| *** zbenjamin is now known as Guest39177 | 01:18 | |
| *** zbenjamin_ is now known as zbenjamin | 01:18 | |
| *** frinring_ is now known as frinring | 03:56 | |
| *** leinir_ is now known as leinir | 16:42 | |
| attah | I'll have to by GitHub minutes before i get a grip on Docker... | 18:49 |
|---|---|---|
| Nico[m] | That's why I usually use custom runners (although with gitlab) | 18:49 |
| attah | as in self-hosted? | 18:51 |
| Nico[m] | Yes, but you can use custom runners also with hosted | 18:51 |
| attah | wow, Docker is really finicky :( | 20:03 |
| Nico[m] | It sometimes really is | 20:04 |
| attah | the job on github fails with mb2(?) saying Fatal: 'SailfishOS-3.3.0.14-armv7hl' is not a known build target | 20:04 |
| Nico[m] | Ah, yeah | 20:05 |
| attah | same thing locally fails to find a file that definitely exists inside the container | 20:05 |
| Nico[m] | I think for that reason I added `sdk-assistant list` to my gitlab job :D | 20:06 |
| attah | wat? | 20:06 |
| attah | oh, now i get it | 20:06 |
| Nico[m] | I had the same issue. I don't remember how I fixed it, I think I needed to pushd before my build | 20:07 |
| Nico[m] | If you want to look at my struggles: https://nheko.im/nheko-reborn/spoon/-/merge_requests/1/pipelines | 20:08 |
| attah | thanks | 20:11 |
| attah | i'll save that for tomorrow | 20:11 |
| Nico[m] | Looking at my own job failure history, it seems like the pushd ~/build fixed it | 20:12 |
| Nico[m] | So that may be worth a try | 20:12 |
| attah | as opposed to a cd? | 20:12 |
| Nico[m] | Well, cd should probably work too | 20:13 |
| Nico[m] | I assume you are also nemo and not root? | 20:13 |
| attah | likely | 20:13 |
| Nico[m] | In the end, I'm not sure, how I fixed it anymore, it's been 5 months :D | 20:14 |
| attah | i mkdired a build dir on /tmp... since i didn't have permissions in the mounted(?) volume | 20:14 |
| Nico[m] | You should have permissions in /home/nemo | 20:14 |
| attah | maybe there is better... but shouldn't matter | 20:15 |
| Nico[m] | The other locations should be read only | 20:15 |
| Nico[m] | You say that now, but... :D | 20:16 |
| attah | So the output of sdk-assistant list is a hierarchy of armv7hl and i486 under the version... on github, it is just the version heading | 20:17 |
| Nico[m] | Yeah, then it can't find the sdk, which also explains your error | 20:18 |
| Nico[m] | You can install them manually, but that takes ages and I don't think you need to do that | 20:18 |
| attah | yes, but since it is the exact same Dockerfile, i don't get it | 20:18 |
| Nico[m] | I think it was really something about the directories | 20:19 |
| attah | damnit, you were right | 20:19 |
| Nico[m] | Can I see your build definition? | 20:19 |
| attah | now it build locally | 20:20 |
| Nico[m] | Heh | 20:20 |
| Nico[m] | It is such a weird issue and the error message does not help at all! | 20:20 |
| Nico[m] | So you're sorted? | 20:20 |
| attah | probably not... | 20:20 |
| attah | still no targets on Github, but will retry | 20:21 |
| attah | still same crap | 20:26 |
| Nico[m] | Hm | 20:26 |
| Nico[m] | Maybe add some whoami and pwds? | 20:27 |
| attah | probably... but it works locally :/ | 20:27 |
| Nico[m] | Are you running the same thing locally= | 20:28 |
| attah | "docker build ." in the workflow repo i'm making | 20:28 |
| Nico[m] | Ah, you are building inside the dockerfile? | 20:29 |
| attah | and then a simple docker "run -v /home/attah/repos/harbour-seaprint:/github/workspace <hash from previous step>" | 20:29 |
| attah | the Dockerfile ENTRYPOINT owns the command to execute | 20:30 |
| Nico[m] | I see. That is somewhat unusual to me coming from gitlab :D | 20:30 |
| attah | well, i have no idea what i *should* be doing, but the abstraction of just deferring to an externally defined build step seemed good at the time | 20:31 |
| Nico[m] | Yeah, I may need to read up on Github Actions. They always looked weird and complex to me | 20:32 |
| Nico[m] | gitlab just has the build steps one after another, which is easier to wrap my mind around | 20:32 |
| attah | well, this is that too, but one step has "uses: attah/buildfish@0.6" instead of a run attribute | 20:33 |
| Nico[m] | Mhm | 20:33 |
| Nico[m] | Did you look at the actions coderus uses? | 20:35 |
| attah | that's what i started with... just tried to break it out | 20:35 |
| Nico[m] | And coderus version worked? | 20:35 |
| attah | no (: | 20:37 |
| Nico[m] | :D | 20:37 |
| Nico[m] | CI sucks, it takes so much effort to find the small issue, that breaks everything | 20:38 |
| Nico[m] | I still would guess the current user is wrong, when you are executing your build step | 20:38 |
| attah | This was a pretty nice experience tbh: https://github.com/attah/bytestream/actions | 20:39 |
| Nico[m] | Looks like it. Not 90% of your history is red! | 20:39 |
| attah | took a few tries to get it right... the ones you see are legitimate errors | 20:40 |
| attah | (i test with -Werror -Wextra -pedantic etc. on both g++ and clang) | 20:41 |
| attah | sometimes i don't bother with both compilers locally... and it bites back | 20:41 |
| Nico[m] | I think I need to try out github actions for nheko too now. Travis sucks and appveyor is barely better | 20:41 |
| Nico[m] | I did not know you could just use ubuntu as the image :3 | 20:42 |
| attah | as opposed to? | 20:42 |
| Nico[m] | I thought you had to define weird actions to do anything | 20:43 |
| Nico[m] | There is an action to create an APT repo but not an RPM one ._. | 20:45 |
| attah | coderus trigger works, as expected | 20:49 |
| Nico[m] | Well, then the issue has to be somewhere in between :D | 20:50 |
| attah | ...the keyboard and the chair | 20:50 |
| Nico[m] | Well, that too ;p | 20:50 |
| attah | oh well... better get some sleep | 20:59 |
| attah | good night! | 20:59 |
| Nico[m] | Good night :3 | 20:59 |
Generated by irclog2html.py 2.17.1 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!