| *** tristan has quit IRC | 05:40 | |
| *** tristan has joined #buildstream | 06:02 | |
| *** Kinnison has joined #buildstream | 06:52 | |
| *** tristan has quit IRC | 07:41 | |
| *** bochecha has joined #buildstream | 08:04 | |
| *** tristan has joined #buildstream | 08:06 | |
| *** pointswaves has joined #buildstream | 08:08 | |
| *** rdale has joined #buildstream | 08:10 | |
| *** phil has joined #buildstream | 08:11 | |
| *** pointswaves has quit IRC | 08:17 | |
| *** raoul has joined #buildstream | 08:48 | |
| gitlab-br-bot | tpollard approved MR !1371 (bschubert/cythonize-valid-char-names->master: _loader/loader: cythonize valid_chars_name) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1371 | 08:54 |
|---|---|---|
| benschubert | jennis: you are a maintainer for bst-plugins-experimental, correct? https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/12 :) | 09:30 |
| jennis | Not that I'm aware of, I think the MAINTAINERS file was just copied from bst-external, but I'll have a look nonetheless | 09:33 |
| benschubert | ah, do you know who's responsible? | 09:33 |
| jennis | IIRC, jjardon created the repo | 09:36 |
| jennis | but this might have just been him actioning a stalling process | 09:36 |
| *** jonathanmaw has joined #buildstream | 09:36 | |
| jennis | I think I have merge rights, so I'll review | 09:37 |
| *** lantw44 has quit IRC | 09:37 | |
| *** lachlan has joined #buildstream | 09:51 | |
| jennis | benschubert, do you mind if I push a commit to your branch, there is a typo in the ostree documentation | 09:56 |
| jennis | May aswell bundle it in with this MR | 09:57 |
| benschubert | sure | 09:57 |
| benschubert | go for it | 09:57 |
| benschubert | I'll have another PR in 2 minutes for more changes | 09:57 |
| benschubert | and then a bigger one a bit later on | 09:57 |
| jennis | Cool | 09:58 |
| *** lachlan has quit IRC | 09:58 | |
| jennis | ahh benschubert, turns out the typo is referring to an invalid cross reference to somewhere in BuildStream's core docs | 10:05 |
| jennis | <benschubert> Do we have a good way for linking to the BuildStream doc from a plugin's docs? | 10:05 |
| jennis | ^ Is one of your MR's for this? Or did you find a way to do this? | 10:06 |
| benschubert | jennis: then wait for my new MR ;) | 10:06 |
| benschubert | it's coming | 10:06 |
| jennis | haha, good timing | 10:06 |
| jennis | Let me merge this one | 10:06 |
| jennis | oh... in fact I don't have write access :@ | 10:07 |
| benschubert | phil: you have access to bst-plugins-experimental, correct? | 10:08 |
| jennis | benschubert, done | 10:10 |
| benschubert | jennis: https://gitlab.com/BuildStream/bst-plugins-experimental/merge_requests/13 is with the additional changes | 10:10 |
| phil | benschubert, I do. I've given jennis access. | 10:11 |
| *** bilelmoussaoui has quit IRC | 10:11 | |
| benschubert | thanks a lot! | 10:11 |
| *** lachlan has joined #buildstream | 10:11 | |
| benschubert | I'll be moving the ostree examples to plugiins-experimental too | 10:11 |
| benschubert | in order to cleanup buildstream master | 10:11 |
| phil | Cool. Feel free to poke me for reviews when you want them :) | 10:13 |
| *** ChanServ sets mode: +o tristan | 10:14 | |
| tristan | Wasn't the plan to move all the plugins to a single repository (bst-plugins-experimental I presume being the logical choice) as a first step of moving plugins out of BuildStream ? | 10:14 |
| benschubert | tristan: I think so yes. I'm only moving examples using ostree that were kept in buildstream when moving ostree to plugins-experimental | 10:14 |
| benschubert | I think it makes little sense to keep the examples for flatpak with ostree in the master, if you need all the plugins from experimental for it | 10:15 |
| tristan | Yeah, I'm just whining again that ostree (like one of our most important plugins) got moved out in advance | 10:15 |
| jennis | benschubert, some of the documentation examples in core use the ostree plugin to build, I'm not sure what we should be doing about this | 10:15 |
| benschubert | jennis: i'm moving them to bst-plugins-experimental | 10:16 |
| tristan | And was sort of hoping that we'd acknowledge that moving ostree outside was just shifting the same problem from one place to another | 10:16 |
| tristan | but meh | 10:16 |
| jennis | benschubert, we can't move BuildStream's example documentation to bst-plugins-experimental | 10:16 |
| benschubert | tristan: that is definitely correct, however, ostree was bringing lots of host dependencies not necessarily available on more legacy systems | 10:16 |
| benschubert | jennis: only the ostree-using part, which is two examples | 10:16 |
| jennis | https://buildstream.gitlab.io/buildstream/examples/flatpak-autotools.html | 10:16 |
| tristan | benschubert, Right, and that was rather the point: Moving all the plugins to bst-plugins-experimental puts them in a place where ostree also is | 10:17 |
| cs-shadow | tristan: I seem to remember the discussion differently. I thought we were going to move plugins to domain-specific repos. I am not sure what it accomplishes to move all plugins if we are again to lump them in the same place | 10:17 |
| tristan | proving that we're not getting rid of the problem anyway, and that anyway, the install of ostree is an optional one (it's not a hard dependency and never was) | 10:17 |
| tristan | cs-shadow, The conversation was long and painful, find the summary... ok I'll find it | 10:17 |
| benschubert | tristan: "make doc" fails on my system because I don't have ostree, that's not super friendly :) | 10:18 |
| tristan | cs-shadow, this: https://mail.gnome.org/archives/buildstream-list/2019-April/msg00082.html ... and then there is another final summary to that | 10:19 |
| tristan | cs-shadow, was sander who suggested we move everything in one block, and then later go domain specific | 10:20 |
| tristan | benschubert, building the docs shouldnt require plugins, only if you are regenerating the bst outputs, which I believe needs to be explicit (with some env var) | 10:21 |
| cs-shadow | yeah, i mean incremental progress is fine. I only mean that it's not the endgame to have all plugins end up in one repo | 10:21 |
| tristan | that was very intentional that nobody should need to download a runtime just to test some changes to a docs contribution | 10:21 |
| tristan | cs-shadow, You know my stance on that - they need to be all in the same repo if they they are going to installation bound instead of project bound, and if they are project bound, we need a better way to get it done than git submodules | 10:22 |
| tristan | but whatever, not happy with this plugins story one bit, I'll stop pouting and keep my discontent to myself | 10:23 |
| * cs-shadow isn't supper happy about it either but isn't looking forward to more lengthy ML threads :) | 10:24 | |
| tristan | yeah, better wait a couple years and then think about creating something that is user ready after all the dev dies down | 10:28 |
| benschubert | tristan: I was trying to force the rebuild yes | 10:28 |
| benschubert | If I interpret this correctly, you'd rather not have this moved to bst-experimental? | 10:28 |
| tristan | benschubert, Originally "If you had BuildStream, then you had all the upstream core plugins, period; if you are missing a host dependency for a project, it will kindly tell you in preflight()"... and if you want to use any *other* plugins, then they should be project specific, shared amongst projects, and always safe to clash with other projects you junction depend on because plugins are namespaced by project | 10:30 |
| tristan | benschubert, Now - people started shipping plugins with pip, can of worms starts opening, now people are sharing unstable plugin collections that are installation bound, rather than project bound as they were intended | 10:31 |
| tristan | Now we wanna split everything out, plot thickens | 10:31 |
| tristan | Not really fun | 10:31 |
| * Kinnison was never particularly happy with the idea of pip sourced plugins | 10:32 | |
| benschubert | tristan: Ok, let me backtrack a bit. The reason why I wanted to move the ostree examples to bst-plugins-experimental, is because we have some HAVE_OSTREE in a tests/testutils/sites.py. I wanted to get rid of this tests/testutils/sites.py since we already have buildstream/testing/site.py. For me the logical conslucion is to move the ostree stuff out since we moved the plugin away. I can hide it in BuildStream if that's what you want | 10:33 |
| * cs-shadow humbly disagrees but doubts we'll resolve it here and now | 10:33 | |
| persia | cs-shadow: From curiosity, with which bit do you disagree? | 10:34 |
| benschubert | tristan: So, would you rather have the sites.py stuff concerning ostree be moved to the ostree example so everything is at the same place, or that we move the ostree-using documentation to where ostree plugin lives? | 10:35 |
| cs-shadow | pip: couple of things but 1. that pip has somehow caused people to share unstable plugins. `pip` is just a delivery mechanism, everyone was free to share whatever plugins they wanted anyway | 10:35 |
| cs-shadow | and also that external plugins have to project bound. If core plugins aren't, I am not sure why external plugins have to be | 10:36 |
| cs-shadow | persia: ^ | 10:36 |
| persia | cs-shadow: Thanks: that makes sense. I just wasn't sure if you disagreed with the plugin model or with moving the examples from first gloss :) | 10:38 |
| cs-shadow | persia: oh sorry, I could've been clearer | 10:38 |
| persia | No worries :) | 10:39 |
| tristan | cs-shadow, Before pip origin there was no opportunity for project authors to share plugins that are (incorrectly) installation bound instead of project bound | 10:41 |
| tristan | cs-shadow, that was just a mistake we made, our reasoning being that "If buildstream installs it's own plugins, shouldnt it be possible to install other system wide packages alongside BuildStream" ? Yeah, but we got it wrong | 10:41 |
| * Kinnison advocates removing that plugin source for bst2 | 10:43 | |
| tristan | benschubert, I'm not sure honestly about that - do we want to fragment the tutorial documentation just because the plugins are fragmenting ? Not sure that is a great idea | 10:43 |
| cs-shadow | tristan: of course shipping all plugins ourselves will be the simplest but I think the power of buildstream comes from being extensible, not from having an all-encompassing set of core plugins | 10:44 |
| tristan | benschubert, Anyway sorry if I am a bit bitter, it is mostly about the knee-jerk nuking ostree plugin itself from BuildStream and treating it differently than everything else which was bothering me | 10:44 |
| benschubert | tristan: agreed. So isolating it inside the ostree example would be ok with you? | 10:44 |
| tristan | cs-shadow, There are two sides of this: Extensible is important, but by design this extensibility is on a per-project basis: The project itself decides exactly which plugin to use; it should never be influenced by what version of what package someone decided to install *beside* buildstream | 10:45 |
| tristan | cs-shadow, the other side of it is providing a useful base set of plugins which we can control and really guarantee API stability on | 10:46 |
| tristan | benschubert, Don't we need the ostree plugin for many tutorials ? Some of them use the alpine tarball, then there is the flatpak example also which needs ostree | 10:46 |
| benschubert | tristan: from what I saw only the autotools-flatpak requires it | 10:47 |
| benschubert | hence why I thought moving it was the best | 10:47 |
| tristan | benschubert, anyway I won't object to moving things around, but someone needs to come up with a plan on how to index this material conveniently :-/ | 10:47 |
| benschubert | Agreed | 10:47 |
| benschubert | I'll keep it in for now | 10:47 |
| benschubert | isolating it more | 10:47 |
| jennis | benschubert, tristan, tlater[m] is working on replacing this alpine tarball import with freedesktop, which will need ostree | 10:48 |
| jennis | Just a heads up | 10:48 |
| jennis | !1300 | 10:48 |
| gitlab-br-bot | MR !1300: WIP: Use freedesktop-sdk as a base for tutorials as much as possible https://gitlab.com/BuildStream/buildstream/merge_requests/1300 | 10:48 |
| benschubert | jennis: ah ok, I'll keep that in then | 10:48 |
| tristan | That is also true, but it could technically not use ostree if a tarball were generated instead | 10:48 |
| tristan | It would have to be hosted, but the point of that MR is to prove some self hosting here - it makes sense to make examples which run from a runtime that is created with BuildStream | 10:49 |
| jennis | tristan, I had a brief conversation with tlater about this MR, I think he wanted to implement this change because the tutorial will show the power/benefits of buildstream, i.e. here are all my components that I'm depending on, not just, let's unpack this tarball which is hosted somewhere random and build on top of it | 10:52 |
| jennis | therefore I'm not sure if a freedesktop tarball is the way he'd want to go, but I don't wanna speak for him :p | 10:52 |
| tristan | jennis, I think that whether it comes in the form of a tarball or not is irrelevant; it is the fact that it was generated with BuildStream that is interesting | 11:01 |
| tpollard | having the artifact command have a --remote option is something that I'd like to see | 11:05 |
| tpollard | if that's the correct place for that | 11:05 |
| benschubert | tpollard: thanks for the heads up, I need to answer to that message | 11:06 |
| tristan | tpollard, Right, I only summarized what was in the mails from last year, we didn't discuss that before | 11:07 |
| tristan | but it would be a good thing to discuss indeed - and raises the question of how that factors in with the same option for `pull` and `push` | 11:07 |
| tpollard | yep! it's good to have the summary | 11:07 |
| tpollard | build also has the --remote constraint too, but obviously that's not under the artifact subgroup | 11:08 |
| tristan | right I guess anything that can pull/push currently can have a remote specified on the CLI - maybe that should also now include even commands like workspaces or shell | 11:09 |
| tpollard | also have the artifacts be proto based, gives us the possibility to encode more information into that which could be beneficial to show/list-content etc | 11:09 |
| tpollard | *having | 11:10 |
| tristan | since we (A) now have SourceCache ... oh maybe that doesnt apply to workspaces anyway... and (B) ... shells can result in pulling a buildtree | 11:10 |
| tpollard | I was thinking maybe the size is a candidate to be included in the proto, which would be quicker to query | 11:10 |
| tristan | In general... the list-content idea has a similar problem to sources... and the fact that we don't have something to show source urls is already kind of problematic | 11:11 |
| tristan | the similarity of the problem is that it is a list member of an element/artifact (and so doesnt work very well with `bst show` format strings) | 11:11 |
| tristan | would be nice if someone had a genius idea of how to express what we want to see from a list of things in an element from a sort of bst show command | 11:12 |
| tristan | I haven't thought of anything particularly elegant for that | 11:12 |
| tristan | Another case of "Shucks we have no way to observe sources": https://gitlab.com/BuildStream/bst-plugins-experimental/issues/2 | 11:13 |
| tristan | (and honestly, it would also be practical to be able to say "Show me the sources which went into this artifact", as much for an artifact as for an element) | 11:14 |
| tpollard | I agree | 11:14 |
| *** bochecha_ has joined #buildstream | 11:22 | |
| *** bochecha has quit IRC | 11:24 | |
| *** bochecha_ is now known as bochecha | 11:24 | |
| *** bilelmoussaoui has joined #buildstream | 11:35 | |
| *** bochecha_ has joined #buildstream | 11:37 | |
| *** bilelmoussaoui has quit IRC | 11:38 | |
| *** bochecha has quit IRC | 11:40 | |
| *** bochecha_ is now known as bochecha | 11:40 | |
| *** slaf has quit IRC | 11:45 | |
| tristan | jennis, So you are asking specifically about storing _depth on elements ? or the patch in general | 11:53 |
| tristan | I just went for a walk through the comments and the MR certainly looks promising (!1344), and it seems juergbi and benschubert have been closely reviewing this | 11:53 |
| jennis | The former, purely because it's probably not inherent to Element, but we can't think of a better way to do it and it's not that bad | 11:54 |
| tristan | Well, that part seems to be quite fine to me, it makes sense to cache some things to make graph traversals more performant | 11:54 |
| jennis | Cool, I'm glad you said that | 11:55 |
| tristan | I'm very interested to see performance impacts of eliminating calls to Element.dependencies() in cache key calculations (i.e. accumulative cache key building) | 11:55 |
| tristan | I guess that hasn't been tackled yet :) | 11:55 |
| jennis | Yeah me too, this will happen once the concept of update sate is (eventually) removed | 11:58 |
| jennis | Ok, I guess that once benschubert's comment about the RequiredQueue is resolved, we can think about marging this :) | 11:58 |
| tristan | jennis, Just added a few comments there, mostly about lack of comments | 12:07 |
| *** slaf has joined #buildstream | 12:12 | |
| *** slaf has joined #buildstream | 12:12 | |
| *** lachlan has quit IRC | 12:16 | |
| *** slaf has quit IRC | 13:08 | |
| jennis | hah, ok thanks tristan | 13:10 |
| *** tpollard has quit IRC | 13:24 | |
| *** tpollard has joined #buildstream | 13:25 | |
| *** tristan has quit IRC | 13:37 | |
| *** slaf has joined #buildstream | 13:38 | |
| *** slaf has joined #buildstream | 13:38 | |
| *** slaf has joined #buildstream | 13:38 | |
| *** slaf has joined #buildstream | 13:38 | |
| *** slaf has joined #buildstream | 13:39 | |
| *** slaf has joined #buildstream | 13:39 | |
| *** slaf has joined #buildstream | 13:39 | |
| *** slaf has joined #buildstream | 13:40 | |
| *** lachlan has joined #buildstream | 13:48 | |
| gitlab-br-bot | aevri opened MR !1373 (aevri/defensive_send_message->master: _scheduler/jobs: refactor, defensive send_message) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1373 | 13:49 |
| benschubert | could someone wake marge up please? :) | 14:08 |
| * jennis will see what he can do | 14:09 | |
| jennis | She should be restarted... | 14:13 |
| benschubert | thanks! | 14:13 |
| *** tristan has joined #buildstream | 14:19 | |
| *** lantw44 has joined #buildstream | 14:25 | |
| gitlab-br-bot | BenjaminSchubert opened MR !1375 (bschubert/site-consolidation->master: Remove tests/testutils/site.py and move everything to buildstream/testing/_utils/site.py) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1375 | 14:25 |
| gitlab-br-bot | marge-bot123 merged MR !1334 (aevri/split_jobs_parent_child->master: Split ChildJob out from Job class) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1334 | 14:40 |
| benschubert | tristan: !1375 is my cleanup of the _site_ business if you are interested | 14:41 |
| *** bilelmoussaoui has joined #buildstream | 14:44 | |
| *** bilelmoussaoui has quit IRC | 14:53 | |
| *** lachlan has quit IRC | 14:54 | |
| gitlab-br-bot | aevri opened (was WIP) MR !1374 (aevri/spawk->master: Rename (spawn, fork) -> 'start process') on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1374 | 14:57 |
| gitlab-br-bot | aevri opened (was WIP) MR !1373 (aevri/defensive_send_message->master: _scheduler/jobs: refactor, defensive send_message) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1373 | 14:57 |
| *** lachlan has joined #buildstream | 15:00 | |
| gitlab-br-bot | tpollard approved MR !1374 (aevri/spawk->master: Rename (spawn, fork) -> 'start process') on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1374 | 15:03 |
| gitlab-br-bot | marge-bot123 closed issue #1025 (AaaP follow-up: Handling of legacy remotes & CAS/Artifact services) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1025 | 15:14 |
| gitlab-br-bot | marge-bot123 merged MR !1366 (raoul/1025-legacy-remotes->master: Improved handling of legacy remotes) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1366 | 15:14 |
| tpollard | woo \o/ | 15:14 |
| * tpollard uncomments legacy remotes from his user conf | 15:16 | |
| *** bochecha has quit IRC | 15:30 | |
| *** phildawson_ has joined #buildstream | 15:34 | |
| *** phil has quit IRC | 15:35 | |
| gitlab-br-bot | marge-bot123 merged MR !1371 (bschubert/cythonize-valid-char-names->master: _loader/loader: cythonize valid_chars_name) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1371 | 16:13 |
| *** bochecha has joined #buildstream | 16:15 | |
| *** xjuan has joined #buildstream | 16:19 | |
| *** phildawson_ has quit IRC | 16:30 | |
| *** shibu has joined #buildstream | 16:48 | |
| *** paulsherwood has quit IRC | 17:04 | |
| *** paulsherwood has joined #buildstream | 17:05 | |
| *** ikerperez has quit IRC | 17:05 | |
| *** adds68 has quit IRC | 17:05 | |
| gitlab-br-bot | BenjaminSchubert opened MR !1376 (bschubert/pylint-integration->master: Ensure pylint runs in tests/integration) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1376 | 17:12 |
| *** raoul has quit IRC | 17:12 | |
| gitlab-br-bot | BenjaminSchubert opened MR !1377 (bschubert/pylint-artifactcache->master: Ensure pylint runs in tests/artifactcache) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1377 | 17:45 |
| *** shibu has quit IRC | 17:45 | |
| *** shibu has joined #buildstream | 17:46 | |
| *** jonathanmaw has quit IRC | 17:56 | |
| *** CTtpollard has joined #buildstream | 17:57 | |
| *** tpollard has quit IRC | 17:59 | |
| gitlab-br-bot | BenjaminSchubert opened MR !1378 (bschubert/pylint-fixes->master: Ensure pylint runs in some tests paths) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1378 | 18:01 |
| *** slaf has quit IRC | 18:06 | |
| *** bochecha has quit IRC | 18:07 | |
| *** laurence has quit IRC | 18:22 | |
| *** bethw has quit IRC | 18:23 | |
| *** johnward has quit IRC | 18:23 | |
| *** jennis has quit IRC | 18:23 | |
| *** WSalmon has quit IRC | 18:23 | |
| *** valentind has quit IRC | 18:24 | |
| *** pointswaves has joined #buildstream | 18:24 | |
| *** WSalmon has joined #buildstream | 18:27 | |
| *** valentind has joined #buildstream | 18:27 | |
| *** laurence has joined #buildstream | 18:31 | |
| *** bethw has joined #buildstream | 18:31 | |
| *** johnward has joined #buildstream | 18:31 | |
| *** jennis has joined #buildstream | 18:32 | |
| *** xjuan has quit IRC | 18:54 | |
| *** shibu has quit IRC | 19:00 | |
| *** lachlan has quit IRC | 19:14 | |
| *** slaf has joined #buildstream | 20:52 | |
| gitlab-br-bot | BenjaminSchubert opened MR !1379 (bschubert/optimize-loader-types->master: Optimize _loader/types.py) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1379 | 21:46 |
| *** pointswaves has quit IRC | 22:12 | |
| *** hergertme has quit IRC | 23:39 | |
| *** hergertme has joined #buildstream | 23:40 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!