| *** jude has joined #buildstream | 06:02 | |
| *** jude has quit IRC | 06:15 | |
| *** tristan has quit IRC | 06:41 | |
| *** tristan has joined #buildstream | 06:53 | |
| *** tlater has joined #buildstream | 08:09 | |
| *** jude has joined #buildstream | 08:24 | |
| *** ssam2 has joined #buildstream | 08:31 | |
| *** tiagogomes has joined #buildstream | 08:32 | |
| ssam2 | is GitLab broken for anyone else today? | 08:38 |
|---|---|---|
| tlater | Doesn't look like it, git fetch works fine | 08:42 |
| ssam2 | that's good at least | 08:42 |
| tlater | ... Glibc is still compiling the output of mkdir --version | 08:48 |
| ssam2 | bah | 08:49 |
| ssam2 | do many things successfully build before that? | 08:49 |
| tlater | Only two things. I've experimentally commented out glibc once, no luck either. | 08:50 |
| ssam2 | fair enough | 08:50 |
| ssam2 | i think its best to focus on cleaning up and submitting the patch as-is today, in any case | 08:51 |
| ssam2 | we could spend forever debugging why glibc doesn't build, but if it gets that far then the basics of `bst source-bundle` must be there | 08:51 |
| tlater | Alright. I guess if we get lucky it's just a fluke on my system. | 08:52 |
| *** jonathanmaw has joined #buildstream | 08:53 | |
| ssam2 | here's a BuildStream build on GitLab CI with latest OSTree: | 09:05 |
| ssam2 | https://gitlab.com/baserock/definitions/-/jobs/19501661 | 09:05 |
| ssam2 | sadly it still fails with "unable to set xattr" | 09:05 |
| ssam2 | oh wait | 09:05 |
| ssam2 | Unable to set xattr: No space left on device (12) | 09:05 |
| ssam2 | hahahaha | 09:05 |
| ssam2 | jjardon[m], is there a way to increase the amount of space that these GitLab runners have available? | 09:06 |
| ssam2 | or that error misleading and there's actually something else wrong? this is in base-sdk.bst so one would think that it's quite early on to be running out of space | 09:07 |
| *** jude has quit IRC | 09:12 | |
| gitlab-br-bot | push on buildstream@master (by Tristan Van Berkom): 1 commit (last: ostree.py source plugin: Fixed for recent changes in Source API contract) https://gitlab.com/BuildStream/buildstream/commit/8d5e2304bf182539f768272d229741f9ae5598d6 | 09:24 |
| tristan | ssam2, the base sdk itself is rather big | 09:30 |
| tristan | maybe by itself it accounts for 20% of what will be required to complete the build ? | 09:30 |
| ssam2 | true.. | 09:30 |
| ssam2 | maybe i should host my alpine sysroot somewhere and try again ... | 09:31 |
| ssam2 | i'm suspicious that it was setting xattrs that failed though | 09:31 |
| tristan | yeah | 09:31 |
| tristan | that can be a permission denied error even if the underlying fs supports xattrs | 09:32 |
| tristan | but no space left on device is pretty telling :) | 09:32 |
| gitlab-br-bot | buildstream: merge request (tristan/source-bundle->master: Add source-bundle command) #35 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/35 | 10:29 |
| tlater | ^ Not 100% sure about how pretty some of what I wrote is. | 10:30 |
| tlater | Anything I can do in the mean time? I could continue compiling different versions of glibc, hoping that one works. | 10:35 |
| ssam2 | stick with the latest (2.25) | 10:35 |
| tlater | Doesn't compile :/ | 10:36 |
| ssam2 | I guess you could try older ones to try and bisect the issue, but there's probably an easier route to investigate | 10:37 |
| tristan | I dont know why you are trying to build the toolchain with that | 10:37 |
| tlater | Hm... I'll try and figure out why it's compiling mkdir, I suppose. | 10:38 |
| tlater | I guess I could also compile a chroot with glibc | 10:38 |
| tristan | tlater, I'll try to make a review quickly before I leave, I will be leaving early cause... you know better not work until midnight on fridays | 10:38 |
| tlater | Heh, yeah, that's alright. If you point out bad things I can fix them myself this time. | 10:39 |
| tristan | tlater, the use case is to compile middleware from an already working toolchain, so that is a better test yes | 10:39 |
| ssam2 | I think we'll need to build a new toolchain as the one built in stage2 is pretty minimal | 10:40 |
| ssam2 | but fine to test with other components that have less wacky build systems first | 10:40 |
| tristan | tlater, Ok I think you have plenty to chew on after my dozen comments on your branch :) | 10:52 |
| tlater | *Half a dozen | 10:53 |
| tlater | ;P | 10:53 |
| tristan | only 6 ? | 10:54 |
| tlater | That's what I can see, anyway. | 10:54 |
| tristan | you are right, I forgot to hit comment button on one of them | 10:55 |
| tristan | only 7 :) | 10:55 |
| tristan | tlater, most important though is the API contract and public API surface | 10:55 |
| tlater | Alright, these seem fine. Surprised you didn't complain about writing and deleting a file. | 10:56 |
| tristan | it was a very quick review | 10:57 |
| tristan | probably I missed that | 10:57 |
| tlater | I had a line number report early on, but I'm pretty sure that breaks sh compatibility. Considering this will be used in minimal systems, I don't think reporting what failed can work. | 10:58 |
| tristan | When you run one of the individual build scripts from your master script, always tee the output into a log and keep track of which script / module is running | 10:59 |
| tristan | then you just say "Err building module ${foo}, see: ${logfile}" | 10:59 |
| tristan | *Error | 10:59 |
| tlater | Ok, cool, I should have thought of that. | 10:59 |
| tlater | If I want to create a template script, how do I manage to get a loop? I'm weary of sh loops, because the directory names don't have to be sane. | 11:33 |
| tristan | I dont understand what you mean | 11:34 |
| tlater | If, instead of having a ScriptTemplate class, I want to create a template script in data | 11:35 |
| tristan | but at generation time, you can get the directory to load a template and subst it with python, from _site.py | 11:35 |
| tristan | I have to run | 11:35 |
| tlater | Yeah, but I need to repeat the "execute module script" bit a certain number of times. | 11:36 |
| tristan | I dont understand how this is a problem | 11:36 |
| tlater | I can either fill in a bunch of module names, that might confuse bash, or repeat this in a python loop. | 11:36 |
| ssam2 | quote module names ? | 11:37 |
| tristan | tlater, an array in bash is a good option I think | 11:37 |
| ssam2 | bash won't be available... busybox sh | 11:37 |
| tlater | sh doesn't have support for that though | 11:37 |
| tlater | I'm weary of quoting, had issues with that before, but I can try. | 11:37 |
| tristan | wordlist, then; bst element names already have some constraints | 11:37 |
| ssam2 | python has a tool than can do shell quoting for you | 11:37 |
| tlater | oh | 11:38 |
| tristan | Probably makes sense at least to say that a bst element name is not allowed to have spaces | 11:38 |
| * tristan & | 11:38 | |
| ssam2 | shlex.quote() | 11:38 |
| tlater | Cool, thanks | 11:38 |
| ssam2 | also what tristan said :-) | 11:38 |
| tlater | Ok, that makes things easier | 11:38 |
| *** tristan has quit IRC | 11:41 | |
| gitlab-br-bot | buildstream: merge request (sam/exception-fix->master: _pipeline.py: Fix bad `except:` statement) #36 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/36 | 13:18 |
| gitlab-br-bot | buildstream: merge request (sam/exception-fix->master: _pipeline.py: Fix bad `except:` statement) #36 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/36 | 13:20 |
| *** tlater has quit IRC | 13:55 | |
| *** tlater has joined #buildstream | 13:57 | |
| tlater | I don't think the functionality to log and display an error message can be implemented in sh. | 14:26 |
| tlater | This is because to do so, I need to call tee as a background process, and that makes it impossible to trap EXIT. | 14:26 |
| ssam2 | hmm | 14:27 |
| ssam2 | i'm not sure why tee has to run in the background | 14:27 |
| ssam2 | can't you just run `./buildscript | tee file` | 14:27 |
| tlater | Because I need to catch the exit code | 14:27 |
| tlater | If I do that set -e has no effect | 14:27 |
| ssam2 | ah yeah, the whole pipefile nightmare | 14:28 |
| tlater | On that command, anyway | 14:28 |
| ssam2 | *pipefail | 14:28 |
| ssam2 | i'm not sure. I prefer keeping the script simple | 14:28 |
| tlater | It's either or here, I can implement one of the two quite easily | 14:29 |
| tlater | Technically the user can use tee to log | 14:30 |
| tlater | So I think it makes more sense to just keep the error message | 14:30 |
| ssam2 | sounds good to me | 14:33 |
| tlater | I assume this time I leave my fix commits as they are? | 14:49 |
| tlater | - no fun rebasing? | 14:49 |
| *** ssam2 has quit IRC | 14:51 | |
| *** ssam2 has joined #buildstream | 14:53 | |
| gitlab-br-bot | buildstream: merge request (tristan/source-bundle->master: Add source-bundle command) #35 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/35 | 14:59 |
| gitlab-br-bot | buildstream: merge request (tristan/source-bundle->master: Add source-bundle command) #35 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/35 | 15:01 |
| *** ssam2_ has joined #buildstream | 15:14 | |
| *** ssam2 has quit IRC | 15:14 | |
| jonathanmaw | \o/ I've got .deb files coming out of a pile of artifacts | 15:22 |
| tlater | \o/ | 15:22 |
| jonathanmaw | now to compare them with what would normally be generated, to see how messed-up they are | 15:23 |
| jonathanmaw | all I've done so far is to reconstitute the /DEBIAN/* files and put the other contents in the right place to run dpkg-deb | 15:23 |
| jonathanmaw | so I suspect there'll be some glaring oversight. | 15:24 |
| jonathanmaw | It looks like I forgot the md5sums file, and the files have a different creation date | 15:31 |
| jonathanmaw | otherwise seems fine | 15:31 |
| *** ssam2_ has quit IRC | 16:24 | |
| *** ssam2_ has joined #buildstream | 16:39 | |
| tlater | I'm off, see you next week o/ | 17:02 |
| ssam2_ | have a good weekend | 17:03 |
| *** ssam2_ has quit IRC | 17:03 | |
| *** tlater has quit IRC | 17:03 | |
| gitlab-br-bot | buildstream: merge request (jonathan/dpkg-build->master: WIP: Jonathan/dpkg build) #37 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/37 | 17:40 |
| *** tiagogomes has quit IRC | 17:45 | |
| gitlab-br-bot | buildstream: merge request (jonathan/dpkg-build->master: WIP: Jonathan/dpkg build) #37 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/37 | 17:45 |
| *** jonathanmaw has quit IRC | 17:49 | |
| *** tristan has joined #buildstream | 20:57 | |
Generated by irclog2html.py 2.14.0 by Marius Gedminas - find it at mg.pov.lt!