| mmikowski | arraybolt3[m]: We will *definitely* want to test for Kfocus devices. All M2 and XE generations have Thunderbolt 3 or 4. We can test for eGPU and display devices. Although, as Erich has said - and I agree - updating TB at this time of the lifecycle of the kernel seems awfully risky. We hadn't seen or heard any complaints up to this time, but we have seen kernel updates break USB and bluetooth before. We'd rather not go through | 00:25 |
|---|---|---|
| mmikowski | that again (although we now meta-package kernels to avoid breaking kernels). | 00:25 |
| arraybolt3[m] | mmikowski: Nice. Makes sense. | 00:28 |
| mmikowski | arraybolt3[m]: Thanks. We meta-package kernels and nv libs to avoid breakage on our hardware but still allow us to use upstream packages. The only time a kernel gets held back is if it breaks the hardware smoke test. NV libs are similar and are correlated with matching CUDA libs for ML. | 00:32 |
| arraybolt3[m] | So basically random hardware flunk-outs with a kernel update won't happen on a Focus? | 00:32 |
| arraybolt3[m] | That sounds very good. Hmm, that makes my choice of upgrade harder - Framework is cheaper and more powerful, but Focus is more dependable. Hmm... hard decisions! | 00:33 |
| mmikowski | arraybolt3[m]: right. We test the kernels and drivers for the hardware. Never say never, but we test against it for all systems we've ever sold. | 00:50 |
| Eickmeyer[m] | RikMills: *ahem* bug 1987934 | 03:54 |
| ubottu | Bug 1987934 in kdenlive (Ubuntu Kinetic) "kdenlive 22.08: missing Recommends: mediainfo" [Medium, Triaged] https://launchpad.net/bugs/1987934 | 03:54 |
| Eickmeyer[m] | Also working on another bug with mlt missing a module altogether. Basically breaking some kdenlive functionality. | 03:55 |
| RikMills | Eickmeyer[m]: you going to upload with that recommends? | 05:20 |
| Eickmeyer[m] | That's my intention, first thing in the morning. It's 10:21pm here. | 05:22 |
| RikMills | okay | 05:22 |
| BluesKaj | Hi all | 12:17 |
| RikMills | Eickmeyer[m]: cheers for the fix. any luck with mlt? | 17:59 |
| RikMills | I have a similar issue with haruna https://bugs.debian.org/cgi-bin/bugreport.cgi?bug=1017593 | 18:00 |
| ubottu | Debian bug 1017593 in haruna "Haruna: haruna watch file should point to the official release tarball location" [Normal, Open] | 18:00 |
| Eickmeyer[m] | RikMills: Well, I'm having trouble with the debian maintainer understanding that he grabbed the wrong tarball. The tarball is missing a bunch of git submodules. | 18:00 |
| Eickmeyer[m] | The upstream developers of mlt didn't even officially release mlt 7.8.0 until July 3rd, despite tagging on June 22nd. Basically, he jumped the gun. | 18:01 |
| Eickmeyer[m] | Oh, big yikes. | 18:02 |
| RikMills | with many KDE things, if you take the tarball from a git tag, then you don't have translations. As they are kept in svn and rolled into release tars via release scripts | 18:03 |
| Eickmeyer[m] | Yeah. There's a lot of audio plugins that are like that with git submodules. | 18:04 |
| Eickmeyer[m] | I even know of one developer that's appalled that he even has to be bothered to release a tarball. 🙄 | 18:05 |
| RikMills | properly release tars can also be signed with checking via uscan. grabbing the wrong ones rules that out | 18:06 |
| Eickmeyer[m] | Yeah. | 18:07 |
| Eickmeyer[m] | So, the struggle right now is proper functionality in kdenlive 22.08 with that mlt module or things appearing broken to the average user. I'd rather have the former. | 18:08 |
| RikMills | agreed | 18:08 |
| RikMills | we can sort it for kinetic with or without cooperation. just going forward it needs to be sorted | 18:09 |
| Eickmeyer[m] | The maintainer in this case mentioned he didn't want to introduce an epoch, I wrote back that is understandable, but changing tarballs doesn't need an epoch, just a +repack or something like that in the suffix of the tarball. | 18:11 |
| RikMills | yes, there are plenty of example in the archive of that being done | 18:12 |
| Eickmeyer[m] | I'd even go for something like +actual | 18:12 |
| RikMills | could be | 18:12 |
| Eickmeyer[m] | But either way, the closer it gets to beta freeze, the less likely I am to let it slide. | 18:13 |
| valorie | stick to your guns, Eickmeyer[m]! Your email was extremely reasonable | 20:19 |
| Eickmeyer[m] | Thanks, valorie . So far, Gianfranco has good advice, but the maintainer seemed to... overreact? I mean, no need to bump an epoch, just give a suffix and call it good! | 20:21 |
| valorie | maintainer could be embarrassed | 20:21 |
| Eickmeyer[m] | True. | 20:21 |
| valorie | gosh, we all make mistakes | 20:22 |
| valorie | chill & fix! | 20:22 |
| Eickmeyer[m] | Well, yeah, but own your mistake, learn from it, and fix it. No need to assume the worst. | 20:22 |
| Eickmeyer[m] | So far, nothing is broken, nobody's cat has died. It just pops up a scary warning. | 20:23 |
| valorie | yep | 20:24 |
| === Feesh is now known as ely-the-kitsune | ||
| Eickmeyer[m] | RikMills: Looks like haruna got accepted into unstable! \o/ | 23:06 |
Generated by irclog2html.py 2.7 by Marius Gedminas - find it at mg.pov.lt!