| *** leopi has joined #buildstream | 01:45 | |
| *** leopi has quit IRC | 03:50 | |
| *** leopi has joined #buildstream | 04:34 | |
| *** slaf has quit IRC | 05:17 | |
| *** slaf has joined #buildstream | 05:19 | |
| *** slaf has joined #buildstream | 05:19 | |
| *** slaf has joined #buildstream | 05:19 | |
| *** slaf has joined #buildstream | 05:19 | |
| *** tristan has joined #buildstream | 06:27 | |
| *** leopi has quit IRC | 06:29 | |
| *** Kinnison has quit IRC | 06:34 | |
| *** Kinnison has joined #buildstream | 06:44 | |
| *** toscalix has joined #buildstream | 06:45 | |
| *** ernestask has joined #buildstream | 06:47 | |
| *** noisecell has joined #buildstream | 07:24 | |
| tristan | juergbi, I've hit another regression I suspect is related to CAS... | 07:30 |
|---|---|---|
| tristan | juergbi, see: https://bpaste.net/show/b16ce4ba725f | 07:30 |
| tristan | and the build log itself: https://bpaste.net/show/4f018af4778c | 07:30 |
| tristan | Something about public data has got screwed... here we are building base/ed.bst, and calling self.integrate() requires it's public data... somehow we run down the codepath to load it from an artifact, but the artifact cannot yet exist | 07:32 |
| tristan | I hit this by bumping my core artifact version ensuring that I get a full rebuild of freedesktop-sdk, I was rather expecting a different error to pop up for git describe | 07:32 |
| tristan | Not sure how we got as far as we did without hitting the public data load error | 07:33 |
| tristan | strange, there is an empty directory at /home/tristan/.cache/buildstream/artifacts/extract/base-sdk/base-ed/cf9edc49dd3173be815d560e69346e588a37f515e03fc1d28753e060cfaae86d | 07:37 |
| tristan | which would probably cause BuildStream to think that the artifact exists, causing it to try to extract the public data maybe | 07:37 |
| tristan | Removing it however, results in the same error | 07:37 |
| tristan | And the directory gets created again | 07:38 |
| *** qinusty has joined #buildstream | 07:55 | |
| *** coldtom has joined #buildstream | 07:58 | |
| Kinnison | tristan: regarding the WIP/MR approach you outlined -- why do you think this is the way Gitlab is suggesting people should work? The way I read things is "Issue assigned, but not MR - someone is working on a fix/feature" "MR created, but WIP, someone's after some feedback but not a full-on review" "MR marked not-WIP, the feature/fix is ready for proper review" | 08:00 |
| Kinnison | It's really hard to filter for non-WIP MRs as far as I can tell | 08:00 |
| Kinnison | So the fewer of those we have, the easier a time of it there'll be for those who want to review something theoretically ready for merge | 08:01 |
| tristan | Kinnison, just experience | 08:01 |
| tristan | Kinnison, i.e. I've seen even merge requests being opened with *no commit* further than the branch point | 08:01 |
| Kinnison | I don't understand why anyone would do that | 08:01 |
| Kinnison | it's not like MR numbers need name-grabbing | 08:01 |
| tristan | Also, when someone pushes a branch the first time, gitlab giving you a URL to create a merge request seems to prompt people to do it | 08:02 |
| tristan | I do agree with your logic, though | 08:02 |
| Kinnison | To my mind, an MR is saying "I require a human to look at this and give me an opinion", especially given gitlab can run pipelines against branches not in MR (IIRC) | 08:02 |
| * Kinnison does wish the message gitlab gave you on push, instead of saying "Create merge request..." said "If you are ready, you may create a merge request by following..." | 08:04 | |
| tristan | Kinnison, to be fair, the older wording was too strict, I'm not against changing this | 08:04 |
| Kinnison | I suggest that too many changes close together will confuse people | 08:04 |
| tristan | it's just I feel like there's a limit to how far we can expect people to follow guidelines | 08:04 |
| Kinnison | let's see how we do for at least a few weeks with your updates | 08:04 |
| * Kinnison goes to wade through the MR list | 08:05 | |
| tristan | Kinnison, in practice I think it's around the same... if you open a WIP branch, you are probably going to ask a peer to review explicitly | 08:06 |
| tristan | while without WIP is a general "I'm ready for formal review" and a sign that we should review that | 08:06 |
| Kinnison | Yes, but if you open a WIP with nothing in it, that's very confusing for anyone coming in to try and help :-) | 08:07 |
| gitlab-br-bot | buildstream: merge request (Qinusty/275->master: Indicate where artifacts are going to and coming from in the log) #553 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/553 | 08:11 |
| *** WSalmon has joined #buildstream | 08:14 | |
| *** jennis has joined #buildstream | 08:16 | |
| qinusty | Cheers for the review tristan, should be sorted. Just waiting on the pipeline. | 08:17 |
| *** finn has joined #buildstream | 08:17 | |
| *** bochecha has joined #buildstream | 08:19 | |
| gitlab-br-bot | buildstream: merge request (tristan/git-stage-with-shared-clone->master: plugins/sources/git.py: Use --shared instead of --hardlinks) #557 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/557 | 08:38 |
| tristan | qinusty, looks like you forgot to rebase against upstream master :) | 08:38 |
| Kinnison | tristan: I've approved your --shared change, it looks sensible and the pipeline passed | 08:38 |
| tristan | anyway looks like there is a race in the queue :) | 08:38 |
| tlater | Kinnison: Personally, sometimes I like having a WIP MR on gitlab just to keep track of the branch I'm working on. | 08:38 |
| tristan | Kinnison, thanks ! | 08:39 |
| Kinnison | tlater: Is that not what an issue assigned to you does? | 08:39 |
| tlater | It doesn't link to the branch | 08:39 |
| tlater | Which means I can forget my branch name, and need to figure out which of the bunch it was... | 08:39 |
| Kinnison | hmm, and a suitable comment on the issue isn't enough? | 08:39 |
| Kinnison | Or naming your branches tlater/issuenum-what-it-is ? | 08:39 |
| tristan | tlater, the way I do that is observe the branches named "tristan/*", and then cleanup occasionally old/irrelevant stuff | 08:39 |
| tlater | Probably would be, but an MR doesn't leave annoying debris | 08:40 |
| Kinnison | tlater: apart from the debris of an MR in place which doesn't represent a request for a review | 08:40 |
| tlater | Kinnison: That debris disappears when the branch is merged :) | 08:40 |
| * Kinnison shrugs, we're all different and I'm happy for whatever solution tristan and others find works best | 08:40 | |
| tristan | but anyway, I dont mind this overflow of the merge queue with WIP stuff, as it's the WIP branch owner's responsibility to track | 08:40 |
| * Kinnison is unlikely to ever be happy with Gitlab anyway :-D | 08:40 | |
| tristan | it may make cleanup more difficult to do periodicially | 08:41 |
| * Kinnison will be chasing a whole bunch of MRs associated with people from the team he's helping later today | 08:41 | |
| Kinnison | Hopefully we'll see some movement on some of them | 08:41 |
| gitlab-br-bot | buildstream: merge request (tristan/git-stage-with-shared-clone->master: plugins/sources/git.py: Use --shared instead of --hardlinks) #557 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/557 | 08:47 |
| *** coldtom has quit IRC | 08:48 | |
| *** tiagogomes has joined #buildstream | 08:55 | |
| qinusty | tristan: By that I assume you mean updating my branch to master, then having my 3 commits after? To avoid a merge commit? | 08:56 |
| gitlab-br-bot | buildstream: merge request (phil/203-BuildStream-crashes-when-dependency-tree-too-deep->master: Phil/203 build stream crashes when dependency tree too deep) #512 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/512 | 08:56 |
| tlater | qinusty: Yep, i.e., you forgot to use `git rebase` | 08:57 |
| tlater | tristan introduced me to that feature, was never allowed to do anything but merge elsewhere so it was a bit of an eye opener (: | 08:58 |
| qinusty | Yeah, I've just merged previously. It seems too good to be true, does it often fail on complex merges? | 09:01 |
| qinusty | s/merges/'merges' | 09:01 |
| *** brlogger has joined #buildstream | 09:02 | |
| *** coldtom has joined #buildstream | 09:02 | |
| tlater | So it doesn't fail per-se - it just tells you that it can't do it automatically | 09:02 |
| * tlater recommends using an editor/IDE that has UI for merge conflicts | 09:03 | |
| laurence | tristan and tlater - can we close the Local Cache Expiry ticket, and open a new one in its place that's not got the 'urgent' label? https://gitlab.com/BuildStream/buildstream/issues/135 | 09:03 |
| tlater | laurence: I've commented there, probably should just have done it in retrospect | 09:04 |
| *** gitlab-br-bot has joined #buildstream | 09:04 | |
| tristan | qinusty, fwiw, we do not allow any merging in master, we have enabled the symbolic "merge commits" as of this weekend, but "merges" are not allowed, fast-forward only on master or stable branches | 09:05 |
| tiagogomes | How does a UI helps for merge conflicts? You still need to manually edit chunks of code | 09:06 |
| tristan | and yes, first time use of `git rebase --interactive` is indeed awesome :) | 09:06 |
| *** tpollard has joined #buildstream | 09:07 | |
| tlater | tiagogomes: UIs for this tend to integrate this with the conflict display. I find it easier to keep track of what I'm doing when I have to solve large conflicts, but it's definitely not required. | 09:07 |
| tristan | tiagogomes, I think the UI people are referring to is the git CLI :) | 09:07 |
| tiagogomes | :) | 09:08 |
| tristan | oh, /me wrong... | 09:08 |
| * tristan never seen such a UI but is happy with the CLI UI | 09:08 | |
| *** bethw has joined #buildstream | 09:09 | |
| tristan | tlater, also bit confused... lets see... what deficit to we still have on that branch ? | 09:09 |
| * tristan thought it was going to be closed too | 09:09 | |
| tlater | tristan: We don't have any of the interactive stuff | 09:10 |
| tristan | right | 09:10 |
| tristan | yeah that's worth an issue | 09:10 |
| tlater | And the issue mentions the edge case of too much quota for total disk space - don't think we have that yet | 09:10 |
| * tlater should really have caught that one | 09:10 | |
| tristan | tlater, great, please file those separately :D | 09:10 |
| tristan | lets not grow that issue and close it | 09:11 |
| tlater | Am in the process of doing so :) | 09:11 |
| tristan | thanks ! | 09:11 |
| gitlab-br-bot | buildstream: merge request (Qinusty/275->master: Indicate where artifacts are going to and coming from in the log) #553 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/553 | 09:11 |
| *** jonathanmaw has joined #buildstream | 09:14 | |
| gitlab-br-bot | buildstream: issue #490 ("Introduce some interactivity to artifact cache expiry") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/490 | 09:18 |
| gitlab-br-bot | buildstream: issue #488 ("Follow-up from "Remove history from build trees"") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/488 | 09:20 |
| gitlab-br-bot | buildstream: issue #488 ("Follow-up from "Remove history from build trees"") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/488 | 09:20 |
| gitlab-br-bot | buildstream: merge request (tristan/git-stage-with-shared-clone->master: plugins/sources/git.py: Use --shared instead of --hardlinks) #557 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/557 | 09:20 |
| gitlab-br-bot | buildstream: merge request (Qinusty/275->master: Indicate where artifacts are going to and coming from in the log) #553 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/553 | 09:22 |
| gitlab-br-bot | buildstream: issue #491 ("Users can specify an artifact cache quota larger than their disk size") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/491 | 09:27 |
| gitlab-br-bot | buildstream: merge request (phil/436-add-ubuntu-install-intructions->master: Phil/436 add ubuntu install intructions) #525 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/525 | 09:28 |
| gitlab-br-bot | buildstream: issue #135 ("Expire artifacts in local cache") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/135 | 09:30 |
| jjardon | Hi, with the move to the new CAS server, does buildstream still depends on ostree? | 09:42 |
| tristan | jonathanmaw, so I gave some review comments regarding the mirroring thing, I think it's very close to landing | 09:42 |
| tristan | jjardon, good point, it is now only a dependency of the ostree source plugin and docs probably need an update | 09:43 |
| tristan | lemme check with setup.py | 09:43 |
| jjardon | tristan: right, I will try to create a patch | 09:43 |
| tristan | indeed | 09:43 |
| tristan | thanks | 09:43 |
| tiagogomes | There's a lot of 'ostree' matches on git grep | 09:44 |
| jjardon | seems there is tests/testutils/repo | 09:44 |
| jonathanmaw | tristan: tvm | 09:44 |
| tristan | jjardon, ah... for tests we have a thing... sec | 09:44 |
| *** mohan43u has joined #buildstream | 09:44 | |
| jmac | Phil: !512 needs rebasing, happy to merge it when that's done | 09:45 |
| tristan | jjardon, tests/testutils/site.py defines HAVE_OSTREE | 09:45 |
| tristan | jjardon, which should be used to disable any ostree using tests, for the repo, it already automatically skips any test which uses that repo in the case you dont have it | 09:45 |
| gitlab-br-bot | buildstream: issue #492 ("ostree is not a core dependency of buildstream anymore") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/492 | 09:46 |
| jjardon | ok, great | 09:46 |
| tristan | jjardon, that way tests are skipped when you dont have optional dependencies, and we just ensure we have the deps we want in CI | 09:46 |
| tristan | jonathanmaw, right so... the one concern I have is really only that SourceFetchers API seeming to be incomplete, i.e. to use it; git.py needs to poke into BuildStream internals | 09:49 |
| jjardon | tristan: Is pygobject needed, now that we do not use the ostree introspection data | 09:49 |
| jjardon | ? | 09:49 |
| tristan | jjardon, nope | 09:49 |
| tristan | it was only used for ostree | 09:49 |
| jjardon | ok | 09:49 |
| tristan | s/was/is | 09:50 |
| tristan | jonathanmaw, anyway I'm about to step out for food but not in a hurry, would you like to discuss the patch at all first ? | 09:54 |
| * jonathanmaw hasn't had time to digest the review yet, will poke you after food | 09:54 | |
| tristan | ok perfect :) | 09:54 |
| gitlab-br-bot | buildstream: issue #275 ("Indicate where artifacts are going to and comming from in the log") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/275 | 09:57 |
| gitlab-br-bot | buildstream: issue #275 ("Indicate where artifacts are going to and comming from in the log") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/275 | 09:57 |
| gitlab-br-bot | buildstream: merge request (Qinusty/275->master: Indicate where artifacts are going to and coming from in the log) #553 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/553 | 09:57 |
| gitlab-br-bot | buildstream: merge request (jjardon/ostree_repo->master: doc/source/install_linux_distro.rst: buildstream doesn't depend on ostree or pygobject anymore) #558 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 09:58 |
| *** tristan has quit IRC | 09:59 | |
| gitlab-br-bot | buildstream: merge request (jjardon/ostree_repo->master: doc/source/install_linux_distro.rst: buildstream doesn't depend on ostree or pygobject anymore) #558 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 09:59 |
| gitlab-br-bot | buildstream: merge request (phil/437-junction-tutorial->master: WIP Phil/437 junction tutorial) #550 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/550 | 10:26 |
| *** bethw has quit IRC | 10:32 | |
| *** bethw has joined #buildstream | 10:32 | |
| *** tristan has joined #buildstream | 10:43 | |
| *** bethw has quit IRC | 10:44 | |
| Nexus | tristan: you around? | 10:46 |
| Nexus | Just to clarify your comments about caching build trees on https://gitlab.com/BuildStream/buildstream/issues/21#note_89508744 and https://gitlab.com/BuildStream/buildstream/issues/21#note_89508744 You want an implementation that is ONLY the caching code? no tests, no workspace stuff, just the caching? | 10:47 |
| *** bethw has joined #buildstream | 10:47 | |
| gitlab-br-bot | buildstream: merge request (Qinusty/397->master: Modify retry behaviour to be opt-in on exceptions during job processing) #559 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/559 | 10:48 |
| gitlab-br-bot | buildstream: merge request (Qinusty/397->master: Modify retry behaviour to be opt-in on exceptions during job processing) #559 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/559 | 10:49 |
| qinusty | This fixes the issue you were seeing with the retries tristan, There may need to be more consideration into which implementations of BstError also get the 'temporary' option in their constructor. | 10:50 |
| gitlab-br-bot | buildstream: merge request (Qinusty/397->master: Modify retry behaviour to be opt-in on exceptions during job processing) #559 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/559 | 10:55 |
| gitlab-br-bot | buildstream: merge request (phil/437-junction-tutorial->master: WIP Phil/437 junction tutorial) #550 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/550 | 11:01 |
| tristan | Nexus, essentially yes, and without project.conf optionality; I didn't think about it not coming with tests but indeed, as long as it exposes no visible feature; it will not be properly testable | 11:04 |
| tristan | Nexus, I expect that to change soon but, if we can land this first part, it makes the cached build trees available sooner, so for instance, we can also use it for https://gitlab.com/BuildStream/buildstream/issues/76 | 11:05 |
| tristan | I also expect things will be easier to land in smaller chunks :) | 11:06 |
| tristan | At this same time, I will file a separate bug about ensuring that we only ever download the build tree separately and at the last possible moment (like only when we know we're going to need it), which I think still blocks on a small feature from CAS | 11:07 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees_limited->master: Adding caching build trees (limited)) #560 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/560 | 11:09 |
| Nexus | tristan: ^ | 11:09 |
| tristan | nice :) | 11:09 |
| * tristan takes a look... | 11:09 | |
| tristan | yeah, note that making it optional is exactly the thing I want to avoid; it's possible we get cornered into it for some reason, though | 11:10 |
| tristan | i.e. it changes the definition of what an artifact is, while optionality at download time is important either way | 11:11 |
| tristan | (having optionality is in the definition of an artifact seems to be potentially expensive code-wise as we add features which use this) | 11:11 |
| Nexus | Just a point for consideration, it may be worth making a plan at some point incase we end up having to make it optional | 11:12 |
| tristan | right | 11:12 |
| Nexus | left | 11:12 |
| gitlab-br-bot | buildstream: merge request (tpollard/386->master: _frontend: Don't print failure summary on build success) #561 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/561 | 11:12 |
| *** jonathanmaw_ has joined #buildstream | 11:13 | |
| *** jonathanmaw has quit IRC | 11:13 | |
| gitlab-br-bot | buildstream: merge request (tpollard/386->master: _frontend: Don't print failure summary on build success) #561 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/561 | 11:14 |
| tristan | hmmm, juergbi is not around today ? | 11:16 |
| Nexus | Holiday | 11:16 |
| Nexus | not that is means he won't be around... Engineers seem to not understand the concept of holidays | 11:17 |
| tristan | jmac, you are pretty familiar with the CAS stuff right ? as I understand it, when creating an artifact from the commit location, objects are copied from there, and the extract location always comes from the CAS object store right ? | 11:18 |
| tristan | I think anyway this makes sense in whatever scenario, so maybe I don't need to ask this question | 11:19 |
| * tristan comments on 560... | 11:19 | |
| tiagogomes | tristan, valentind Agustin would like me to work on ways to reducing the effort required for code review, and improve our confidence that when we release, we release with no regressions. Does any of you have any ideas of ways to achieve this? | 11:19 |
| jmac | tristan: Things extracted from the CAS have internal directory structure, but the CAS doesn't say anything about where a directory should be extracted to | 11:20 |
| tristan | jmac, err I phrased it wrong or confusing | 11:21 |
| paulsherwood | tiagogomes: smaller changes | 11:22 |
| tristan | jmac, I mean ~/.cache/buildstream/artifacts/extract is always hardlinks coming from ~/.cache/buildstream/artifacts/cas, correct ? | 11:22 |
| tristan | jmac, i.e. it never runs the opposite way, it would be possible that they were hardlinked from the build directory | 11:22 |
| paulsherwood | tiagogomes: and end-to-end (system) tests, rather than unit tests | 11:22 |
| tristan | but I dont think that could have happened to the code | 11:22 |
| jmac | tristan: I don't know what ~/.cache/buildstream/artifacts/extract is meant to contain at all, I'm afraid | 11:23 |
| tristan | tiagogomes, I am worried that with so much activity we're going to have regressions and we need to mitigate with better CI indeed | 11:23 |
| tristan | tiagogomes, we should discuss more... | 11:23 |
| tristan | jmac, ah, it's currently how we *use* CAS :) | 11:24 |
| paulsherwood | start by defining the core usecases we must not break | 11:24 |
| paulsherwood | (for ybd, this was cache-key calculations and full build of specific definitions checkouts, with and without artifact caches) | 11:25 |
| jmac | tristan: Is there any difference between hardlinking from the outside of CAS to inside and vice versa? I thought hard links were all done with ref counts | 11:25 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees_limited->master: Adding caching build trees (limited)) #560 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/560 | 11:25 |
| tlater | paulsherwood: We currently have no unit tests, only end-to-end tests | 11:25 |
| * Nexus bumps BST_CORE_ARTIFACT_VERSION | 11:26 | |
| tlater | Oh, well, besides some unit tests for our yaml parsers | 11:26 |
| Nexus | Didn't even know we did that :p | 11:26 |
| paulsherwood | tlater: cool :-) | 11:26 |
| tiagogomes | paulsherwood are you saying that ybd had a test script that cloned definitions and verified the tool output? | 11:26 |
| paulsherwood | yes | 11:26 |
| paulsherwood | for some value of 'verified' | 11:26 |
| paulsherwood | it still does | 11:27 |
| tristan | Nexus, all reviewed :) | 11:27 |
| tlater | paulsherwood, tiagogomes: We have such tests for buildstream at present :) | 11:27 |
| paulsherwood | so for example https://gitlab.com/baserock/ybd/blob/master/.gitlab-ci.yml#L30 | 11:27 |
| paulsherwood | this checks that any new version of ybd still calculates a v1 artifact cache-key correctly | 11:28 |
| paulsherwood | tlater: so if this is the case, how come tristan is worried about regressions? | 11:28 |
| tlater | tiagogomes: If you want to have a look at our test suite, I think that would still be appreciated. There may be some tests that are starting to show their age - but generally we trust our tests quite a lot. | 11:29 |
| tiagogomes | That would be a bit odd to depend in an external repo for the testing | 11:29 |
| paulsherwood | tiagogomes: what would be? | 11:29 |
| tlater | paulsherwood: Tests are never perfect - with the number of changes I think being a little paranoid is absolutely warranted. | 11:30 |
| paulsherwood | (the rest of the tests build a whole system on various OSes, with and without cached artifacts 'validate' is taken to mean 'built ok') | 11:30 |
| gitlab-br-bot | buildstream: merge request (Qinusty/397->master: WIP: Modify retry behaviour to be opt-in on exceptions during job processing) #559 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/559 | 11:30 |
| paulsherwood | tlater: i totally agree | 11:30 |
| * paulsherwood is more than a little paranoid himself | 11:31 | |
| tristan | There are parts which have eluded tests, source-bundle is an example, miracle it's still working but we've hand tested it | 11:37 |
| tristan | Another is everything interactive, this is a bigger concern as it's a growing surface which is easy to regress | 11:37 |
| tristan | We could use a technique to run bst in the test harness in such a way that we can expect what it prompts us, and we can answer it's user prompts | 11:37 |
| tristan | test that we can interactively debug a build, etc | 11:38 |
| tristan | Nexus, another thing popped up in my mind about that... it would be better I think to create the directory in the artifact unconditionally... because we'll need to download it and it will provide a more consistent structure for error handling around that | 11:39 |
| tristan | lemme comment in the issue | 11:39 |
| Nexus | i thought that too and have done, but comment it anyway for the record :) | 11:39 |
| tristan | ah :) | 11:40 |
| tristan | done | 11:41 |
| tiagogomes | Another thing suggested was automatic license and copyright check on the sources files | 11:42 |
| Nexus | tristan: should i still do the try:/except: ? since no other directory made that was is checked | 11:42 |
| tristan | tiagogomes, to cover paulsherwood's type of test case, we do have a plan for that since around GUADEC, jjardon suggested that we use a branch of freedesktop-sdk for testing the latest buildstream against those builds on a nightly basis | 11:42 |
| Nexus | s/was/way/ | 11:42 |
| tristan | tiagogomes, the rationale being, we want our CI to be brief and test small things individually, adding small tests as we go for regressions (and more importantly, being sure that we're testing against a constant set of data, so we know only a commit to BuildStream can cause BuildStream's CI to break) | 11:44 |
| tristan | however, having a nightly which is not attached to CI, and takes hours to run, to check latest BuildStream against a well known complex build, is also desirable | 11:45 |
| tristan | Nexus, I dont get "since no other directory made that was is checked", but it would be better I think yeah, you are basically testing whether the source dir exists, but ensuring the target still exists if no source dir existed | 11:47 |
| Nexus | tristan: for example `os.mkdir(logsdir)` is done, and logsdir is later used without any check being done | 11:48 |
| tristan | Nexus, creating the empty dir in the except case where the error was (I think FileNotFoundError ?), would be slightly better, even if mostly for encouraging better practices to others who read the code | 11:48 |
| Nexus | tristan: i was going to just do an `os.mkdir(buildtreedir)` seperate from the build tree linking logic, so it was made in the same way as the other directories | 11:49 |
| tristan | Nexus, I think that the distinction here is that we do not know if self.get_variable('build-root') exists at all | 11:49 |
| tristan | in the other cases, we've created it; in the case of the 'build-root' variable, it depends on the element | 11:50 |
| tristan | Nexus, ah you mean create the directory regardless, at the *source* side ? no don't do that | 11:51 |
| Nexus | tristan: there;s a default for the 'build-root' variable, it should always exist | 11:51 |
| tristan | Nope :) | 11:51 |
| tristan | Nexus, the result of doing that, will be to create a /buildstream/${element-name}.build/ directory unconditionally... in every import element artifact, for instance | 11:51 |
| tristan | Nexus, because an import element's files output is the root of the sandbox | 11:52 |
| tristan | Nexus, so that really has to be left up to the element | 11:52 |
| Nexus | tristan: i was refering to this: ./buildstream/data/projectconfig.yaml:57: build-root: /buildstream/%{project-name}/%{element-name} | 11:53 |
| tristan | Nexus, yes I know... I was improvising with my path name :) | 11:54 |
| Nexus | i mean as the default, assuming there is a project and an element (which there should be) it will have a default | 11:54 |
| Nexus | or am i misunderstanding something? | 11:54 |
| tristan | Nexus, did you understand what I mean about import elements ? | 11:54 |
| tristan | Nexus, the variable has a default yes, but not all elements use that variable | 11:55 |
| Nexus | ah | 11:55 |
| tristan | import will create an artifact of everything it imported (and if you create a directory for %{build-root}, you'll be adding that to the import outputs) | 11:56 |
| tristan | I think you understood anyway :) | 11:56 |
| Nexus | So what should i use as the dir in the case of it not having a "build-root" var? | 11:58 |
| tristan | Nexus, the build-root variable will be declared no matter what | 12:06 |
| tristan | Nexus, whether an Element uses that (whether sources or a build tree is even relevant for that element), is another question | 12:07 |
| tristan | Nexus, i.e. if the build-root does not exist in the sandbox, just fallback to making sure the 'buildtree' directory exists in the artifact we're creating | 12:07 |
| Nexus | tristan: ok thats fine, in which case do i really want a try:except:? since i won't be raising any erryor if the file doesn't exist, but instead just supplying an empty directory and not doing any linking | 12:22 |
| Nexus | surely just a check of "Does this variable exist" would suffice | 12:22 |
| Nexus | if self.get_variable('build-root'): Link_files else: just create and empty dir | 12:23 |
| Nexus | kind of thing | 12:23 |
| Nexus | s/and/an/ | 12:23 |
| tristan | Nexus, the try/except is in order to completely avoid os.path.isdir() | 12:26 |
| tristan | Nexus, consider it pretty low priority and partly stylistic. In general, it's just always better form to try the syscall and handle a specific error, than to issue a syscall in order to predict whether what you're going to do might fail | 12:27 |
| tlater | tristan: I think you're talking about slightly different things :) | 12:28 |
| tristan | Nexus, it's not immensely important, the biggest benefit we get from changing this to try/except is for other people reading the code to follow the same style | 12:28 |
| tristan | tlater, oh indeed | 12:28 |
| tristan | oops | 12:28 |
| tristan | Ummm, Nexus changing the structure such that in some cases 'build-root' is unset, is a larger activity | 12:29 |
| tristan | and I'm not 100% sure that redefining how that works would be an API compatible change | 12:29 |
| tristan | Nexus, currently, 'build-root' is always set; whether the element uses that and creates a directory is up to the element, that's the current contract | 12:30 |
| Nexus | kk | 12:32 |
| tristan | toscalix, I've just removed the ChangeLog part of the merge request template as we previously discussed | 12:53 |
| toscalix | ok | 12:53 |
| gitlab-br-bot | buildstream: issue #493 ("Allow staging part of a source") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/493 | 12:54 |
| qinusty | Do we have some tests which sometimes pass and sometimes fail? https://gitlab.com/BuildStream/buildstream/-/jobs/83660495 | 12:57 |
| tristan | qinusty, test_push_pull is the one I'm aware of... there has been some spurious errors around unmounting the sandbox | 13:04 |
| qinusty | test_build_track_track_first appears to have just done it on me tristan. If we're keeping track | 13:04 |
| tristan | qinusty, that looks new to me | 13:05 |
| tristan | i.e. failed to move the cloned repo into place in the sources cache | 13:06 |
| tristan | qinusty, it certainly also points out that we're failing to include the formatted OS error in the propagated error | 13:06 |
| tristan | git.py: line 121: should be adding the error into the formatted string | 13:07 |
| gitlab-br-bot | buildstream: merge request (Qinusty/397->master: WIP: Modify retry behaviour to be opt-in on exceptions during job processing) #559 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/559 | 13:08 |
| qinusty | Okay, I fixed the two lint issues I was getting. That MR should be good to fix #397 once the pipeline completes. | 13:08 |
| tristan | qinusty, looks like it could be racy though, i.e. one of our first bug reports... | 13:09 |
| tristan | https://gitlab.com/BuildStream/buildstream/issues/5 | 13:09 |
| qinusty | Consideration may need to be taken into what other BstError based Exceptions will need to have 'temporary' in their constructors tristan. | 13:09 |
| qinusty | 5, that's an early one | 13:09 |
| tristan | qinusty, if you have two parallel fetches of the same thing, one will beat the other and the second one it would appear, from git.py:121, would trigger that error | 13:10 |
| tristan | qinusty, since it appears you've added it to BstError, I don't think that is really a concern | 13:12 |
| tristan | qinusty, I.e. at the soonest point where we need to trigger another core error as 'temporary', that would be the time we add the 'temporary' option to chain into the specific error's constructor | 13:13 |
| qinusty | Okay, that sounds good to me. | 13:14 |
| jennis | qinusty, do you remember which buildstream command failed when you didn't have the ostree packages from the debian backports installed? | 13:18 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees_limited->master: Adding caching build trees (limited)) #560 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/560 | 13:18 |
| Nexus | tristan: It turns out that if you give utils.link_files() a non-existant dir as a source, it just does nothing | 13:19 |
| skullman | os.walk() will yield no results if the directory doesn't exist | 13:20 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees_limited->master: Adding caching build trees (limited)) #560 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/560 | 13:21 |
| tristan | Nexus, that is unfortunate... and contradictory to what the utils.link_files() API statement :-/ | 13:22 |
| qinusty | jennis: https://gitlab.com/BuildStream/buildstream/issues/480 is my closed issue | 13:22 |
| tristan | Nexus, I think that in itself is a bug, given that `ignore_missing` defaulting to false is exactly for this... and probably stems from how we handle directories differently than other files | 13:22 |
| * Nexus wonders why even the smallest of jobs he is given becomes bigger | 13:23 | |
| tristan | Nexus, or rather, its for exactly the reason skullman highlights, from list_relative_paths() | 13:23 |
| gitlab-br-bot | buildstream: merge request (Qinusty/397->master: WIP: Modify retry behaviour to be opt-in on exceptions during job processing) #559 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/559 | 13:24 |
| tristan | Nexus, it's alright, go ahead with os.path.isdir() for now, let's file this separately | 13:24 |
| tlater | Nexus: "Small" jobs are just how this whole programming thing works ;p | 13:25 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees_limited->master: Adding caching build trees (limited)) #560 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/560 | 13:26 |
| Nexus | tristan: done | 13:27 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees_limited->master: Adding caching build trees (limited)) #560 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/560 | 13:27 |
| jennis | mhmm, qinusty do you remember if you tried this *after* the CAS-based cache landed? | 13:30 |
| qinusty | Yup | 13:30 |
| qinusty | It was Thursday/Wednesday last week? | 13:31 |
| jennis | Yes thats what I thought | 13:31 |
| jennis | Because now I would expect that we don't require to install ostree from the backports repo as it's no longer a hard dependency | 13:31 |
| jennis | jjardon, your MR deals with this ^^ | 13:32 |
| jennis | ahh! | 13:32 |
| jennis | iirc, that example uses the ostree plugin | 13:32 |
| gitlab-br-bot | buildstream: merge request (Qinusty/397->master: WIP: Modify retry behaviour to be opt-in on exceptions during job processing) #559 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/559 | 13:34 |
| jennis | https://gitlab.com/BuildStream/buildstream/blob/master/doc/examples/flatpak-autotools/elements/base/sdk.bst, so it looks like we still need to ostree packages from the backport repo | 13:34 |
| tlater | Oh, right | 13:41 |
| tlater | Damn, I should remove the docker images issue | 13:42 |
| tlater | It's an optional dependency... | 13:42 |
| tlater | Also, we should (since it's now optional) depend on something more common for our tutorials | 13:42 |
| tlater | I.e., get OS tarballs instead of ostree links | 13:42 |
| jennis | tlater, it was seeing your activity on that issue that reminded me of all this :p | 13:46 |
| jennis | However, it's just the flatpak-autotools example that depends on this | 13:46 |
| tlater | jennis: To build flatpaks you need ostree ;) | 13:47 |
| tlater | Still, I think we should at least mention in the tutorial that this has to be installed | 13:47 |
| jennis | Indeed | 13:47 |
| tlater | Not sure if the current patch does that | 13:47 |
| jennis | It does not, I'll update the ticket | 13:47 |
| gitlab-br-bot | buildstream: issue #494 ("Only download the build tree if absolutely necessary") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/494 | 13:51 |
| *** Phil has quit IRC | 13:51 | |
| *** Phil has joined #buildstream | 13:51 | |
| tristan | Nexus, note that I filed https://gitlab.com/BuildStream/buildstream/issues/494 to account for optionality in downloads | 13:51 |
| toscalix | Filled my first bug in GNOME bugzilla. A couple more weeks and I see myself joining the foundation | 13:52 |
| tristan | :) | 13:57 |
| gitlab-br-bot | buildstream: merge request (328-support-for-downloading-sources-from-mirrors->master: Resolve "Support for downloading sources from mirrors") #404 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/404 | 13:58 |
| toscalix | laurence: if I recall correctly, tomorrow's monthly meeting is "mine" right? | 14:00 |
| laurence | yes | 14:01 |
| tristan | speaking of which... when is our next IRC team meeting coming up... | 14:03 |
| laurence | it's 14:00 UTC tomorrow | 14:03 |
| * laurence wonders if any countries in the world are in 'tomorrow' yet | 14:04 | |
| laurence | yes, but i don't think we have any contributors from those countries :) | 14:04 |
| Kinnison | laurence: it's gone midnight where tristan is IIRC | 14:05 |
| laurence | true! | 14:06 |
| jjardon | Hi, any idea what is the minimum required ostree version to use the ostree source plugin? | 14:06 |
| Kinnison | IIRC it's in the docs | 14:06 |
| tlater | jjardon: The docker image also lists it | 14:06 |
| * tlater recalls 2017.something | 14:06 | |
| Kinnison | libostree >= v2017.8 with introspection data | 14:06 |
| Kinnison | says https://buildstream.gitlab.io/buildstream/install_linux_distro.html | 14:06 |
| jjardon | that was when ostree was used ass a core dependency, we do not depend on ostree anymore | 14:07 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees_limited->master: Adding caching build trees (limited)) #560 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/560 | 14:07 |
| jjardon | only if you want to use the ostree source plugin | 14:07 |
| jjardon | I wonder if the ostree required version is more relaxed for that use case? | 14:07 |
| tlater | jjardon: 2017.8, apparently | 14:07 |
| qinusty | Regarding https://gitlab.com/BuildStream/buildstream/issues/481 How do people feel about https://photos.google.com/share/AF1QipPR9pI3MOW7oDGu37D-sWVtDl-LyaHXMkEeAVVN2Rssy6W047CdkMQ00sG3gMNtOQ?key=c2FOVlRkd2xLb2ZqTUhRSzctSDRuQjNNREJwQTFR? | 14:07 |
| tlater | Ah | 14:07 |
| qinusty | Thats a long link.... How do people feel about https://photos.app.goo.gl/bzwkiYVRWgGMpJdP7 **** | 14:08 |
| tlater | jjardon: I think the introspection was required to access it from python at all, which in turn meant we needed a specific version | 14:08 |
| jjardon | tlater: we do not need the instrospection at all now | 14:08 |
| tlater | Yep, that should bring down the dependency version a lot | 14:09 |
| tlater | I suspect we just need a version that allows all the commands we call | 14:09 |
| jjardon | context https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 14:09 |
| * toscalix has added tristan to the calendar event so he gets a mail notification about the IRC monthly meetings. | 14:10 | |
| tlater | qinusty: It's easier if you add that screenshot to your MR/issue and ask for comments on there - hard to see what you're showing with no context :) | 14:10 |
| gitlab-br-bot | buildstream: issue #495 ("utils.link_files()/utils.copy_files() APIs do not fail for non-existing source") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/495 | 14:10 |
| tlater | Cute terminal theme, I suppose? ;p | 14:10 |
| tlater | Solarized, or a variant of that, I think | 14:10 |
| qinusty | ;) The MR was also linked :P | 14:10 |
| qinusty | But I agree | 14:11 |
| tlater | Ah, missed the first link due to longness, sorry. | 14:11 |
| tristan | ... and added https://gitlab.com/BuildStream/buildstream/issues/495 to track the weird misbehavior of the utils.link_files()/utils.copy_files() public APIs | 14:12 |
| tlater | qinusty: Looks good. Interestingly, we never used SKIPPED for anything but jobs, so seeing it there might be a bit confusing. | 14:13 |
| tlater | But I think it's clear enough | 14:13 |
| tristan | toscalix, ummm, calendars aside, I think you've taken over laurence's responsibility for the call for agenda items and announcement around 1 week before the meeting | 14:13 |
| tiagogomes | erm, I can't build freedesktop-sdk because downloading http://download.icu-project.org/files/icu4c/61.1/icu4c-61_1-src.tgz is timing out | 14:13 |
| tristan | laurence, gah... the IRC team meeting is tomorrow ? | 14:14 |
| tristan | Oh is that what you meant by "mine" ? | 14:14 |
| tristan | Ok I dont mind how you guys want to share that | 14:14 |
| toscalix | it is | 14:14 |
| tristan | its a bit late though, did we not make noise in advance of the meeting ? | 14:14 |
| jjardon | tlater: seems bst still uses pygobject to access ostree at buildstream/_ostree.py, so I will rework the patch | 14:14 |
| tristan | Ah I found it | 14:14 |
| tristan | toscalix, sent out last tuesday right ? | 14:15 |
| jjardon | tristan: pygobject still needed for the ostree source plugin ^ | 14:15 |
| tristan | toscalix, thanks :D | 14:15 |
| tristan | jjardon, exactly | 14:15 |
| tristan | jjardon, but that means it's still an optional dependency | 14:15 |
| qinusty | rip tlater, snap | 14:15 |
| tlater | haha | 14:15 |
| tlater | qinusty: Do ask for code review from someone else, I should get back to "real" work | 14:16 |
| qinusty | :P No worries | 14:16 |
| tristan | ok so meeting is 11pm... I'll make a point to sleep in tomorrow haha | 14:17 |
| gitlab-br-bot | buildstream: merge request (jjardon/ostree_repo->master: WIP: doc/source/install_linux_distro.rst: buildstream doesn't depend on ostree or pygobject anymore) #558 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 14:17 |
| gitlab-br-bot | buildstream: issue #21 ("Caching of build trees") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/21 | 14:20 |
| * tiagogomes misses trove already | 14:21 | |
| laurence | i sent an email about tomorrow's meeting one week ago | 14:23 |
| noisecell | tiagogomes, trove is still a thing ;-) | 14:23 |
| laurence | ah, you found it | 14:23 |
| tristan | laurence, I didn't even realize it was you who sent it heh, drowning in emails, issues and merge requests | 14:24 |
| tiagogomes | Is there any option to force building everything from source instead of using the artifact cache? | 14:25 |
| laurence | very understandable | 14:25 |
| noisecell | tiagogomes, you can comment out the artifact cache in your project.conf | 14:26 |
| laurence | although it's pretty odd that we completely ignore the agenda that was sent a whole week in advance and then send that one today | 14:26 |
| tristan | tiagogomes, not currently, a request was made I think to ignore the artifact share downloads, but I think your request is slightly different ? | 14:26 |
| tristan | i.e., you'd also like to re-build locally cached artifacts ? | 14:27 |
| tiagogomes | That's correct | 14:27 |
| tristan | yeah, I think at the same level, a force rebuild thing has been discussed, but; I dont know if all of this is in the issue tracker | 14:28 |
| tristan | tiagogomes, it could be nice to ensure that we have an issue, for the most part, requests like yours so far have amounted to retrying a single element, so I think we might desire both possibilities (force rebuild a specific thing, or everything) | 14:29 |
| tristan | tiagogomes, as far as force rebuilding a specific thing, I've been thinking that a potential `bst artifact delete <element> <cache key>` API might be a decent solution | 14:30 |
| tristan | That, along with `bst artifact list` and the possibility of having something to manipulate artifacts directly, was discussed in a fairly recent issue before GUADEC... | 14:30 |
| tristan | lemme try to find it | 14:30 |
| tristan | https://gitlab.com/BuildStream/buildstream/issues/416 | 14:31 |
| tristan | tiagogomes, I should properly triage that issue I think, it's becoming too much of a side tracked discussion | 14:32 |
| * tristan adds a todo item for tomorrow morning... it's already 11:30 :) | 14:32 | |
| *** Phil has quit IRC | 14:39 | |
| *** Phil has joined #buildstream | 14:39 | |
| *** tristan has quit IRC | 14:41 | |
| tiagogomes | thanks :) | 14:41 |
| gitlab-br-bot | buildstream: merge request (jjardon/ostree_repo->master: WIP: doc/source/install_linux_distro.rst: buildstream doesn't depend on ostree or pygobject anymore) #558 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 14:42 |
| *** bethw has quit IRC | 14:42 | |
| gitlab-br-bot | buildstream: merge request (jjardon/ostree_repo->master: WIP: doc/source/install_linux_distro.rst: buildstream doesn't depend on ostree or pygobject anymore) #558 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 14:43 |
| gitlab-br-bot | buildstream: merge request (jjardon/ostree_repo->master: doc/source/install_linux_distro.rst: buildstream doesn't depend on ostree or pygobject anymore) #558 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 14:46 |
| *** bethw has joined #buildstream | 14:50 | |
| *** tristan has joined #buildstream | 14:52 | |
| gitlab-br-bot | buildstream: merge request (jmac/virtual_directories->master: WIP: Abstract directory class and filesystem-backed implementation) #445 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/445 | 14:53 |
| *** xjuan has joined #buildstream | 15:03 | |
| gitlab-br-bot | buildstream: merge request (Qinusty/481->master: Skipped sessions will no longer be reported as SUCCESS) #562 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/562 | 15:04 |
| qinusty | This MR is ready for review if anyone finds themselves with some free time https://gitlab.com/BuildStream/buildstream/merge_requests/562 | 15:06 |
| gitlab-br-bot | buildstream: merge request (phil/437-junction-tutorial->master: WIP Phil/437 junction tutorial) #550 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/550 | 15:17 |
| gitlab-br-bot | buildstream: merge request (jjardon/ostree_repo->master: doc/source/install_linux_distro.rst: buildstream doesn't depend on ostree or pygobject anymore) #558 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/558 | 15:29 |
| *** adds68 has quit IRC | 15:50 | |
| *** adds68 has joined #buildstream | 15:52 | |
| laurence | https://gitlab.com/BuildStream/buildstream/merge_requests/562 | 15:52 |
| laurence | i've never seen an image embedded in a ticket before | 15:53 |
| laurence | cool | 15:53 |
| *** toscalix has quit IRC | 15:57 | |
| gitlab-br-bot | buildstream: merge request (phil/437-junction-tutorial->master: WIP Phil/437 junction tutorial) #550 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/550 | 16:01 |
| gitlab-br-bot | buildstream: merge request (phil/437-junction-tutorial->master: Phil/437 junction tutorial) #550 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/550 | 16:12 |
| gitlab-br-bot | buildstream: merge request (jmac/virtual_directories->master: WIP: Abstract directory class and filesystem-backed implementation) #445 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/445 | 16:14 |
| gitlab-br-bot | buildstream: merge request (caching_build_trees->master: Caching buildtrees) #474 changed state ("closed"): https://gitlab.com/BuildStream/buildstream/merge_requests/474 | 16:33 |
| *** toscalix has joined #buildstream | 16:36 | |
| gitlab-br-bot | buildstream: merge request (jmac/virtual_directories->master: WIP: Abstract directory class and filesystem-backed implementation) #445 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/445 | 16:46 |
| WSalmon | i am trying to build https://gitlab.com/willsalmon/buildstream-kicad, i have run a few commands before but i have just run `bst track kicad.bst` then `bst build kicad.bst` then i get https://hastebin.com/ivuzakofih.md | 16:48 |
| tlater | WSalmon: Do the "wrong" shasums stay consistent between track attempts? | 16:49 |
| WSalmon | i think so, and i have tried deleting the files from ~/.cache/buildstream$ but they consistently down load to the same hash | 16:51 |
| tlater | Hrm, that sounds pretty bad | 16:52 |
| tlater | WSalmon: What version of buildstream are you running? | 16:52 |
| tlater | Should be in your output, hang, I'll stop being lazy | 16:52 |
| gitlab-br-bot | buildstream: merge request (jmac/virtual_directories->master: WIP: Abstract directory class and filesystem-backed implementation) #445 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/445 | 16:53 |
| tlater | WSalmon: Ah, you snipped that out | 16:53 |
| tlater | Buildstream version, please :) | 16:53 |
| tlater | (We try to keep everything functional at all times, but between MRs a broken commit *may* sneak in on occasion) | 16:54 |
| gitlab-br-bot | buildstream: merge request (bst_workspace_open_force_does_nothing->master: _stream.py: Added functionality for workspace open -f) #549 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/549 | 16:54 |
| gitlab-br-bot | buildstream: merge request (bst_workspace_open_force_does_nothing->master: _stream.py: Added functionality for workspace open -f) #549 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/549 | 16:55 |
| WSalmon | i have just re run both commands and get the same error. bst --version | 16:56 |
| WSalmon | 1.1.3+177.gba964287 | 16:56 |
| WSalmon | it was master last week | 16:56 |
| WSalmon | i will get master and try again | 16:56 |
| gitlab-br-bot | buildstream: merge request (Qinusty/491->master: Cache size is now restricted to available disk space) #563 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/563 | 16:58 |
| *** qinusty has quit IRC | 16:58 | |
| tlater | WSalmon: If that doesn't help, see if stable works | 16:59 |
| tlater | I'll see if I can reproduce it with your definitions in the mean time | 16:59 |
| tlater | Actually, let me try it with stable... | 16:59 |
| * tlater is really happy with bst-test, it makes trying out others' branches really simple :) | 17:02 | |
| WSalmon | fyi i dont think that project will work but it should get further than it dose | 17:02 |
| tlater | Oof, tracking's going to take a little while... | 17:03 |
| tlater | WSalmon: Ok, I see the same thing on stable, so I'm going to guess it's not a bug | 17:05 |
| tlater | Or if it is, it's related to http headers, and that site sends us the wrong file again | 17:06 |
| WSalmon | also just tride master and has grpc been recently added as a requirement? i get `ImportError: No module named 'grpc'` | 17:08 |
| jmac | It was added about a week ago | 17:08 |
| tlater | Hrm | 17:09 |
| tlater | Checking the downloaded source in the cache manually, it appears to have the sha256sum buildstream claims to be correct | 17:09 |
| tlater | It is just when buildstream actually checks whether the checksum matches that it's somehow wrong | 17:10 |
| * tlater wonders if there is a sha collision or something odd going on | 17:10 | |
| *** tristan has quit IRC | 17:17 | |
| tlater | WSalmon: This seems to only happen when building through the junction | 17:17 |
| tlater | I.e. this succeeds: bst build --track freedesktop-sdk.bst:base/gnutls.bst freedesktop-sdk.bst:base/gnutls.bst | 17:18 |
| tlater | I don't know enough about junctions to help much, unfortunately... | 17:18 |
| tlater | jmac and jonathanmaw_ still seem to be around, maybe they can help out, but I think we'd like a juergbi | 17:18 |
| gitlab-br-bot | buildstream: merge request (328-support-for-downloading-sources-from-mirrors->master: Resolve "Support for downloading sources from mirrors") #404 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/404 | 17:19 |
| *** ernestask has quit IRC | 17:19 | |
| Nexus | 404 merge equest not found | 17:21 |
| jmac | That's a weird one, I've not done much with tracking before | 17:22 |
| *** toscalix has quit IRC | 17:31 | |
| gitlab-br-bot | buildstream: merge request (tlater/message-helpers->master: WIP: Improve message helpers) #527 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/527 | 17:38 |
| gitlab-br-bot | buildstream: merge request (Qinusty/491->master: WIP: Cache size is now restricted to available disk space) #563 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/563 | 17:40 |
| gitlab-br-bot | buildstream: merge request (bst_workspace_open_force_does_nothing->master: _stream.py: Added functionality for workspace open -f) #549 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/549 | 18:00 |
| gitlab-br-bot | buildstream: merge request (328-support-for-downloading-sources-from-mirrors->master: Resolve "Support for downloading sources from mirrors") #404 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/404 | 18:17 |
| *** jonathanmaw_ has quit IRC | 18:21 | |
| *** jennis has quit IRC | 18:34 | |
| *** xjuan has quit IRC | 20:33 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!