| *** mcatanzaro has quit IRC | 04:25 | |
| *** mcatanzaro has joined #buildstream | 04:26 | |
| *** mcatanzaro has quit IRC | 04:43 | |
| gitlab-br-bot | buildstream: merge request (juerg/element-state->master: element state updates) #215 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/215 | 05:07 |
|---|---|---|
| *** tristan has joined #buildstream | 07:05 | |
| gitlab-br-bot | buildstream: merge request (refactor-error-details-backport->bst-1.0: Refactor error details backport) #223 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/223 | 07:22 |
| *** valentind has joined #buildstream | 07:59 | |
| gitlab-br-bot | buildstream: merge request (remove-secret-command-lists->master: buildelement.py: Remove secret undocumented command lists) #224 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/224 | 08:07 |
| tristan | juergbi, so I'm looking at !215... | 08:09 |
| tristan | juergbi, I know it's dang early but I'm not gonna be around late today, as it is with fridays, lets see if we can merge it | 08:10 |
| * tristan not sure if juergbi is awake at this time | 08:10 | |
| tlater | It's about 9 am here, it's possible | 08:11 |
| tristan | Element._update_state() is a decent start, afaics this does achieve centralizing the logic of where a cache key is resolved | 08:13 |
| tristan | I wonder now about how the pipeline is initialized and such | 08:13 |
| gitlab-br-bot | buildstream: merge request (remove-secret-command-lists-backport->bst-1.0: Remove secret command lists backport) #225 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/225 | 08:17 |
| tristan | ok so we're still doing Element._force_inconsistent() for elements which will be tracked at load time | 08:21 |
| tristan | which is still a bit fragile | 08:21 |
| tristan | and we still have that Element._consistency() `recalculate` argument | 08:23 |
| tristan | juergbi, your patch is certainly an improvement, but it would be nice if it also covered source consistency in the way it handles element cached state | 08:25 |
| tristan | what I find interesting also about the Element._update_state() is that it could eventually take a `why` argument too | 08:26 |
| tristan | Which is kind of similar to the idea I had about sending events to some central logic which resolves state in one place | 08:26 |
| tristan | for now it's just one event "hey something changed" | 08:26 |
| tristan | (like how we had app->update_ui() in Glade, which made things horribly slow... hehe but the comparison is not really valid) | 08:27 |
| tristan | It seems also unfortunate that we cannot easily test that `a cache key is resolved only once` | 08:28 |
| tristan | if we could, then we could throw a bunch of scenarios at the test case which asserts that | 08:28 |
| gitlab-br-bot | buildstream: merge request (remove-secret-command-lists->master: buildelement.py: Remove secret undocumented command lists) #224 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/224 | 08:34 |
| gitlab-br-bot | buildstream: merge request (remove-secret-command-lists-backport->bst-1.0: Remove secret command lists backport) #225 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/225 | 08:34 |
| gitlab-br-bot | buildstream: merge request (refactor-error-details-backport->bst-1.0: Refactor error details backport) #223 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/223 | 08:36 |
| juergbi | tristan: i was considering doing something similar for sources but it seems to be sufficiently orthogonal that it should at least be a separate commit anyway, so i wanted to get the element part merged first | 09:03 |
| juergbi | also, there is no performance benefit on the source side. it would still be better for consistency, though | 09:04 |
| juergbi | testing: not sure i really like such a white box text | 09:05 |
| *** jennis has joined #buildstream | 09:05 | |
| tristan | Yeah; I care about a hundred times more about the code sanity this adds over the performance benefit it adds | 09:05 |
| tristan | juergbi, indeed, in the bug report I added comments which I think agree with that regarding tests | 09:05 |
| tristan | juergbi, I think maybe we might achieve that in a sneaky way when landing event hooks | 09:05 |
| juergbi | oh, didn't get notification for those, let me check | 09:06 |
| *** noisecell has joined #buildstream | 09:06 | |
| tristan | if we have interesting events to notify; we can use those to examine side effects | 09:06 |
| tristan | juergbi, yeah I didnt add much more than that detail there, other than what is discussed here :) | 09:06 |
| juergbi | there is also a case where the cache key will be determined (although not calculated) twice | 09:06 |
| * tristan was rather worried he would have to step out before juergbi arrives | 09:06 | |
| juergbi | i was actually here earlier to rebase it | 09:06 |
| tristan | hmmm, where is the issue report for event hooks :-S | 09:08 |
| * tlater doesn't think there is one | 09:09 | |
| tristan | well - I guess I imagined event hooks | 09:09 |
| * tlater should make one | 09:09 | |
| tlater | Hm, I guess it's only on the ML atm | 09:10 |
| *** jennis has quit IRC | 09:13 | |
| juergbi | tristan: on the source side things are already pretty explicit. it doesn't have a single _update_state()-like method but that's because the changes in state can't be determined automatically | 09:14 |
| juergbi | e.g., for tracked elements, we anyway need to have a call such as force_inconsistent, otherwise the element won't know this | 09:14 |
| juergbi | we still have the recalculate thing, though, maybe i can get rid of that | 09:16 |
| tristan | juergbi, I wonder if "force inconsistent" is the right thing too, the naming here really makes a huge difference in the perception of roles and responsibilities | 09:17 |
| juergbi | good point. that name initially confused me | 09:18 |
| tristan | juergbi, i.e. if we were telling the element rather that we're "scheduling a track", then it's the responsibility of the element | 09:18 |
| tristan | to resolve it at the right time | 09:18 |
| juergbi | yes, we could have something like that | 09:18 |
| juergbi | and then move update_state out of the else | 09:19 |
| juergbi | in pipeline | 09:19 |
| tristan | So, right now I am tempted to merge this in order to unblock other work depending on it | 09:19 |
| gitlab-br-bot | buildstream: issue #188 ("Trigger external commands on certain events") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/188 | 09:19 |
| tristan | And only reluctant for a paranoid worry that the rest of this cleanup will take the back seat | 09:20 |
| tristan | juergbi, shall we merge straight away, and I leave it to you to do the rest ? | 09:21 |
| tristan | I'm okay with that, but would really love to get this worked out too... | 09:21 |
| juergbi | yes, i can improve the source/consistency aspect | 09:21 |
| juergbi | might be Monday, though, don't really have time today | 09:21 |
| tristan | yeah no worry | 09:22 |
| tristan | more reason to merge earlier | 09:22 |
| tristan | juergbi, I think it's not a lot of work for you, and I will at the same time be tied up with a real review of junctions | 09:22 |
| juergbi | yes, shouldn't be much work | 09:23 |
| juergbi | i expect i will also be working on rebasing junction support on Monday now that multi-cache is in master | 09:23 |
| tlater | UM | 09:24 |
| tristan | It's highly probably that the only reason to ever hold back on resolving consistency is for the sake of tracking (why I had it explicitly the other way around initially), but we we're wrong before | 09:24 |
| * tlater just created a branch called 'e' on accident | 09:24 | |
| tlater | Damnit gitlab | 09:24 |
| tristan | tlater, git push origin :e | 09:24 |
| gitlab-br-bot | buildstream: merge request (188-trigger-external-commands-on-certain-events->master: Resolve "Trigger external commands on certain events") #226 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/226 | 09:25 |
| gitlab-br-bot | buildstream: merge request (188-trigger-external-commands-on-certain-events->master: WIP: Resolve "Trigger external commands on certain events") #226 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/226 | 09:26 |
| *** jude has joined #buildstream | 09:26 | |
| tlater | The integration tests currently take about 30 mins on my machine :| | 09:33 |
| gitlab-br-bot | buildstream: merge request (show_sources->master: Added a flag to bst show) #222 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/222 | 09:34 |
| * nexus guesses that's not a good thing | 09:35 | |
| gitlab-br-bot | buildstream: merge request (modAndTest->master: Making changes to various documents:) #206 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/206 | 09:37 |
| tlater | 42 minutes so far on gitlab, but that's a bit deceptive since it hasn't cached the base platform yet | 09:37 |
| * tlater wonders if we could justify adding a 3G tarfile of a flatpak runtime for integration tests. | 09:37 | |
| nexus | How fast is your machine? | 09:39 |
| tristan | caches on gitlab seem broken | 09:39 |
| tristan | since maybe the upgrade last week | 09:39 |
| nexus | Toaster, Average, Fast | 09:39 |
| nexus | how so? | 09:39 |
| tlater | i5 | 09:39 |
| nexus | that's the equivalent of saying "it's a blue one" | 09:40 |
| tlater | But that barely matters, overhead is writing speeds | 09:40 |
| tlater | 2GHz | 09:40 |
| nexus | ah ok | 09:40 |
| nexus | SSD? | 09:40 |
| tlater | But I do have an SSD on this | 09:40 |
| tlater | Yeah | 09:40 |
| nexus | try running resource manager as you do an int test and see how much it uses | 09:40 |
| nexus | then you can see if its a hardware bottleneck | 09:41 |
| tlater | Well, the problem is always going to be running this on gitlab, not locally | 09:41 |
| nexus | ahh i see | 09:42 |
| tlater | It just seems that removing the second runtime didn't affect performance much | 09:43 |
| tlater | Which means the revolution isn't that revolutionary... | 09:43 |
| tlater | :| | 09:43 |
| nexus | :( | 09:43 |
| gitlab-br-bot | buildstream: merge request (remove-secret-command-lists-backport->bst-1.0: Remove secret command lists backport) #225 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/225 | 10:03 |
| *** jonathanmaw has joined #buildstream | 10:05 | |
| gitlab-br-bot | buildstream: merge request (remove-secret-command-lists->master: buildelement.py: Remove secret undocumented command lists) #224 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/224 | 10:06 |
| *** valentind has quit IRC | 10:07 | |
| nexus | How am i supposed to learn the codebase if people are keeping secrets from me!? | 10:12 |
| tlater | Does anyone have an idea why gitlab's CI doesn't include my symlinks? | 10:15 |
| * tlater can't get his "usrmerge" element to work | 10:15 | |
| tlater | I guess I can try writing a script element that does it... | 10:16 |
| *** ssam2 has joined #buildstream | 10:18 | |
| tristan | juergbi, ok so I just landed some minor things... and escaping presently | 10:20 |
| tristan | juergbi, can you just rebase and merge !215 | 10:20 |
| tristan | ? | 10:20 |
| juergbi | sure, will do | 10:20 |
| juergbi | enjoy your dinner | 10:21 |
| tristan | thanks ! | 10:21 |
| gitlab-br-bot | buildstream: merge request (juerg/element-state->master: element state updates) #215 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/215 | 10:24 |
| *** tristan has quit IRC | 10:26 | |
| *** csoriano has quit IRC | 10:39 | |
| *** csoriano has joined #buildstream | 10:40 | |
| gitlab-br-bot | buildstream: merge request (modAndTest->master: Making changes to various documents:) #206 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/206 | 11:04 |
| jonathanmaw | tristan: hrm, it looks like I'm not seeing the "detail" field get printed in the tests' output | 11:19 |
| jonathanmaw | should I be pulling it out of the exception that the Result produces, instead? | 11:20 |
| jonathanmaw | Or is there a way to use assert_task_error to pull it out? | 11:20 |
| jonathanmaw | also, after rebasing against master, a lot of my tests are failing with "AttributeError: 'LocalPath' object has no attribute 'startswith'". I'm guessing that this is a python3.4 vs. python3.5 problem. | 11:21 |
| tlater | jonathanmaw: We are supposed to support 3.4, probably worth fixing | 11:28 |
| *** adds68 has quit IRC | 11:36 | |
| *** ltu has left #buildstream | 11:45 | |
| *** ltu has joined #buildstream | 11:46 | |
| gitlab-br-bot | buildstream: issue #167 ("Workspaced element's cache keys are calculated 4 times during a build") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/167 | 11:52 |
| gitlab-br-bot | buildstream: issue #149 ("Busyloop while calculating cache keys in `bst build --track`") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/149 | 11:52 |
| gitlab-br-bot | buildstream: issue #149 ("Busyloop while calculating cache keys in `bst build --track`") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/149 | 11:52 |
| gitlab-br-bot | buildstream: issue #173 ("Element state / cache key handling") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/173 | 11:52 |
| gitlab-br-bot | buildstream: issue #173 ("Element state / cache key handling") changed state ("closed") https://gitlab.com/BuildStream/buildstream/issues/173 | 11:52 |
| gitlab-br-bot | buildstream: merge request (juerg/element-state->master: element state updates) #215 changed state ("merged"): https://gitlab.com/BuildStream/buildstream/merge_requests/215 | 11:52 |
| tlater | Ack, apparently pytest doesn't propagate environment variables | 11:55 |
| * tlater isn't sure how to trigger unix tests now... | 11:55 | |
| ssam2 | i think it can if you tell it to | 11:56 |
| ssam2 | you can set 'env' in the pytest.ini file | 11:56 |
| ssam2 | not sure how to do that conditionally though | 11:56 |
| tlater | I suppose I can write that file in .gitlab.ci.yaml | 11:56 |
| ssam2 | yeah | 11:57 |
| *** adds68 has joined #buildstream | 12:11 | |
| *** xjuan has joined #buildstream | 12:26 | |
| gitlab-br-bot | buildstream: merge request (issue-182_Closing_non-existing_workspace->master: WIP: Fix for issue 182 Closing non-existing workspace) #227 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/227 | 12:36 |
| gitlab-br-bot | buildstream: merge request (issue-182_Closing_non-existing_workspace->master: WIP: Fix for issue 182 Closing non-existing workspace) #227 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/227 | 12:37 |
| gitlab-br-bot | buildstream: merge request (issue-182_Closing_non-existing_workspace->master: Fix for issue 182 Closing non-existing workspace) #227 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/227 | 12:38 |
| tlater | Integration tests passing on linux \o/ | 12:51 |
| gitlab-br-bot | buildstream: merge request (issue-181_bst_build_--track-except->master: Fix for issue #181) #228 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/228 | 13:01 |
| jonathanmaw | aha! the detail disappeared from my error message because it was being reraised as a PipelineError | 13:05 |
| jonathanmaw | hrm, I need to figure out the right way to reraise it | 13:06 |
| jonathanmaw | since PipelineError already has a detail field | 13:06 |
| *** ernestask has joined #buildstream | 14:11 | |
| *** mcatanzaro has joined #buildstream | 14:18 | |
| jjardon[m] | tristan: we can have freebsd runners | 14:35 |
| jjardon[m] | but I do not know anything about freebsd so it will take some time | 14:35 |
| jjardon[m] | to setup | 14:35 |
| jjardon[m] | maybe open an issue and assign it to me? | 14:35 |
| persia | lantw44 may also be interested | 14:37 |
| gitlab-br-bot | buildstream: merge request (164-minimise-overlaps-by-having-overlaps-raise-exceptions-unless-configured-not-to->master: Resolve "Minimise overlaps by having overlaps raise exceptions unless configured not to") #181 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/181 | 15:21 |
| *** ssam2 has quit IRC | 16:02 | |
| *** ssam2 has joined #buildstream | 16:04 | |
| *** jmac has joined #buildstream | 16:13 | |
| *** cs_shadow has joined #buildstream | 16:32 | |
| *** valentind has joined #buildstream | 16:44 | |
| *** adds68_ has joined #buildstream | 17:01 | |
| *** adds68 has quit IRC | 17:03 | |
| cs_shadow | Hi. https://gitlab.com/BuildStream/buildstream/blob/master/integration-tests/compose-test/elements/dependencies/amhello-full.bst#L10 looks a bit suspect. The url starts with "file:///home/tristanmaat/Documents...", is that intentional? | 17:08 |
| cs_shadow | Similar url scheme is also present in couple other tests | 17:08 |
| gitlab-br-bot | buildstream: merge request (incremental-build->master: WIP: Add support for doing incremental builds) #126 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/126 | 17:11 |
| persia | cs_shadow: Good catch. Even if the test works in CI, the patch should probably be set to a role name. | 17:11 |
| cs_shadow | I agree, will submit an issue | 17:17 |
| ssam2 | at least we can tell who introduced the bug :-) | 17:34 |
| tlater | cs_shadow, yep, these are being sorted out | 17:37 |
| cs_shadow | great, thanks! | 17:38 |
| tlater | They are remnants of too heavy-handed commits on my end, though the suite patched the paths out for now | 17:38 |
| *** noisecell has quit IRC | 17:39 | |
| *** luc14n0 has quit IRC | 17:41 | |
| *** luc14n0 has joined #buildstream | 17:41 | |
| gitlab-br-bot | buildstream: issue #189 ("Performance issues with running the artifact cache") changed state ("opened") https://gitlab.com/BuildStream/buildstream/issues/189 | 17:49 |
| gitlab-br-bot | buildstream: merge request (jonathan/fix-tests-python34->master: Make artifact cache tests work in python3.4) #229 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/229 | 17:56 |
| *** ssam2 has quit IRC | 18:03 | |
| jjardon[m] | Hi, my build have failed but I can not see the problem in the normal output: https://gitlab.com/freedesktop-sdk/freedesktop-sdk/-/jobs/47780877 Is there an option to tell bst to show all the build logs, or where buildstream store the build logs? | 18:18 |
| *** jude has quit IRC | 18:32 | |
| *** tiago has quit IRC | 18:33 | |
| gitlab-br-bot | buildstream: merge request (incremental-build->master: WIP: Add support for doing incremental builds) #126 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/126 | 19:01 |
| *** jonathanmaw has quit IRC | 19:02 | |
| *** xjuan has quit IRC | 21:05 | |
| *** ernestask has quit IRC | 21:16 | |
| *** adds68__ has joined #buildstream | 21:40 | |
| *** adds68_ has quit IRC | 21:41 | |
| *** tristan has joined #buildstream | 22:35 | |
| *** luc14n0 has left #buildstream | 22:55 | |
| *** mcatanzaro has quit IRC | 23:10 | |
| *** mcatanzaro has joined #buildstream | 23:10 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!