| *** xjuan has quit IRC | 00:52 | |
| gitlab-br-bot | coldtom approved MR !1549 (abderrahim/backport-947->bst-1: plugins/elements/cmake.yaml: always specify variable types) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1549 | 07:12 |
|---|---|---|
| *** rdale has joined #buildstream | 08:21 | |
| *** bochecha has joined #buildstream | 08:24 | |
| *** tme5 has joined #buildstream | 08:38 | |
| *** rdale has quit IRC | 08:53 | |
| *** rdale has joined #buildstream | 08:54 | |
| *** rdale_ct has joined #buildstream | 08:54 | |
| *** rdale has quit IRC | 08:57 | |
| *** rdale_ct has quit IRC | 08:57 | |
| *** rdale has joined #buildstream | 08:57 | |
| tlater[m] | juergbi: Regarding your comment on !1540 | 09:01 |
| gitlab-br-bot | MR !1540: WIP: Support separate end points for artifact caches https://gitlab.com/BuildStream/buildstream/merge_requests/1540 | 09:01 |
| tlater[m] | How would we find our blobs if we didn't link the servers? | 09:02 |
| juergbi | tlater[m]: we could check CAS servers one by one until we have all blobs | 09:10 |
| juergbi | typical case will anyway be single CAS server | 09:10 |
| juergbi | it's difficult to say whether this would adversely affect multi-CAS use cases as they aren't clearly defined | 09:12 |
| tlater[m] | Hm | 09:12 |
| tlater[m] | Yeah, I feel like it could be slow in the latter case | 09:12 |
| tlater[m] | But I'm also struggling to see a use case for that | 09:12 |
| juergbi | you could have a CAS server that is read-only to developers, only populated by CI | 09:14 |
| juergbi | and then a secondary CAS server for free use | 09:14 |
| juergbi | as we batch blob retrieval, I don't expect the performance to be a big issue as long as we're just talking about 2-3 CAS servers | 09:15 |
| juergbi | also, I think we should rather try to go towards the approach of 'unified CAS' from the buildstream point of view (following https://docs.google.com/document/d/17WJ4cz150IHeTgvJGxcbSGK1Zg2vsLpcgUSVffrwGfk/edit# ) | 09:16 |
| juergbi | I'm not suggesting to drop support for multiple servers, however, I don't think we should worry too much about performance of hypothetical use cases | 09:18 |
| *** rdale_ct has joined #buildstream | 09:18 | |
| tlater[m] | Alright, patch rework coming up! | 09:18 |
| tlater[m] | Haha | 09:18 |
| *** rdale_ct has quit IRC | 09:18 | |
| tlater[m] | Thanks juergbi | 09:18 |
| tlater[m] | I suppose if anyone ever comes to complain that their CAS pushes are slow we can rework then. | 09:19 |
| juergbi | tlater[m]: that said, it could be sensible to talk to fdo-sdk devs to make sure we're not forgetting real use cases | 09:19 |
| tlater[m] | juergbi: Their use case is simply using BuildStream with normal remote execution build servers. | 09:20 |
| tlater[m] | Which means they'd have two CASes at most | 09:20 |
| tlater[m] | And for that an iterative search won't be too bad. If we keep configuration order then users can even configure the cache that is typically addressed to be checked first. | 09:21 |
| juergbi | ok, maybe also reply to your ML thread about this | 09:22 |
| tlater[m] | Will do | 09:23 |
| juergbi | tristan's reply actually described both cases 'If it is [not] important to express that a given artifact server is related to a given CAS server...' | 09:23 |
| juergbi | but there was no further public discussion, afaict | 09:23 |
| *** lachlan has joined #buildstream | 09:27 | |
| *** traveltissues has joined #buildstream | 09:39 | |
| *** jonathanmaw has joined #buildstream | 09:39 | |
| *** lachlan has quit IRC | 10:21 | |
| *** lachlan has joined #buildstream | 10:40 | |
| tlater[m] | Do we have EnumNodes? | 10:45 |
| tlater[m] | Ah, there's a `get_enum()` | 10:46 |
| *** lachlan has quit IRC | 10:48 | |
| *** lachlan has joined #buildstream | 10:53 | |
| *** lachlan has quit IRC | 10:58 | |
| *** toscalix has joined #buildstream | 11:27 | |
| *** lachlan has joined #buildstream | 11:34 | |
| *** lachlan has quit IRC | 11:40 | |
| *** lachlan has joined #buildstream | 11:48 | |
| *** lachlan has quit IRC | 12:36 | |
| gitlab-br-bot | danielsilverstone-ct opened MR !1551 (danielsilverstone-ct/fix-node-provenance->master: Improve node provenance in error reporting) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1551 | 12:58 |
| gitlab-br-bot | jennis approved MR !1551 (danielsilverstone-ct/fix-node-provenance->master: Improve node provenance in error reporting) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1551 | 13:16 |
| *** lachlan has joined #buildstream | 13:36 | |
| gitlab-br-bot | marge-bot123 closed issue #1059 (Regression: Unhelpful error message on missing variables) on buildstream https://gitlab.com/BuildStream/buildstream/issues/1059 | 13:50 |
| gitlab-br-bot | marge-bot123 merged MR !1551 (danielsilverstone-ct/fix-node-provenance->master: Improve node provenance in error reporting) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1551 | 13:50 |
| gitlab-br-bot | danielsilverstone-ct approved MR !1511 (aevri/spawn->master: Support the 'spawn' method of starting processes) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1511 | 13:58 |
| Kinnison | tme5: Is it your intent to close/abandon !1541, or to clean up the last bits and merge? | 13:58 |
| gitlab-br-bot | MR !1541: Clarify bst artifact subcommand help text https://gitlab.com/BuildStream/buildstream/merge_requests/1541 | 13:58 |
| *** toscalix has quit IRC | 13:59 | |
| tme5 | Kinnison, i'm not sure, i've since realised the command help text is inconsistent for a reason, because the commands actually don't behave the same right now | 14:03 |
| Kinnison | oh dear | 14:04 |
| Kinnison | :D | 14:04 |
| Kinnison | Check with jennis -- he was doing some artifacty stuff recently | 14:04 |
| tme5 | something like my MR ought to be done eventually but idk if its sensible to do beforehand | 14:04 |
| Kinnison | Maybe you can combine efforts | 14:04 |
| tme5 | yeah, he is working on the features that i assumed were there | 14:04 |
| Kinnison | heh | 14:05 |
| jennis | tme5, I like the fact that you've moved the general artifact subcommand help text into `def artifact`, we should at least take that change from your MR and land that | 14:15 |
| jennis | It's silly that we're repeating ourselves per artifact subcommand | 14:15 |
| gitlab-br-bot | jennis approved MR !1529 (becky/artifact_list_contents->master: Addition of bst artifact list-contents) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1529 | 14:17 |
| * Kinnison concurs | 14:18 | |
| tme5 | ah ok | 14:20 |
| tme5 | it seems that the other thing is that the other artifact commands (checkout/pull/push) only fully operate on elements, but currently accept and partially operate on artifact refs | 14:20 |
| jennis | Yeah this is something I'm working on | 14:21 |
| tme5 | ahh thought so | 14:21 |
| jennis | Long story short, it's not easy to reconstruct a dependency graph from a given artifact (which could be from any point in the past) | 14:21 |
| jennis | So we can't support --deps options (with artifact refs) for these commands, yet | 14:22 |
| tme5 | makes sense | 14:23 |
| tme5 | ok, so I can clear up my MR, or do you want to cherry-pick the commit? not sure how such things are done | 14:23 |
| jennis | So your MR makes a lot of changes where checkout/pull/push now takes "artifacts/artifact" instead of element | 14:24 |
| jennis | Although we do eventually want this, perhaps we just hold off for now | 14:24 |
| jennis | clear up the MR :) no need for another one | 14:24 |
| tme5 | yeah I agree, seeing as you're adding artifact show and Becky has just done list-contents, artifact commands are changing a lot :) | 14:25 |
| tme5 | will do | 14:25 |
| jennis | Indeed | 14:25 |
| gitlab-br-bot | tlater approved MR !1529 (becky/artifact_list_contents->master: Addition of bst artifact list-contents) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1529 | 14:25 |
| jennis | Also, we intend to split artifact related commands from the project (you shouldn't need to be inside a project to do artifact stuff) | 14:25 |
| tlater[m] | jennis: That'll be weird. Running `bst` outside of a project directory and actually getting it to do something | 14:28 |
| jennis | But if you want to do stuff with your artifacts, why do you need to be in a project? | 14:28 |
| jennis | tristan gave some cool use cases for this, one sec | 14:29 |
| tlater[m] | Nah, I'm with you. It'll just be weird because I'm so used to BuildStream refusing to work outside a project dir :) | 14:29 |
| jennis | hehe yeah | 14:30 |
| jennis | It still won't, for most commands anyway | 14:30 |
| *** jonathanmaw has quit IRC | 14:32 | |
| gitlab-br-bot | jennis approved MR !1529 (becky/artifact_list_contents->master: Addition of bst artifact list-contents) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1529 | 14:32 |
| gitlab-br-bot | tlater approved MR !1529 (becky/artifact_list_contents->master: Addition of bst artifact list-contents) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1529 | 14:32 |
| *** lachlan has quit IRC | 14:58 | |
| gitlab-br-bot | marge-bot123 merged MR !1529 (becky/artifact_list_contents->master: Addition of bst artifact list-contents) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1529 | 15:01 |
| gitlab-br-bot | beckyella16 opened issue #1106 (Addition of '--file' option to 'bst artifact log') on buildstream https://gitlab.com/BuildStream/buildstream/issues/1106 | 15:07 |
| tme5 | is the unique key returned by a Source plugin used to store anything? or is it input into the cache key of the element? | 15:14 |
| tlater[m] | tme5: I think it was the latter. Check in element.py for cache key creation. | 15:18 |
| jennis | tme5, have you seen this: https://buildstream.gitlab.io/buildstream/buildstream.plugin.html#buildstream.plugin.Plugin.get_unique_key ? | 15:19 |
| tme5 | ahh | 15:19 |
| *** lachlan has joined #buildstream | 15:19 | |
| tlater[m] | jennis: Handy doc! | 15:21 |
| tme5 | yeah that does answer my question, sorry hahah, ty | 15:21 |
| *** tiagogomes has quit IRC | 15:33 | |
| *** tiagogomes has joined #buildstream | 15:34 | |
| *** tiagogomes has quit IRC | 15:52 | |
| *** tiagogomes has joined #buildstream | 15:56 | |
| *** tristan has left #buildstream | 15:59 | |
| *** tristan has joined #buildstream | 15:59 | |
| *** ChanServ sets mode: +o tristan | 15:59 | |
| *** tpollard has quit IRC | 16:04 | |
| *** tiagogomes has quit IRC | 16:06 | |
| gitlab-br-bot | beckyella16 opened MR !1552 (becky/artifact_log_file_option->master: Addition of --file option to bst artifact log) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1552 | 16:32 |
| *** bochecha has quit IRC | 16:32 | |
| *** lachlan has quit IRC | 16:35 | |
| *** tme5 has quit IRC | 16:37 | |
| *** lachlan has joined #buildstream | 16:40 | |
| *** lachlan has quit IRC | 16:53 | |
| *** lachlan has joined #buildstream | 17:02 | |
| *** lachlan has quit IRC | 17:10 | |
| *** lachlan has joined #buildstream | 17:14 | |
| *** lachlan has quit IRC | 17:20 | |
| *** lachlan has joined #buildstream | 17:30 | |
| gitlab-br-bot | marge-bot123 merged MR !1511 (aevri/spawn->master: Support the 'spawn' method of starting processes) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1511 | 17:37 |
| *** lachlan has quit IRC | 17:45 | |
| *** lachlan has joined #buildstream | 17:48 | |
| *** lachlan has quit IRC | 18:24 | |
| *** traveltissues has quit IRC | 18:34 | |
| *** lachlan has joined #buildstream | 18:51 | |
| *** rdale has quit IRC | 19:30 | |
| *** lachlan has quit IRC | 19:34 | |
| *** benschubert has quit IRC | 22:16 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!