| *** xjuan has quit IRC | 01:12 | |
| *** nimish has quit IRC | 01:37 | |
| *** nimish has joined #buildstream | 03:03 | |
| *** nimish has quit IRC | 03:06 | |
| *** nimish has joined #buildstream | 03:22 | |
| *** bochecha has joined #buildstream | 03:29 | |
| *** kapil___ has joined #buildstream | 03:58 | |
| *** bochecha has quit IRC | 05:56 | |
| *** kapil___ has quit IRC | 06:37 | |
| *** kapil___ has joined #buildstream | 07:54 | |
| *** toscalix has joined #buildstream | 08:28 | |
| *** alatiera has joined #buildstream | 09:27 | |
| *** jonathanmaw has joined #buildstream | 09:39 | |
| *** raoul has joined #buildstream | 10:10 | |
| jjardon | Maybe of interest here: http://julio.meroh.net/2019/02/hello-sandboxfs-0.1.0.html | 10:29 |
|---|---|---|
| Kinnison | I saw that yesterday, yeah it's similar to the intent for buildbox-fuse I think | 10:31 |
| gitlab-br-bot | aevri opened (was WIP) MR !891 (aevri/include-error->master: More user-friendly reporting on include errors) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/891 | 10:32 |
| juergbi | not CAS-based, though | 10:32 |
| Kinnison | indeed | 10:33 |
| Kinnison | I think Bazel only uses CAS for RE | 10:33 |
| *** nimish has quit IRC | 10:34 | |
| *** sambishop has joined #buildstream | 10:35 | |
| valentind | Interesting thing. On bst 1.2, the permissions of files are different depending on the builder. | 10:42 |
| valentind | Sometimes files are in 644, some times in 600. And the reason is that we link files from the cache. | 10:43 |
| Kinnison | U asks? | 10:44 |
| Kinnison | Gah | 10:44 |
| Kinnison | umasks? | 10:44 |
| valentind | My umask is 022. | 10:44 |
| valentind | So there is something else on my machine that get it to 077. | 10:44 |
| phildawson | looks like the overnight tests failed last night, though I don't understand why from the logs. https://gitlab.com/BuildStream/buildstream/pipelines/46277177 | 10:44 |
| valentind | Not all my files are in 0600. | 10:44 |
| valentind | Some are 644. | 10:45 |
| gitlab-br-bot | tpollard opened issue #896 (Optional creation of buildtrees) on buildstream https://gitlab.com/BuildStream/buildstream/issues/896 | 10:45 |
| jjardon | phildawson: those test have been failing for months; " c++: fatal error: Killed signal terminated program cc1plus" seems the runner got out of memory. Maybe related with https://gitlab.com/BuildStream/buildstream/issues/874 ? | 10:49 |
| phildawson | Ah I see. Thanks jjardon | 10:51 |
| jjardon | Does buildstream have its own cache server? Seems the overnight build is building everything from scratch, probably because the cache key of bst master is different from the one in bst-1.2 | 10:54 |
| tpollard | nafaik | 10:58 |
| *** lachlan has joined #buildstream | 10:59 | |
| gitlab-br-bot | valentindavid opened MR !1132 (valentindavid/pull-chmod-bug-1.2->bst-1.2: buildstream/_artifactcache/cascache.py: Set 0644 rights to pulled files) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1132 | 11:00 |
| valentind | Kinnison, found the bug. I did it. Fixed it in !1132 | 11:01 |
| *** nimish has joined #buildstream | 11:02 | |
| * Kinnison responds to your MR :D | 11:02 | |
| gitlab-br-bot | jjardon opened issue #897 (Setup cache server) on buildstream https://gitlab.com/BuildStream/buildstream/issues/897 | 11:04 |
| tpollard | aha, needed to reload bst with pip to pick up roaringbitmap landing | 11:15 |
| *** nimish has quit IRC | 11:17 | |
| tpollard | hmm, pip claims to have installed it but I can't run the tests, 'ModuleNotFoundError: No module named 'roaringbitmap'' | 11:17 |
| tpollard | any ideas? | 11:17 |
| *** nimish has joined #buildstream | 11:17 | |
| jjardon | Are the docker images provided by buildstream updated with the new dependency? IIRC there was some automation for that but not sure | 11:19 |
| gitlab-br-bot | martinblanchard opened issue #898 (Implement the REAPI Platform message) on buildstream https://gitlab.com/BuildStream/buildstream/issues/898 | 11:25 |
| jjardon | Is this new dependency? https://pypi.org/project/roaringbitmap/ | 11:26 |
| benschubert | jjardon: yes, it has been added yesterday | 11:26 |
| benschubert | the docker images should have it yes | 11:26 |
| tpollard | it's added to requirements.txt, but I'm not sure what's happening | 11:27 |
| tpollard | pip3 claims it's installed, and tox loads requirements.txt | 11:27 |
| jjardon | I think the license of that project is incompatible with BuildStream (roaringbtimap is GPL, but buildstream is LGPL) | 11:28 |
| *** toscalix has quit IRC | 11:30 | |
| benschubert | ouch true | 11:31 |
| *** tpollard has quit IRC | 11:31 | |
| *** WSalmon_ has quit IRC | 11:31 | |
| benschubert | I thought it was only for linking against, which we are not? (Not a lawyer though) | 11:31 |
| *** sambishop has quit IRC | 11:31 | |
| *** lachlan has quit IRC | 11:31 | |
| benschubert | Kinnison: I guess we need to go with the other fix then | 11:32 |
| *** jonathanmaw has quit IRC | 11:35 | |
| *** sambishop has joined #buildstream | 11:35 | |
| *** tpollard has joined #buildstream | 11:35 | |
| *** WSalmon_ has joined #buildstream | 11:35 | |
| *** phildawson_ has joined #buildstream | 11:35 | |
| *** phildawson has quit IRC | 11:35 | |
| *** toscalix has joined #buildstream | 11:35 | |
| *** raoul_ has joined #buildstream | 11:35 | |
| *** jonathanmaw has joined #buildstream | 11:35 | |
| *** raoul has quit IRC | 11:35 | |
| tpollard | tests work outside of tox at least | 11:41 |
| benschubert | tpollard: does tox --recreate works? | 11:43 |
| *** lachlan has joined #buildstream | 11:46 | |
| tpollard | benschubert: that's done the trick | 11:46 |
| Kinnison | jjardon: Hrm, urgh | 11:48 |
| Kinnison | benschubert: Perhaps yes | 11:49 |
| tpollard | somehow I missed that section in contributing... cheers benschubert | 11:49 |
| *** kapil___ has quit IRC | 12:14 | |
| *** kapil___ has joined #buildstream | 12:25 | |
| *** slaf_ has joined #buildstream | 12:25 | |
| *** slaf_ has joined #buildstream | 12:25 | |
| *** slaf has quit IRC | 12:27 | |
| *** slaf_ is now known as slaf | 12:27 | |
| *** raoul_ has quit IRC | 12:30 | |
| *** raoul_ has joined #buildstream | 12:36 | |
| WSalmon | https://gitlab.com/BuildStream/buildstream/merge_requests/1125#note_138688047 this was caused by repacing all the tabs in a file with spaces, is there a reson that most of our licence blocks have a random tab in them? | 12:49 |
| WSalmon | juergbi, tristan etc ^ | 12:50 |
| juergbi | oh, no idea, that sounds wrong to me | 12:51 |
| WSalmon | i forgot to set my editor to spaces only and then fixed it and that one got caught, i think i have had similar things before | 12:52 |
| gitlab-br-bot | jjardon opened issue #899 (Possible license incompatibility; roaringbitmap is GPL while BuildStream is LGPL) on buildstream https://gitlab.com/BuildStream/buildstream/issues/899 | 13:10 |
| *** bochecha has joined #buildstream | 13:12 | |
| tpollard | urk | 13:13 |
| Kinnison | Dear tox, please notice when I update requirements.txt. No love, me. | 13:34 |
| benschubert | Kinnison: that is the case when you don't use a requirement file x') | 13:38 |
| * Kinnison is confuse | 13:38 | |
| gitlab-br-bot | danielsilverstone-ct opened MR !1133 (danielsilverstone-ct/other-roaring->master: loadelement.py: Switch from roaringbitmap to pyroaring) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1133 | 13:46 |
| Kinnison | jjardon: https://gitlab.com/BuildStream/buildstream/merge_requests/1133 switches us from roaringbitmap to pyroaring (which is MIT licensed) | 13:47 |
| Kinnison | jjardon: it also saves a bit more RAM which is nice | 13:47 |
| *** raoul has joined #buildstream | 13:47 | |
| *** raoul_ has joined #buildstream | 14:00 | |
| *** raoul has quit IRC | 14:02 | |
| gitlab-br-bot | BenjaminSchubert approved MR !1133 (danielsilverstone-ct/other-roaring->master: loadelement.py: Switch from roaringbitmap to pyroaring) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1133 | 14:08 |
| *** tristan has joined #buildstream | 14:11 | |
| jjardon | Kinnison: cool! I've updated the Mr to close #899 automatically when it gets merged | 14:20 |
| *** kapil___ has quit IRC | 14:34 | |
| *** raoul_ has quit IRC | 14:47 | |
| *** raoul_ has joined #buildstream | 14:48 | |
| Kinnison | jjardon: Thanks | 15:01 |
| Kinnison | jjardon: Would you like to hit the 'merge on pipeline success' too? | 15:02 |
| jjardon | Kinnison: sure, but you should be able to do that too | 15:02 |
| Kinnison | Nope | 15:03 |
| * Kinnison does not have merge rights | 15:03 | |
| * Kinnison avoids having them because he's too nice to people and might merge stuff prematurely | 15:03 | |
| jjardon | AFAIK everyone has merge rigths | 15:04 |
| * jjardon press button | 15:04 | |
| Kinnison | :D | 15:05 |
| Kinnison | erp, everything failed | 15:05 |
| Kinnison | gcc: error trying to exec 'cc1plus': execvp: No such file or directory | 15:06 |
| Kinnison | :( | 15:06 |
| Kinnison | So the unix, fedora, and centos dockers lack a C++ compiler by the looks of things | 15:06 |
| Kinnison | This is not-good-TM | 15:06 |
| Kinnison | Can we add the C++ compilers to the dockers easily? | 15:08 |
| * Kinnison isn't familiar with how the docker images are constructed | 15:09 | |
| benschubert | let me do that | 15:09 |
| benschubert | may I push a new commit on top of your PR to add the change? | 15:09 |
| benschubert | or should I do another PR first? | 15:09 |
| Kinnison | please feel free to add to my branch | 15:10 |
| * Kinnison is grateful for you doing this, docker is still a bit of a dark art to me (though I managed to run a docker engine the other day without needing to subsequently erase my entire computer, which is a step up from the previous time I'd tried) | 15:10 | |
| WSalmon | i have updated https://gitlab.com/BuildStream/buildstream/merge_requests/1125#note_138688047, juergbi etc | 15:11 |
| benschubert | https://gitlab.com/BuildStream/buildstream-docker-images/merge_requests/101 for review :) | 15:12 |
| *** nimish has quit IRC | 15:12 | |
| Kinnison | benschubert: looks sane to me, for what I understand at least | 15:13 |
| Kinnison | amsusingly I appear to be able to hit merge for that repo | 15:13 |
| * Kinnison wonders why | 15:13 | |
| benschubert | :'D | 15:14 |
| * Kinnison closes the tab before he's tempted to | 15:14 | |
| juergbi | WSalmon: testing it I see different color for the microseconds | 15:15 |
| juergbi | is this intentional? | 15:15 |
| WSalmon | nope | 15:15 |
| juergbi | I don't see it for elapsed-us | 15:15 |
| WSalmon | what the... | 15:15 |
| juergbi | and I don't actually see it in the code... | 15:15 |
| juergbi | ah, should probably be content_profile instead of format_profile | 15:16 |
| juergbi | commented, sorry I missed it in the previous review | 15:16 |
| juergbi | feel free to directly merge it after the fix (when you've tested it yourself) | 15:17 |
| WSalmon | i have just spoted it | 15:22 |
| WSalmon | and tested in my terminal | 15:23 |
| WSalmon | how odd | 15:23 |
| *** kapil___ has joined #buildstream | 15:23 | |
| gitlab-br-bot | jonathanmaw closed issue #852 (Cover the use of WSL in CI) on buildstream https://gitlab.com/BuildStream/buildstream/issues/852 | 15:29 |
| gitlab-br-bot | jonathanmaw merged MR !1108 (jonathan/wsl-tests->master: Add pre-merge tests that use a WSL runner) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1108 | 15:29 |
| tpollard | nice! | 15:37 |
| *** sambishop has quit IRC | 15:41 | |
| *** tristan has quit IRC | 15:49 | |
| *** sambishop has joined #buildstream | 15:49 | |
| *** raoul_ is now known as raoul | 15:50 | |
| raoul | juergbi, I've realised with the patch plugin requiring previous sources to be staged, I'm not sure where during a pipeline it makes sense to stage sources into a local CAS. For this plugin staging only makes sense when an elements sources are staged, and really we don't want to be creating exceptions for the patch plugin. | 15:51 |
| raoul | Maybe it would make more sense to stage sources for an entire element, but that doesn't seem ideal | 15:51 |
| Kinnison | raoul: does this imply you can only stage sources as a whole for an element, rather than individually for each source? | 15:52 |
| Kinnison | snap | 15:52 |
| raoul | haha | 15:52 |
| Kinnison | How about if sources carry an indicator of whether they can stage independently, or layer on top? | 15:52 |
| Kinnison | then you can individually cache independent sources, and deal with the exceptions (which for now might only be patch, but it leaves it open for later source plugins too) | 15:52 |
| juergbi | oh, good point | 15:52 |
| juergbi | or maybe separate methods instead of a flag | 15:53 |
| raoul | hmm could have something along the lines of the BST_REQUIRE_PREVIOUS_SOURCE_FETCH, but for staging | 15:53 |
| juergbi | hm, no, that wouldn't work for cargo and co. | 15:53 |
| juergbi | what if we internally always combine the key of the previous source(s) with the current source? | 15:55 |
| *** tristan has joined #buildstream | 15:56 | |
| juergbi | we could make it a flag, if needed, but as a start we could always do this | 15:56 |
| raoul | what do you mean by combine the key? | 15:56 |
| juergbi | e.g., the sourcecache key for the patch source would be a combination of the unique key for the underlying tar/git source and the unique key of the patch source itself | 15:57 |
| juergbi | and similar if there are multiple underlying sources | 15:58 |
| juergbi | and when we move away from real temp. directories, we'd pass in the (CAS)Directory of the underlying source | 15:58 |
| juergbi | it should probably really be a flag long-term, though | 15:59 |
| juergbi | but for a flag we need to discuss exact semantics for the yaml source list | 15:59 |
| juergbi | i.e., what happens if two sources are staged at the same location and the second source does not set that flag | 15:59 |
| juergbi | should probably result in merging the trees | 16:00 |
| raoul | So all sources for an element would be staged on top of each other, rather than individually? | 16:00 |
| raoul | yeah we'd definitely need a think about what that flag would do | 16:00 |
| juergbi | yes, without a flag we would always stage them on top of each other | 16:02 |
| juergbi | if we introduce a flag, we'd only do that for those with the flag | 16:03 |
| raoul | makes sense, I think in this case it probably makes sense to create source cache refs when an element stages its sources? | 16:05 |
| *** ChanServ sets mode: +o tristan | 16:05 | |
| tristan | juergbi, only catching the tail end of this, but don't we already have this flag ? | 16:05 |
| juergbi | we only have it for tracking | 16:06 |
| tristan | Ah only for track and fetch ! | 16:06 |
| tristan | right | 16:06 |
| tristan | I see that makes sense | 16:06 |
| *** sambishop_ has joined #buildstream | 16:09 | |
| *** sambishop has quit IRC | 16:11 | |
| *** bochecha has quit IRC | 16:25 | |
| *** sambishop_ has quit IRC | 16:38 | |
| *** tristan has quit IRC | 17:08 | |
| *** jonathanmaw has quit IRC | 17:15 | |
| *** jonathanmaw has joined #buildstream | 17:15 | |
| *** kapil___ has quit IRC | 17:33 | |
| *** lachlan has quit IRC | 17:36 | |
| *** lachlan has joined #buildstream | 17:53 | |
| *** lachlan has quit IRC | 17:57 | |
| *** jonathanmaw has quit IRC | 18:00 | |
| *** toscalix has quit IRC | 18:11 | |
| *** lachlan has joined #buildstream | 18:14 | |
| gitlab-br-bot | BenjaminSchubert approved MR !1133 (danielsilverstone-ct/other-roaring->master: loadelement.py: Switch from roaringbitmap to pyroaring) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1133 | 18:25 |
| *** raoul has quit IRC | 18:27 | |
| gitlab-br-bot | BenjaminSchubert closed issue #899 (Possible license incompatibility; roaringbitmap is GPL while BuildStream is LGPL) on buildstream https://gitlab.com/BuildStream/buildstream/issues/899 | 19:34 |
| gitlab-br-bot | BenjaminSchubert merged MR !1133 (danielsilverstone-ct/other-roaring->master: loadelement.py: Switch from roaringbitmap to pyroaring) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1133 | 19:34 |
| *** slaf has quit IRC | 20:09 | |
| *** lachlan has quit IRC | 20:19 | |
| *** slaf has joined #buildstream | 20:23 | |
| *** slaf has joined #buildstream | 20:23 | |
| *** kapil___ has joined #buildstream | 20:28 | |
| *** benschubert has quit IRC | 21:10 | |
| *** slaf has quit IRC | 22:05 | |
| *** slaf has joined #buildstream | 22:06 | |
| *** slaf has joined #buildstream | 22:06 | |
| *** slaf has joined #buildstream | 22:06 | |
| *** slaf has joined #buildstream | 22:07 | |
| *** slaf has joined #buildstream | 22:07 | |
| *** slaf has joined #buildstream | 22:07 | |
| *** slaf has joined #buildstream | 22:07 | |
| *** slaf has joined #buildstream | 22:08 | |
| *** slaf has joined #buildstream | 22:08 | |
| *** slaf has joined #buildstream | 22:08 | |
| *** skullman has quit IRC | 22:24 | |
| *** skullman has joined #buildstream | 22:30 | |
| *** alatiera has quit IRC | 22:44 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!