IRC logs for #buildstream for Tuesday, 2019-01-15

« Monday, 2019-01-14 Index Wednesday, 2019-01-16 »
gitlab-br-botcs-shadow approved MR !1068 (jjardon/bst_external_0_9->master: .gitlab-ci.yml: Use latest bst-external) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/106800:01
gitlab-br-botcs-shadow closed issue #349 (Add plugin to generate docker (OCI) images) on buildstream https://gitlab.com/BuildStream/buildstream/issues/34900:05
*** igor has quit IRC00:10
*** igor has joined #buildstream00:10
*** ironfoot has joined #buildstream00:12
*** segfault3[m] has joined #buildstream00:19
gitlab-br-botcs-shadow opened issue #863 (Some tests fail when run as root) on buildstream https://gitlab.com/BuildStream/buildstream/issues/86300:19
*** ironfoot has quit IRC00:24
*** igor has quit IRC00:25
*** igor has joined #buildstream00:25
*** ironfoot has joined #buildstream00:27
*** ironfoot has quit IRC00:31
*** ironfoot has joined #buildstream00:32
*** ChanServ sets mode: +o ironfoot00:35
gitlab-br-botjjardon closed issue #861 (overnight-tests are failing: "buildstream-external does not contain a plugin named 'flatpak_repo'") on buildstream https://gitlab.com/BuildStream/buildstream/issues/86100:39
gitlab-br-botjjardon merged MR !1068 (jjardon/bst_external_0_9->master: .gitlab-ci.yml: Use latest bst-external) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/106800:39
*** igor has quit IRC00:40
*** igor has joined #buildstream00:40
gitlab-br-botcs-shadow opened (was WIP) MR !1038 (chandan/reqs-update-test->master: .gitlab-ci.yml: Add job to attemp to update dependencies) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/103800:54
*** igor has quit IRC00:55
*** igor has joined #buildstream00:55
jjardonbst has been packaged for debian! https://gitlab.com/BuildStream/buildstream/issues/549#note_13156334101:05
*** nimish has joined #buildstream01:07
*** nimish has quit IRC01:10
*** igor has quit IRC01:10
*** igor has joined #buildstream01:10
*** nimish has joined #buildstream01:14
gitlab-br-botcs-shadow merged MR !1038 (chandan/reqs-update-test->master: .gitlab-ci.yml: Add job to attempt to update dependencies) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/103801:21
*** igor has quit IRC01:25
*** igor has joined #buildstream01:25
*** igor has quit IRC01:40
*** igor has joined #buildstream01:40
*** phil has quit IRC01:53
*** igor has quit IRC01:55
*** igor has joined #buildstream01:55
*** tristan has joined #buildstream01:58
*** igor has quit IRC02:10
*** igor has joined #buildstream02:10
*** igor has quit IRC02:25
*** igor has joined #buildstream02:25
*** igor has quit IRC02:40
*** igor has joined #buildstream02:40
*** igor has quit IRC02:55
*** igor has joined #buildstream02:55
*** igor has quit IRC03:10
*** igor has joined #buildstream03:10
*** nimish has quit IRC03:12
*** igor has quit IRC03:25
*** igor has joined #buildstream03:25
*** igor has quit IRC03:40
*** igor has joined #buildstream03:40
*** igor has quit IRC03:55
*** igor has joined #buildstream03:55
*** igor has quit IRC04:10
*** igor has joined #buildstream04:10
*** igor has quit IRC04:25
*** igor has joined #buildstream04:25
*** mohan43u has quit IRC04:39
*** mohan43u has joined #buildstream04:39
*** igor has quit IRC04:40
*** igor has joined #buildstream04:40
*** igor has quit IRC04:55
*** igor has joined #buildstream04:55
*** igor has quit IRC05:10
*** igor has joined #buildstream05:10
*** igor has quit IRC05:25
*** igor has joined #buildstream05:25
*** tristan has quit IRC05:35
*** igor has quit IRC05:40
*** igor has joined #buildstream05:40
*** igor has quit IRC05:55
*** igor has joined #buildstream05:55
*** igor has quit IRC06:00
*** igor has joined #buildstream06:00
*** igor has quit IRC06:10
*** igor has joined #buildstream06:10
*** igor has quit IRC06:25
*** igor has joined #buildstream06:25
*** igor has quit IRC06:40
*** igor has joined #buildstream06:40
*** igor has quit IRC06:55
*** igor has joined #buildstream06:55
*** igor has quit IRC07:10
*** igor has joined #buildstream07:10
*** kapil___ has joined #buildstream07:24
*** igor has quit IRC07:25
*** igor has joined #buildstream07:25
*** igor has quit IRC07:40
*** igor has joined #buildstream07:40
*** igor has quit IRC07:55
*** igor has joined #buildstream07:55
*** alatiera has joined #buildstream08:09
*** igor has quit IRC08:10
*** igor has joined #buildstream08:10
*** igor has quit IRC08:25
*** igor has joined #buildstream08:25
*** igor has quit IRC08:40
*** igor has joined #buildstream08:40
*** igor has quit IRC08:55
*** igor has joined #buildstream08:55
*** igor has quit IRC09:10
*** igor has joined #buildstream09:10
*** igor has quit IRC09:25
*** igor has joined #buildstream09:25
*** phildawson_ has joined #buildstream09:26
*** toscalix has joined #buildstream09:29
gitlab-br-botBenjaminSchubert opened issue #864 (Local cache doesn't cleanup properly) on buildstream https://gitlab.com/BuildStream/buildstream/issues/86409:31
*** sambishop has joined #buildstream09:33
*** igor has quit IRC09:40
*** igor has joined #buildstream09:40
*** raoul has joined #buildstream09:45
*** igor has quit IRC09:55
*** toscalix has quit IRC09:55
*** igor has joined #buildstream09:55
*** jonathanmaw has joined #buildstream09:59
*** igor has quit IRC10:10
*** igor has joined #buildstream10:10
*** skullman has joined #buildstream10:12
*** nimish has joined #buildstream10:17
skullmanvalentind: in case you've not seen, we could do with your input on https://gitlab.com/BuildStream/buildstream/issues/833#note_131409089 to see if https://gitlab.com/BuildStream/buildstream/merge_requests/1066 is an appropriate fix10:19
valentindskullman, yes, I am looking at it.10:20
valentindI have a question about one of your comment.10:21
valentindhttps://gitlab.com/BuildStream/buildstream/issues/833#note_13132988510:21
valentindI am missing some context here I think.10:21
valentindWhat is the first graph produced by?10:22
skullmanthat was from me looking at what history the git describe test produced10:23
valentindWas this a comparison old git vs new git?10:23
valentindskullman, I am posting my comments. But I am still not sure about your fix, what it is for.10:24
skullmanThe first graph is the same in both. Old git produces the second, new git produces a diamond without the I commit of the first graph.10:24
*** igor has quit IRC10:25
valentindskullman, I am afraid your fix is for just one case here. There might be lots of possible graphs and it is hard to see whether it would fix all of them.10:25
*** igor has joined #buildstream10:25
valentindBut if we wanted to just fix it for old git versions, I would think we just traverse manually. We just remake the traversal from new git in BuildStream.10:26
valentindBut I think we are fine not fixing it.10:26
valentindThe reason is in my comment.10:26
*** lachlan has joined #buildstream10:31
valentindskullman, I think the best would be just to warn.10:32
valentindAnd disable the tests.10:32
*** nimish has quit IRC10:36
*** nimish has joined #buildstream10:37
*** igor has quit IRC10:40
*** igor has joined #buildstream10:40
*** lachlan has quit IRC10:47
*** igor has quit IRC10:55
*** igor has joined #buildstream10:55
skullmanvalentind: thanks for the ruling, so it's an unintended feature, but old git is an unusual enough circumstance that it's not something to warp the whole test-suite to support.11:01
skullmanI think this implies that we don't need the change to make it say exactly the same describe output too.11:02
valentindskullman, do you think the bug you found with some tags being marked as shallow could be reproduced with a newer git?11:03
*** lachlan has joined #buildstream11:03
valentindCould a light tag mark a commit as shallow and `git describe` not find the annotated tag that is an ancestor of it? For example.11:04
valentindWell, I think this should not happen. That was what "included" was for.11:06
valentindI suppose the differences you saw was with different tracking.11:07
skullmanyeah, and tracking included meant there was enough information to know that we _could_ not mark the commit as shallow since we had all its parents11:07
valentindOK, then it is all fine.11:08
skullmanI think it's not entirely a result of different tracking. We still had all the commits, just not the tag. I can experiment more.11:08
skullmanWe shouldn't have had those commits without the tag really, but that's a result of commit ranges giving counter-intuitive results unless you're using git log with --ancestry-path.11:09
valentindDo we include an extra commit?11:10
*** igor has quit IRC11:10
*** igor has joined #buildstream11:10
valentindskullman, I see a bug.11:11
valentindrev = line.lstrip('-')11:11
valentindThis should happen after testing line[0] == '-'11:11
valentindThis is strange, how did the tests pass?11:12
valentindOh no it is fine.11:13
valentindIt is line and rev.11:13
skullmanre: extra commits, yeah, in that diamond graph, if we didn't see tag1, we should only include the commits for HEAD and tag2, so M and F2.11:17
*** nimish has quit IRC11:17
skullmanWe get more because we work out which commits to include with `git rev-list $tag..HEAD`.11:17
skullman$tag..HEAD means "All commits reachable from HEAD minus those reachable by $tag" which means we get the left branch up until F1 too.11:17
skullmanIf we were to use `git log` instead we could use `--ancestry-path` which makes it only include the shortest path between them.11:17
*** kapil___ has quit IRC11:19
*** igor has quit IRC11:25
*** igor has joined #buildstream11:25
valentindskullman, I do not see the issue here actually. I got confused with the direction of the graph.11:27
valentindIt seems to me that we have the right commits. But we might have extra edges in the graphs.11:29
skullmanIt's times like this I wish I had a whiteboard…11:31
valentindskullman, how did you generate the second graph?11:31
valentindAre there any shared whiteboard online?11:31
skullmanthe second graph is what _rebuild_git does with an old version of git11:32
valentindSo rev-list behaved differently?11:32
valentindMaybe we need --full-history11:33
skullman  F1 ← refs/tags/tag111:33
skullman /11:33
skullmanF3  F2 ← refs/tags/tag211:33
skullman \  /11:33
skullman  M ← master, HEAD11:33
skullmanold version of git rev-list is why there isn't an edge between F1 and F2 I think.11:34
valentindLet me spawn a container with centos instead of asking lots of stupid questions.11:34
valentindI think they changed the default behavior of rev-list.11:34
*** nimish has joined #buildstream11:35
skullmanthe weird bit is that F3 and F1 get included even though we're not including tag1, since --first-parent doesn't exist in old git11:35
valentindskullman, agreed, this is strange.11:36
*** nimish has quit IRC11:38
*** igor has quit IRC11:40
*** igor has joined #buildstream11:40
valentindskullman, it seems there is a bug in --boundary in older git.11:41
valentindIt is missing the first parent as boundary.11:42
gitlab-br-botaevri opened (was WIP) MR !1061 (aevri/are_you_sure->master: BREAK:remove unconditional 'are you sure?' prompts) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/106111:48
*** lachlan has quit IRC11:50
*** lachlan has joined #buildstream11:52
valentindskullman, probably this: https://github.com/git/git/commit/4d826608e9610851d29440ca290e54b70192160811:52
*** igor has quit IRC11:55
*** lachlan has quit IRC11:55
*** igor has joined #buildstream11:55
valentindI was wrong maybe.12:05
*** igor has quit IRC12:10
*** igor has joined #buildstream12:10
*** igor has quit IRC12:25
*** igor has joined #buildstream12:25
*** lachlan has joined #buildstream12:26
gitlab-br-botvalentindavid opened MR !1069 (valentindavid/git-reduced-history->master: buildstream/_gitsourcebase.py: Reduce git history for git describe.) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/106912:26
*** lachlan has quit IRC12:30
*** igor has quit IRC12:31
*** raoul has quit IRC12:50
*** lachlan has joined #buildstream13:01
gitlab-br-botvalentindavid opened (was WIP) MR !1069 (valentindavid/git-reduced-history->master: buildstream/_gitsourcebase.py: Reduce git history for git describe.) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/106913:03
*** raoul has joined #buildstream13:38
*** nimish has joined #buildstream13:43
*** nimish has quit IRC13:46
*** nimish has joined #buildstream13:49
*** tristan has joined #buildstream13:59
*** tristan_ has joined #buildstream14:00
laurencemonthly irc team meeting starting on #buildstream-meetings14:00
*** tristan_ is now known as tristan14:01
*** nimish has quit IRC14:01
*** nimish has joined #buildstream14:03
*** nimish has quit IRC14:05
*** nimish has joined #buildstream14:05
*** lachlan has quit IRC14:32
*** lachlan has joined #buildstream14:38
*** ChanServ sets mode: +o tristan14:54
tristanphildawson, https://gitlab.com/BuildStream/buildstream/merge_requests/1067#note_131764842 ... did you get the version where the artifact cache expiry test hangs forever ?14:54
tristanphildawson, or is it your first time building the tox environment ?14:54
*** nimish has quit IRC14:55
*** nimish has joined #buildstream14:56
tristanlooks like tests are still having a hard time...14:56
*** lachlan has quit IRC15:01
phildawson_tristan, using tox for the tests works fine for me normally15:05
*** nimish has quit IRC15:06
phildawson_I went back and ran tox on master and repeated a few times to make sure it wasn't just a hiccough15:06
*** nimish has joined #buildstream15:06
laurenceDoes this page render in a strange way for anyone else? https://wiki.gnome.org/action/login/Projects/BuildStream/Events/Gathering-20190115:06
laurenceThe emails aren't displayed as hyperlinks15:06
phildawson_Renders fine for me15:07
laurencehhhmmm15:09
laurenceI wasn't logged in, now I am and it does....15:11
*** lachlan has joined #buildstream15:11
phildawson_tristan, I've re-fetched the branch and it seems to run more sensibly now. I guess I just pulled at the wrong time :)15:13
tristanphildawson_, yeah it's still got a last kink I overlooked15:14
tristanit runs generally well when I build, but is failing around where the BuildQueue goes on a personal quest to queue a Job on it's lonesome15:15
*** lachlan has quit IRC15:21
phildawson_tristan, Out of interest, have you looked at running any largish, say 500+ elements, pipelines using that branch? It'd be interesting to have an idea of the performance relative to master, especially as past 'fixes' for #712 have resulted in an overall slowdown.15:22
gitlab-br-botIssue #712: BuildStream spends a long time pulling/looking to pull before doing anything useful https://gitlab.com/BuildStream/buildstream/issues/71215:22
*** lachlan has joined #buildstream15:23
*** nimish has quit IRC15:26
*** nimish has joined #buildstream15:27
*** lachlan has quit IRC15:41
tristanphildawson_, I'm using it for freedesktop-sdk and building all from scratch, which was triggering the pull everything right away15:42
tristanphildawson_, it's not 500 no... but it doesn't seem to perform worse, and it reduces the overall data structures15:43
tristanThere is one area where I think there is potential to behave worse15:43
KinnisonIf it reduces the number of times Element._update_state() is called then it's a good thing15:44
tristanphildawson_, The thing is, tlater[m]'s changes caused the Queue.status() to stop being queried once it is ready - but it did so at the cost of choosing the wrong elements order to process15:45
KinnisonIf it increases the number of times Element._update_state() is called then it'll be a serious performance regression until such time as we can deal with _update_state() in general15:45
tristanBut really, it is the data model's place to do caching and ensure that expensive cache interrogations only happen when something changes or needs to be discovered15:46
tristanKinnison, it should not result in that, but it may result in additional calls to say, Source.get_consistency()... *IF* the Source class is not properly caching the value (but I think it is)15:48
tristanKinnison, or, additional calls to ArtifactCache lookups, if the Element class hasn't properly cached whether it is in the artifact cache or not15:49
tristanit's the questions that will be asked more often of Queue.status(), but that is necessary for the intended algorithm15:50
tristanand the buck should stop when the question reaches the Element or Source which returns immediately with a cached value if the call was redundant15:50
*** lachlan has joined #buildstream15:51
*** lachlan has quit IRC15:55
*** nimish has quit IRC15:57
*** nimish has joined #buildstream15:57
*** lachlan has joined #buildstream16:00
*** tristan has quit IRC16:02
gitlab-br-botBenjaminSchubert opened MR !1070 (bschubert/pipeline->master: Refactor _update_state() to be called only when needed) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107016:04
gitlab-br-botraoul.hidalgocharman opened MR !1071 (raoul/cas-refactor->master: Cas refactor) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107116:07
*** nimish has quit IRC16:27
*** nimish has joined #buildstream16:28
*** lachlan has quit IRC16:29
*** mohan43u has quit IRC16:29
*** nimish has quit IRC16:33
*** lachlan has joined #buildstream16:33
gitlab-br-botjjardon closed issue #806 (overnight tests are failing: "Error loading project: Invalid value for arch") on buildstream https://gitlab.com/BuildStream/buildstream/issues/80616:46
*** bochecha has quit IRC16:51
*** bochecha has joined #buildstream16:51
gitlab-br-botjennis opened MR !1072 (jennis/doc_fixes_in_context->master: Small documentaion/comment fixes in context.py) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107216:56
jennisHi guys, I've split out a couple fixes I found when working on the deprecation of push/pull: https://gitlab.com/BuildStream/buildstream/merge_requests/1072/diffs - should be a quick review16:57
*** lachlan has quit IRC17:00
gitlab-br-botjjardon closed issue #768 (Overnigth tests are failing for bst-1.2 branch) on buildstream https://gitlab.com/BuildStream/buildstream/issues/76817:01
gitlab-br-botjjardon closed issue #742 (Overnigth tests are failing: too many client connections to a specific repo) on buildstream https://gitlab.com/BuildStream/buildstream/issues/74217:02
*** abderrahim32 has joined #buildstream17:03
*** lachlan has joined #buildstream17:04
gitlab-br-botjmacarthur approved MR !1072 (jennis/doc_fixes_in_context->master: Small documentaion/comment fixes in context.py) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107217:04
jmacjennis: One minor comment on there but I'm happy with it17:05
gitlab-br-botjjardon opened issue #865 (Overnigth tests are failing: Seems runners doesn't have enough memory) on buildstream https://gitlab.com/BuildStream/buildstream/issues/86517:06
*** alatiera_ has joined #buildstream17:06
*** alatiera has quit IRC17:07
*** alatiera_ is now known as alatiera17:07
jennista jmac17:11
*** lachlan has quit IRC17:20
*** sambishop has quit IRC17:32
*** lachlan has joined #buildstream17:43
*** jonathanmaw has quit IRC18:01
*** xjuan has joined #buildstream18:06
*** raoul has quit IRC18:12
gitlab-br-botBenjaminSchubert opened (was WIP) MR !1070 (bschubert/pipeline->master: Refactor _update_state() to be called only when needed) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107018:18
gitlab-br-botBenjaminSchubert opened issue #866 (Local cache configuration is unclear and error prone) on buildstream https://gitlab.com/BuildStream/buildstream/issues/86618:24
*** nimish has joined #buildstream18:32
*** lachlan has quit IRC18:35
*** phildawson_ has quit IRC18:38
*** phildawson_ has joined #buildstream18:39
*** phil has joined #buildstream18:50
*** phildawson_ has quit IRC18:50
*** phil has quit IRC19:05
*** phil has joined #buildstream19:05
*** nimish has quit IRC19:10
*** xjuan has quit IRC20:24
gitlab-br-botcs-shadow opened MR !1073 (chandan/fix-pytest-get-marker->master: conftest.py: Don't use deprecated get_marker() function) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107320:50
*** xjuan has joined #buildstream21:24
*** alatiera has quit IRC21:36
*** bochecha has quit IRC22:03
gitlab-br-botcs-shadow opened MR !1074 (chandan/python37-tests->master: .gitlab-ci.yml: Add tests for python 3.7) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107422:44
gitlab-br-botjjardon approved MR !1074 (chandan/python37-tests->master: .gitlab-ci.yml: Add tests for python 3.7) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107422:48
gitlab-br-botjjardon closed issue #838 (Add tests for python 3.7) on buildstream https://gitlab.com/BuildStream/buildstream/issues/83822:48
gitlab-br-botjjardon merged MR !1074 (chandan/python37-tests->master: .gitlab-ci.yml: Add tests for python 3.7) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107422:48
*** xjuan has quit IRC23:02
*** xjuan has joined #buildstream23:03
*** xjuan has quit IRC23:18
*** xjuan has joined #buildstream23:19
gitlab-br-botjjardon merged MR !1073 (chandan/fix-pytest-get-marker->master: conftest.py: Don't use deprecated get_marker() function) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/107323:23
« Monday, 2019-01-14 Index Wednesday, 2019-01-16 »

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!