| *** w0rd has joined #buildstream | 00:26 | |
| *** bochecha has quit IRC | 00:41 | |
| *** toscalix has joined #buildstream | 06:40 | |
| *** mohan43u has quit IRC | 08:07 | |
| *** mohan43u has joined #buildstream | 08:11 | |
| *** mohan43u has quit IRC | 08:15 | |
| *** mohan43u has joined #buildstream | 08:18 | |
| *** finn has joined #buildstream | 08:56 | |
| gitlab-br-bot | jjardon opened issue #763 (Overnigth tests are failing: "Error response from daemon: Conflict. The container name "/runner-xxx" is already in use by container") on buildstream https://gitlab.com/BuildStream/buildstream/issues/763 | 09:00 |
|---|---|---|
| laurence | Is the meeting going ahead? | 09:01 |
| laurence | ah yes, had to re-join. | 09:02 |
| toscalix | there is something wrong.... I had issues joining and I cannot see or hear anything | 09:09 |
| toscalix | let's move to Google Meet. Let me add the link | 09:09 |
| toscalix | or is it me alone? | 09:09 |
| *** WSalmon has joined #buildstream | 09:10 | |
| toscalix | valentind: laurence jjardon ^ | 09:11 |
| *** benschubert has joined #buildstream | 09:11 | |
| jjardon | toscalix: Its only you | 09:12 |
| laurence | toscalix, we are all on appear.in | 09:12 |
| laurence | going through the issues | 09:12 |
| toscalix | ah, ok | 09:12 |
| toscalix | will keep working on it then | 09:12 |
| *** WS has joined #buildstream | 09:28 | |
| *** WSalmon has quit IRC | 09:28 | |
| *** WS is now known as WSalmon | 09:29 | |
| *** alatiera_ has joined #buildstream | 09:39 | |
| *** jonathanmaw has joined #buildstream | 09:41 | |
| *** tiagogomes has joined #buildstream | 09:49 | |
| *** tristan has joined #buildstream | 09:49 | |
| *** tiagogomes has quit IRC | 09:55 | |
| *** tiagogomes has joined #buildstream | 09:56 | |
| *** tpollard has joined #buildstream | 09:59 | |
| *** alatiera__ has joined #buildstream | 10:08 | |
| *** alatiera_ has quit IRC | 10:08 | |
| *** alatiera__ is now known as alatiera_ | 10:14 | |
| *** raoul has quit IRC | 10:19 | |
| *** lachlan has joined #buildstream | 10:31 | |
| *** jfhbrook has joined #buildstream | 10:37 | |
| *** rdale has joined #buildstream | 10:40 | |
| *** jonathanmaw_ has joined #buildstream | 10:42 | |
| *** jonathanmaw has quit IRC | 10:43 | |
| gitlab-br-bot | richardmaw-codethink approved MR !945 (bschubert/fix-silence-stopiteration-pep-0479->master: source.py: don't let StopIteration propagate to silence() contextmanager) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/945 | 11:05 |
| gitlab-br-bot | richardmaw-codethink merged MR !945 (bschubert/fix-silence-stopiteration-pep-0479->master: source.py: don't let StopIteration propagate to silence() contextmanager) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/945 | 11:05 |
| Nexus | hey all, i'd really appreciate some reviews on: | 11:07 |
| Nexus | https://gitlab.com/BuildStream/buildstream/merge_requests/925 | 11:07 |
| Nexus | https://gitlab.com/BuildStream/buildstream/merge_requests/926 | 11:07 |
| Nexus | thanks! | 11:07 |
| Nexus | 926 is an interesting one, as it's adding a --world flag to bst build, which some people dislike the name of, so i think a discussion would be useful | 11:08 |
| *** lachlan has quit IRC | 11:11 | |
| jennis | https://wiki.codethink.co.uk/bl013/log/bl013:_Standup:_Mon_12th_November_2018/ | 11:13 |
| jennis | woops | 11:15 |
| *** lachlan has joined #buildstream | 11:15 | |
| Nexus | :I | 11:16 |
| Nexus | smh | 11:16 |
| *** jonathanmaw_ is now known as jonathanmaw | 11:17 | |
| jonathanmaw | how odd, when I run a test with "--no-cov" it passes, when I run it without, it fails because it thinks the element hasn't been built :S | 11:19 |
| *** lachlan has quit IRC | 11:20 | |
| *** raoul has joined #buildstream | 11:23 | |
| *** lachlan has joined #buildstream | 11:28 | |
| benschubert | @Nexus here you go :) | 11:29 |
| benschubert | I'd also appreciate reviews on !895 and !938 if people have time | 11:35 |
| gitlab-br-bot | MR !895: Don't cache sandbox failures https://gitlab.com/BuildStream/buildstream/merge_requests/895 | 11:35 |
| gitlab-br-bot | MR !938: Fix os.rename in git source element to correctly handle error codes https://gitlab.com/BuildStream/buildstream/merge_requests/938 | 11:35 |
| Nexus | benschubert: you raise a good point regarding the --world/--all flags, and i agree that we should just have one, but which? or a 3rd option? As --all already exists in some contexts, but not in the same way as this | 11:50 |
| Nexus | and i've heard from some people that they don't like the --world convention | 11:50 |
| benschubert | `--world` is at least the convention in the Gentoo world, it's easy to remember and doesn't conflict with any other --all we could have (assuming the --all all have the same meaning). I don't see any other solution that is "commonly" used in projects. And I agree that keeping the meaning the same for options accross commands should be done | 11:52 |
| Kinnison | I think --world also comes from Yocto | 11:53 |
| Kinnison | so it's common in meta buildsystems | 11:53 |
| tpollard | bitbake does indeed use world | 11:53 |
| Kinnison | If there's objection to 'world' then I'd go for '--all-elements' | 11:53 |
| Kinnison | or '--everything' | 11:53 |
| * Kinnison likes nice wordy options | 11:53 | |
| benschubert | I would suggest against --all-elements, especially with autocompletion :) | 11:54 |
| benschubert | and --world has the advantage that "-w" is also unused, so we would have a convenient short option | 11:55 |
| tiagogomes | or bst build \* | 11:56 |
| skullman | * is a valid filename | 11:56 |
| skullman | an alternative might be a --glob option to disambiguate it | 11:57 |
| tiagogomes | I think the case of having a '*' named file with build definitions is rare enough case to not be worth to consider | 11:58 |
| tiagogomes | Also assuming good citizens name files with the .bst extension | 11:59 |
| benschubert | I'd be against *, typing a \ everytime would be a pain | 11:59 |
| Kinnison | I still think we should have mandated the .bst extension and removed it from element names | 11:59 |
| * Kinnison runs away | 12:00 | |
| Nexus | i agree that we shouldrequire .bst | 12:00 |
| skullman | should ≠ must | 12:00 |
| tiagogomes | I could turn the argument to bst build a valid glob as in (https://docs.python.org/3/library/glob.html) | 12:01 |
| tiagogomes | This I think would remain backwards compatible and flexible | 12:02 |
| tiagogomes | So to build everything: bst build "**" | 12:02 |
| * tiagogomes goes for lunch | 12:02 | |
| benschubert | tiagogomes: that's something that seems safer :D | 12:02 |
| gitlab-br-bot | valentindavid opened MR !948 (valentindavid/integration-tests-aarch64->master: Run tests on aarch64) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/948 | 12:08 |
| *** lachlan has quit IRC | 12:10 | |
| valentind | tristan, Maybe you could have a look at #144 to be sure we can close it. | 12:12 |
| gitlab-br-bot | Issue #144: Possible race condition - inconsistent passing test https://gitlab.com/BuildStream/buildstream/issues/144 | 12:12 |
| *** lachlan has joined #buildstream | 12:13 | |
| laurence | jennis, I have an email from Gitlab telling me you commented on this issue about cache clean-up... https://gitlab.com/BuildStream/buildstream/issues/737#note_115262786 | 12:32 |
| laurence | but i cannot see the comment on gitlab's UI - is this a gitlab bug or did you remve it? | 12:33 |
| *** lachlan has quit IRC | 12:37 | |
| *** jonathanmaw_ has joined #buildstream | 12:40 | |
| *** jonathanmaw has quit IRC | 12:41 | |
| *** alatiera_ has quit IRC | 12:46 | |
| *** alatiera_ has joined #buildstream | 12:46 | |
| tpollard | tristan: I've tried to cover your points and closed discussions that I felt were mine to close on https://gitlab.com/BuildStream/buildstream/merge_requests/786/ | 12:48 |
| *** lachlan has joined #buildstream | 12:53 | |
| *** lachlan has quit IRC | 12:58 | |
| Nexus | Any further thoughts on "bst build "**"" ? | 13:17 |
| * skullman would prefer `bst build --glob "**"` or `bst build -g "**"` | 13:18 | |
| Nexus | that starts to become far less intuitive imo | 13:19 |
| skullman | I'm just a grump about making assumptions about file path validity. | 13:21 |
| Nexus | would we be introducing "--glob" as a different function then? different to the current "Just build everything" functionality ? | 13:23 |
| skullman | yeah, you'd be doing the same tree walk to find things (though depending on how it's implemented you don't need to walk the whole tree) and filtering using the glob expression | 13:24 |
| Kinnison | One concern I have with globbing is that if someone has READMEs or notes files, or somesuch in the tree, that we're going to hit issues. | 13:27 |
| skullman | so `-g "**.bst"` rather than `--world` and assume everything under elements/ is a bst definition? | 13:29 |
| Kinnison | That'd be okay for me | 13:29 |
| Kinnison | though a good shell would let you type **/*.bst and hit TAB and have it DTRT anyway | 13:30 |
| Nexus | DTRT? | 13:30 |
| Nexus | and this still requires .bst to be mandatory doesn't it? | 13:31 |
| benschubert | I don't think you can assume everything under elements will be a bst definitions, one thing we have is ".inc" files that contain general definitions for groups of elements | 13:31 |
| *** starquake has joined #buildstream | 13:33 | |
| Nexus | I don't see how it's possible to have a global "build all" without either asserting that only elements exist in the element-path or that all elements end in .bst | 13:34 |
| benschubert | One thing that could be done though is use a "element-extensions:" config that would allow specifying the elements? | 13:34 |
| benschubert | *the elements extensions | 13:34 |
| benschubert | with a default to "[.bst]" for example (or to *, if we don't want to break anyone) | 13:35 |
| Nexus | it might be worth making an email thread for this :/ | 13:40 |
| gitlab-br-bot | BenjaminSchubert opened MR !949 (bschubert/add-fedora-nobwrap-tests->master: Add tests for fedora with no bubblewrap available) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/949 | 13:51 |
| *** lachlan has joined #buildstream | 13:52 | |
| benschubert | Kinnison: this PR is close to what you requested in the issue I closed without paying attention, would you mind having a look into this? | 13:52 |
| Kinnison | benschubert: which? | 13:54 |
| Kinnison | (MR number) | 13:54 |
| benschubert | !949 sorry | 13:54 |
| benschubert | and for issue #696 | 13:55 |
| gitlab-br-bot | Issue #696: Missing tests for Linux-non-bwrap-non-FUSE situations https://gitlab.com/BuildStream/buildstream/issues/696 | 13:55 |
| Kinnison | is there a reason it's a large number of commits? | 13:55 |
| benschubert | I refactored the gitlab-ci file to remove duplication and did it step by step for better clarity | 13:56 |
| benschubert | all commits are passing one after the other though :) | 13:56 |
| benschubert | I can squash some together if needed | 13:56 |
| Kinnison | If they all pass then that's fine I guess | 14:13 |
| Kinnison | :_) | 14:13 |
| benschubert | thanks! | 14:17 |
| *** jonathanmaw has joined #buildstream | 14:32 | |
| *** jonathanmaw_ has quit IRC | 14:33 | |
| *** lachlan has quit IRC | 14:33 | |
| *** Kinnison has quit IRC | 14:42 | |
| *** Kinnison has joined #buildstream | 14:42 | |
| *** lachlan has joined #buildstream | 14:50 | |
| *** lachlan has quit IRC | 15:25 | |
| gitlab-br-bot | juergbi merged MR !937 (element-path_not_validated->master: Element path not validated before use) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/937 | 15:44 |
| jonathanmaw | :-/ looks like the reason why my tests were failing when I passed "--no-cov" is because the coverage file is generated in the current directory, which changes the cache key | 15:57 |
| Kinnison | oops | 16:00 |
| * jonathanmaw tries replacing the "cwd" kwarg in cli.run with adding "-C" to the args. | 16:01 | |
| jonathanmaw | well, that works! | 16:05 |
| benschubert | I'm not sure to grasp exactly how the mounting of directories works in the sandbox. Do we have documentation somewhere? I'm particularily interested in knowing what directories are mounted in what order and with what options (trying to replace the fuse layer just for playing) | 16:20 |
| *** lachlan has joined #buildstream | 16:25 | |
| skullman | not really, and it varies by platform too i.e. sandboxchroot mounts the fuse layer on top of /dev, while sandboxbwrap mounts /dev on top of the fuse layer | 16:25 |
| benschubert | if I understand well, the FUSE layer is just here to provide a COW experience so we don't corrupt the real data on disk correct? But in the end only /buildstream and /buildstream-install are saved. Correct? | 16:26 |
| skullman | correct, though I was told at one point that buildbox continuously uploads all the filesystem changes to CAS, so everything else gets there too at some point | 16:27 |
| benschubert | So I would need to save the changes made on the rest of the layer everytime, I wouldn't be able of discarding it right? Do we have somewhere a spec about the format of the data we upload? | 16:28 |
| skullman | BuildStream doesn't care about the contents of anywhere but /buildstream and /buildstream-install, I think BuildBox might care since it wants to be more generic or as a side-effect of how its CAS fuse layer works, I'm not sure of the details. | 16:31 |
| skullman | The spec of the upload format is the gRPC protos. | 16:31 |
| juergbi | with local execution, changes in the rootfs will not be uploaded with BuildBox, although they will be in local cache | 16:32 |
| benschubert | juergbi: so they are cached locally? The context: I'm trying to add a docker sandbox, to be able to run on Windows. But I don't have access to fuse so I was thinking doing that with docker layers, but I've got trouble making sure I'm actually doing the correct thing | 16:33 |
| juergbi | with remote execution all changes are needed in the remote CAS to preserve them across commands | 16:33 |
| *** lachlan has quit IRC | 16:34 | |
| juergbi | benschubert: yes, they will be cached locally but in CAS, not as regular filesystem structure | 16:34 |
| juergbi | that's for the Build | 16:35 |
| juergbi | Box case | 16:35 |
| juergbi | with the current local execution, a separate rootfs temp directory will be created where COW changes are cached | 16:36 |
| juergbi | for Docker with overlayfs you might need two separate directories, one for the lower read-only layer and one for the changes | 16:36 |
| benschubert | Docker by default is COW, so saving the layer after the command run should be sufficient, I'm just not sure if the format would be correct. And for the /buildstream{,-install}, are they also COW ? | 16:37 |
| skullman | benschubert: closer to volumes AIUI | 16:39 |
| juergbi | benschubert: in practice the format of those COW changes should only matter for the sandbox backend itself | 16:39 |
| juergbi | /buildstream{,-install} are not COW, they are regular native directories, empty at startup | 16:39 |
| juergbi | so they should likely be volumes as skullman mentions | 16:40 |
| benschubert | or bind-mounts so we have direct access to them I guess, ok I think I have enough information to move, thanks a lot! | 16:41 |
| skullman | juergbi: sorry to bother you, but have you had a chance to look at the addition of the BST_GRANULAR_STAGE flag to elements in https://gitlab.com/BuildStream/buildstream/merge_requests/909 and I could ideally do with you looking at https://gitlab.com/BuildStream/buildstream/merge_requests/920 since you did the work to split up the artifact and cas cache classes and it's a new user. | 16:41 |
| juergbi | I think I had a brief look. let's check | 16:43 |
| * skullman didn't see any comments, but GitLab is like that sometimes | 16:50 | |
| *** lachlan has joined #buildstream | 16:54 | |
| benschubert | juergbi: if you have time, i'd like a maintainer review on !895, since it has to do with caching | 17:04 |
| gitlab-br-bot | MR !895: Don't cache sandbox failures https://gitlab.com/BuildStream/buildstream/merge_requests/895 | 17:04 |
| juergbi | I've added it to my list | 17:05 |
| benschubert | Thanks! | 17:05 |
| *** lachlan has quit IRC | 17:07 | |
| gitlab-br-bot | valentindavid opened (was WIP) MR !948 (valentindavid/integration-tests-aarch64->master: Run tests on aarch64) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/948 | 17:10 |
| *** toscalix has quit IRC | 17:13 | |
| *** lachlan has joined #buildstream | 17:15 | |
| juergbi | skullman: added an initial comment | 17:18 |
| juergbi | also added !920 to my list | 17:18 |
| gitlab-br-bot | MR !920: Add artifact log command https://gitlab.com/BuildStream/buildstream/merge_requests/920 | 17:18 |
| skullman | thanks | 17:18 |
| jonathanmaw | If anyone has time for review, I've fixed the CI problem I was having with https://gitlab.com/BuildStream/buildstream/merge_requests/924 | 17:24 |
| *** lachlan has quit IRC | 17:25 | |
| *** abderrahim3 has joined #buildstream | 17:30 | |
| *** abderrahim2 has quit IRC | 17:31 | |
| *** lachlan has joined #buildstream | 17:37 | |
| *** lachlan has quit IRC | 18:04 | |
| *** lachlan has joined #buildstream | 18:13 | |
| *** toscalix has joined #buildstream | 18:23 | |
| *** toscalix has quit IRC | 18:26 | |
| *** rdale has quit IRC | 18:40 | |
| *** alatiera_ has quit IRC | 18:42 | |
| *** raoul has quit IRC | 18:45 | |
| *** rdale has joined #buildstream | 18:50 | |
| *** lachlan has quit IRC | 19:31 | |
| *** lachlan has joined #buildstream | 19:50 | |
| *** lachlan has quit IRC | 19:58 | |
| *** lachlan has joined #buildstream | 20:09 | |
| *** tristan has quit IRC | 20:39 | |
| *** lachlan has quit IRC | 20:45 | |
| *** catonano has joined #buildstream | 20:46 | |
| *** jonathanmaw has quit IRC | 20:56 | |
| *** catonano has quit IRC | 23:03 | |
| *** samouy has joined #buildstream | 23:45 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!