IRC logs for #buildstream for Tuesday, 2019-12-03

« Monday, 2019-12-02 Index Wednesday, 2019-12-04 »
*** m_22[m] has joined #buildstream01:20
*** abderrahim[m] has joined #buildstream01:44
*** waltervargas[m] has joined #buildstream01:46
*** dbuch has joined #buildstream02:07
*** reuben640[m] has joined #buildstream02:16
*** Trevinho[m] has joined #buildstream02:24
*** ssssam[m] has joined #buildstream02:44
*** tchaik[m] has joined #buildstream02:49
*** doras has joined #buildstream03:12
*** mattiasb has joined #buildstream03:13
*** kailueke[m] has joined #buildstream03:16
*** verdre[m] has joined #buildstream03:21
*** nielsdg has joined #buildstream03:21
*** traveltissues has joined #buildstream07:39
*** Kinnison has quit IRC08:30
*** Kinnison has joined #buildstream08:30
*** Kinnison has quit IRC08:41
*** Kinnison has joined #buildstream08:42
*** rdale has joined #buildstream09:03
*** tiagogomes has joined #buildstream09:28
*** santi has joined #buildstream09:35
*** ChanServ sets mode: +o ironfoot09:43
*** tme5 has joined #buildstream10:04
*** jonathanmaw has joined #buildstream10:16
*** toscalix has joined #buildstream10:17
gitlab-br-botjuergbi opened issue #1226 (Add CI job for buildbox-run-bubblewrap with buildbox-fuse) on buildstream https://gitlab.com/BuildStream/buildstream/issues/122610:39
tpollarddo we have any plans to tackle https://gitlab.com/BuildStream/buildstream/issues/336 ?10:41
benschubertas long as we spawn subprocess for every job, we will hit it10:43
benschubertAnd I've seen it up to at lease 2500 times in 30 seconds :'D10:44
gitlab-br-bottlater opened (was WIP) MR !1645 (tlater/artifactserver-casd->master: Refactor casserver.py: Stop relying on the buildstream-internal `CASCache` implementation) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/164510:45
benschuberttlater[m]: ^ the buildbox bug is fixed already?10:46
tlater[m]benschubert: No, but it's unrelated to my branch10:46
benschubertah fair :)10:46
tlater[m]https://gitlab.com/BuildStream/buildstream/merge_requests/1645#note_25341700510:46
tlater[m]I.e., same thing affecting every other MR - I just happened to be the first to stumble upon it10:46
benschubertSo you removed the 'fix' right?10:47
tlater[m]yeah :)10:47
benschubertawesome :)10:49
tlater[m]I wonder why the bandid-sast thing isn't updating10:52
tlater[m]My MR from yesterday fixed all but 2 of the "high" things10:52
tlater[m]They're still there after 3 or so pipelines though; despite the code they're pointing to having changed completely.10:52
benschubertx')10:55
*** cs-shadow has joined #buildstream10:56
*** santi has quit IRC11:00
*** santi has joined #buildstream11:02
gitlab-br-bottlater closed issue #1167 (Make the bst-artifact-server proxy CAS storage calls to buildbox-casd) on buildstream https://gitlab.com/BuildStream/buildstream/issues/116711:38
gitlab-br-bottlater merged MR !1645 (tlater/artifactserver-casd->master: Refactor casserver.py: Stop relying on the buildstream-internal `CASCache` implementation) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/164511:38
tlater[m]\o/11:39
*** phoenix has joined #buildstream11:47
*** phildawson_ has joined #buildstream11:54
*** phildawson has quit IRC11:55
juergbithe list of blockers is shrinking \o/12:18
tme5can someone give me a summary of the work to move plugins around?12:27
tme5it seems like you're moving all the plugins out of the main repo?12:27
juergbitme5: this should provide some context: https://mail.gnome.org/archives/buildstream-list/2019-February/msg00067.html12:30
tme5sounds like a lot of work12:37
*** phildawson_ has quit IRC13:08
gitlab-br-botjuergbi merged MR !1737 (juerg/casd-separate-user->master: Support buildbox-casd running as separate user) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/173713:10
*** santi has quit IRC13:20
*** phoenix_ has joined #buildstream13:21
*** phoenix has quit IRC13:22
*** phoenix_ is now known as phoenix13:23
*** santi has joined #buildstream14:03
*** phildawson_ has joined #buildstream14:08
*** narispo has quit IRC14:20
*** narispo has joined #buildstream14:20
*** coldtom8 is now known as coldtom15:08
coldtomis there a way i can get `bst artifact push` to push artifacts from cross-junction dependencies?15:08
tpollardcoldtom: so you want to push junction elements to your parent remote?15:13
tpollardcoldtom: if so, in your junction element set 'cache-junction-elements": True' & push with --deps all?15:14
WSalmonyep, I did the --deps all trick for populating the `local` cache for elce15:15
tpollardby default master won't push junction elements to your parent15:15
tpollardbut setting that bool should change that15:15
tpollardthat's a breaking change since 1.4, iirc15:16
coldtomthat did it, ta tpollard15:17
*** phoenix has quit IRC15:51
tme5has anyone considered merging git and git_tag ?15:57
coldtomtme5: git_tag exists because upstream did not want that functionality, i assume this has not changed15:58
coldtomthe git plugin has been abstracted, and the intention is to make git_tag inherit from it to reduce duplication15:59
coldtomhowever, i wanted to wait until bst2 was close to release, as otherwise it'll be too much of a moving target15:59
tme5who is upstream?16:02
KinnisonBuildStream16:02
Kinnisongit_tag is an fdsdk thing IIRC16:02
tlater[m](us!)16:02
tlater[m]coldtom: We used to not be able to inherit from other plugins, has that changed?16:03
* tlater[m] still thinks we should be able to16:03
* Kinnison thinks that most if not all of the git plugin's behaviour is in a superclass now16:04
coldtomtlater[m]: i believe there is now an abstract git source that can be inherited from?16:04
KinnisonI don't think you can inherit from plugins per-se, but that's usable16:04
tlater[m]Ah16:04
* tlater[m] still wishes there was inter-plugin inheritance16:04
tme5but if you're moving all your plugins out, won't git and git_tag end up in the same place?16:05
tlater[m]It would make a lot of these use cases a lot simpler16:05
tlater[m]tme5: git_tag is a local plugin in fdsdk16:05
tme5coldtom, there is but it's not exactly useful right now as far as I can see16:05
*** traveltissues has quit IRC16:05
tme5tlater[m], so, where are you moving the plugins to?16:06
tme5bst-external or bst_plugins_experimental?16:06
tlater[m]tme5: bst-plugins-experimental, iirc16:06
tlater[m]To become bst-plugins16:06
tme5what is this then? https://gitlab.com/BuildStream/bst-plugins-experimental/blob/master/src/bst_plugins_experimental/sources/git_tag.py16:06
coldtomthat is the git_tag plugin, it is hosted in bst-plugins-experimental, not fdsdk16:07
coldtomwe wanted it to be easily shared16:07
coldtomso put it in bst-external16:07
tlater[m]Huh, in that case upstream has accepted git-tag16:07
tme5you mean plugins_experimental?16:07
tlater[m]Pre-2.0 it was in fdsdk16:07
coldtomnope, it's always been in bst-external16:07
coldtombst-plugins-experimental is the master version of bst-external, which was made separate for some reason16:08
tlater[m]It has? Interesting, well, in either case, it's simply part of the plugins that buildstream lives with then.16:08
tme5hahahah16:08
tlater[m]coldtom: That's not *entirely* correct16:08
tlater[m]bst-plugins-experimental is the "breeding ground" for 2.0 plugins16:08
tlater[m]2.0 plugins must live in a different repository, since they are incompatible with 1.0 plugins16:09
tlater[m]bst-plugins-experimental will be distilled off into further repositories by the time 2.0 hits16:09
tme5ok, so, returning... has anyone considered merging git and git_tag ? I'm doing a bit of work on the git plugin now so I could give this a go16:09
coldtomthe plan was to use the abstracted git source to basically eliminate the diff16:10
tlater[m]tme5: I'd suggest reading into the discussion that created git_tag, and deciding whether their reasoning means you should not merge them, and then send a ML post.16:10
* tlater[m] suspects the two will end up in the same repo temporarily, and then be split off when we create `bst-plugins` and `bst-plugins-external` or somesuch16:10
coldtomthat would be my expectation16:12
tlater[m]Ergo, as coldtom says, make it depend on the base class, and expect the plugins to remain separate.16:12
tlater[m]Unless you think they should be merged - in which case, pick back up the discussion.16:13
coldtomtme5: here's the original issue from which git_tag was birthed https://gitlab.com/BuildStream/buildstream/issues/22016:13
tme5I do think that, i'll have a look at the mailing list16:13
*** phildawson has joined #buildstream16:29
gitlab-br-botjuergbi opened (was WIP) MR !1738 (juerg/buildbox-run->master: Add buildbox-run sandboxing backend) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/173816:29
*** phildawson_ has quit IRC16:29
tlater[m]juergbi: ta for that16:32
* tlater[m] wonders how that docker image changed16:32
tlater[m]Oh, you just uploaded your own?16:32
juergbiwhat do you mean?16:33
juergbithe docker image was built by CI16:33
juergbiit's in the juerg/userchroot branch (was based on your branch)16:33
tlater[m]Ah16:33
juergbiI should update/replace the MR to avoid confusion16:33
juergbitlater[m]: I've changed the structure a bit such that it sould be easy to add buildbox-run-bubblewrap and buildbox-fuse16:34
tlater[m]Yeah, that looks right16:35
* tlater[m] missed the review comments on that MR D:16:36
*** santi has quit IRC16:44
*** tpollard has quit IRC17:03
*** santi has joined #buildstream17:05
*** slaf has quit IRC17:29
*** slaf_ has joined #buildstream17:29
*** slaf has joined #buildstream17:29
*** slaf has joined #buildstream17:29
*** slaf has joined #buildstream17:29
*** slaf has joined #buildstream17:30
*** slaf has joined #buildstream17:30
*** slaf has joined #buildstream17:30
*** jonathanmaw has quit IRC17:33
*** slaf_ has joined #buildstream17:34
*** slaf_ has joined #buildstream17:34
*** slaf_ has joined #buildstream17:35
*** slaf has quit IRC17:35
*** slaf has joined #buildstream17:35
*** slaf has quit IRC17:35
*** slaf has joined #buildstream17:35
*** slaf has joined #buildstream17:35
*** slaf has joined #buildstream17:36
*** slaf has joined #buildstream17:36
*** slaf has joined #buildstream17:36
*** slaf has joined #buildstream17:36
*** slaf has joined #buildstream17:37
*** slaf has joined #buildstream17:37
*** slaf has joined #buildstream17:37
*** slaf has joined #buildstream17:38
*** slaf has joined #buildstream17:38
*** slaf has joined #buildstream17:38
*** slaf has joined #buildstream17:39
*** slaf has joined #buildstream17:39
*** slaf has joined #buildstream17:39
*** slaf has joined #buildstream17:39
*** slaf has joined #buildstream17:40
*** slaf has joined #buildstream17:40
*** slaf has joined #buildstream17:40
*** slaf has joined #buildstream17:40
*** tme5 has quit IRC17:43
*** toscalix has quit IRC17:50
*** santi has quit IRC18:16
coldtomwhat are people's thoughts on changing the default quota for bst-artifact-server? 10G seems unreasonably small for a server to me18:35
*** rdale has quit IRC18:40
juergbicoldtom: I thought we didn't have a default quota. where do you see that default?18:42
coldtomjuergbi: https://docs.buildstream.build/master/using_configuring_cache_server.html#cmdoption-bst-artifact-server-quota18:48
juergbicoldtom: right. forgot about the default on the click level. I'd be happy to completely remove the quota18:51
juergbiwe don't have a default quota on the client side and we do trigger expiry also without quota if we're running out of disk space18:52
coldtomjuergbi: that sounds _much_ better18:54
* coldtom has forgotten to change from 10G several times and wondered why artifacts have been vanishing18:54
juergbiour initial expiry/cleanup code on the server only worked in terms of free disk space, not in terms of disk usage/quota18:56
juergbithere we cleaned 10GB of data when we hit the 2GB limit18:56
juergbiand somehow that remained as default value for the new quota option18:57
juergbilooks like this was my fault and I don't think that was intentional18:57
*** phildawson has quit IRC19:01
*** cs-shadow has quit IRC22:12
« Monday, 2019-12-02 Index Wednesday, 2019-12-04 »

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