgrade didn't work | 12:44 |
| Saviq | Conan Kudo here are the logs from the process as above: https://pastebin.ubuntu.com/p/tftCr45zC2/ | 12:48 |
| Saviq | I have a container open after this so can introspect… should I reboot? 🤷 | 12:49 |
| Eighth_Doctor | The rawhide repo isn't enabled apparently | 13:07 |
| Eighth_Doctor | Try adding --enablerepo=rawhide to the last command | 13:08 |
| Saviq | That looks like it's doing more, thanks! | 13:10 |
| Saviq | And yes, fc38 packages coming in | 13:10 |
| Saviq | Oops. | 14:24 |
| -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq opened [issue #2823](https://github.com/MirServer/mir/issues/2823): VirtualPointerV1Test.when_virtual_pointer_scrolls_with_steps_client_sees_axis_descrete failing... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/19742d338388ea6b91600aeaff9caad64b1286d8>) | 14:26 | |
| alan_g[m] | Oh, so more renaming needed? | 14:34 |
| grayson-g[m] | <alan_g[m]> "Oh, so more renaming needed?" <- If this is referring to the bug above, it doesnt seem like a “virutal” misspelling issue. It’s got something to do with the introduction of smooth scrolling. | 15:20 |
| Saviq | grayson-g but that's something that only started failing today - and we haven't touched smooth scrolling recently did we? | 15:26 |
| Saviq | I imagine there could have been a skip encoded somewhere… but couldn't find it | 15:33 |
| Saviq | Unless I'm missing something? 🤷 | 15:35 |
| Saviq | I mean, why wasn't it failing in our CI before the wlcs typo fix? | 15:35 |
| alan_g[m] | Try a revert just to confirm | 15:36 |
| Saviq | Yeah doing now | 15:36 |
| Saviq | Oh it's the protocol.xml update 'innit | 15:45 |
| Saviq | Yeah it is. That makes more sense. | 15:47 |
| -GitHub[m]:#mir-server- **[MirServer/mir]** Saviq edited [issue #2823](https://github.com/MirServer/mir/issues/2823): VirtualPointerV1Test.when_virtual_pointer_scrolls_with_steps_client_sees_axis_descrete failing | 15:47 | |
| Saviq | Probably need to revisit https://github.com/MirServer/wlcs/blob/f2161a11c57529f40b3ae208bfca151ed381ba1c/spread/build/ubuntu/task.yaml#L28-L33 to avoid this in the future | 15:48 |
| -GitHub[m]:#mir-server- **[MirServer/wlcs]** Saviq opened [issue #271](https://github.com/MirServer/wlcs/issues/271): We're not considering test failures a problem... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/c8b76709c7dc83cf205f7e164f0bbcb4547547c6>) | 15:50 | |
| alan_g[m] | Test failures are not necessarily a problem (it could be Mir that is wrong). But they need to be explained | 16:09 |
| Saviq | Sure, I'm not saying we should just flip it, but rather find a way to manage known/expected failures | 16:10 |
| Saviq | I suppose that's what the TODO there is | 16:10 |
| grayson-g[m] | <alan_g[m]> "Test failures are not necessaril..." <- My suspicion (I didn't get to look too deeply into this, but it may help someone debugging) is that, to my memory, the protocol now prefers sending value120 scroll events instead of discrete events if the server supports it. Now that the updated `wayland.xml` supports smooth scrolling, we're probably sending value120 events instead of discrete, so I believe this means Mir has the | 16:58 |
| grayson-g[m] | proper behavior but the test fails as it's still expecting the old format. | 16:58 |
| grayson-g[m] | Take everything I just said with a grain of salt ;) | 16:58 |
| alan_g[m] | Yes, it is likely that the tests are right for the version they were written to, but wrong for the latest version. (I.e. we need different tests for different versions.) | 17:03 |
| alan_g[m] | For the moment though, reverting the breaking PR is the way forward | 17:03 |
| -GitHub[m]:#mir-server- **[MirServer/wlcs]** Saviq drafted [pull request #272](https://github.com/MirServer/wlcs/pull/272): Update wayland xml | 17:21 | |
| grayson-g[m] | GitHub[m]: I'll take a look at #267 today. I think I know what it takes, I just didn't get finished with it by yesterday. | 17:26 |
| grayson-g[m] | grayson-g[m]: It was debugging this issue yesterday that brought the "virutal" misspelling to my attention. | 17:28 |
| grayson-g[m] | "What do you mean there's no errors with 'virtual' in the name, Control + F? I'm looking at one right now!" | 17:28 |
| Saviq | It would've been worse if (as I suspected) we had XPASS's encoded with the typo somewhere ;) | 17:29 |
| Saviq | *XFAIL | 17:29 |
| grayson-g[m] | <grayson-g[m]> "I'll take a look at #267 today..." <- Update: Sending discrete scroll events is deprecated since wl_pointer v8 and the event is not sent to clients running v8 or higher. | 17:56 |
| grayson-g[m] | I'm assuming the proper approach is to skip this test if the compositor supports wl_pointer v8 and run a similar test but with value120, and vice-versa should it not support v8. My question is, how can I know what version of wl_pointer the compositor supports? | 17:56 |
| grayson-g[m] | I know how to see if I can bind to a Wayland interface, but I'm confused what the process is for something that's not an extension. | 17:56 |
| grayson-g[m] | > <@grayson-g:matrix.org> I'll take a look at #267 today. I think I know what it takes, I just didn't get finished with it by yesterday. | 17:57 |
| grayson-g[m] | * Update: Sending discrete scroll events is deprecated since wl\_pointer v8 and the event is not sent to clients running v8 or higher. | 17:57 |
| grayson-g[m] | I'm assuming the proper approach is to skip this test if the compositor supports wl\_pointer v8 and run a similar test but with value120, and vice-versa should it not support v8. | 17:57 |
| grayson-g[m] | EDIT: One question resolved. | 17:57 |
| alan_g[m] | The PROPER approach is more complicated that. It is to attach to the compositor using each version of the interface up to and including the highest it supports and to run the right tests for each of these versions | 17:58 |
| alan_g[m] | But we don't have any test infrastructure to support that yet | 17:59 |
| grayson-g[m] | > <@grayson-g:matrix.org> Update: Sending discrete scroll events is deprecated since wl\_pointer v8 and the event is not sent to clients running v8 or higher.... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/fd89e9a5d4e047bc931b7b7f6ef97c093c479672>) | 19:19 |
| RAOF | > <@grayson-g:matrix.org> I figure it's EOD across the board so answer this tomorrow if so, but I'm struggling to get which version of wl_output a compositor supports. I'm trying to see if I get a pointer back from `receive_client.bind_if_supported(wl_pointer_interface, wlcs::AtLeastVersion(8))` but it seems like when I... (full message at <https://libera.ems.host/_matrix/media/v3/download/libera.chat/d4b279890517037716bdcd8baa46dd2c4c9ae3bc>) | 22:17 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!