| *** cs-shadow has quit IRC | 01:27 | |
| *** nimish has joined #buildstream | 02:29 | |
| *** abderrah1 has joined #buildstream | 03:15 | |
| *** abderrahim has quit IRC | 03:17 | |
| *** tristan has joined #buildstream | 05:17 | |
| *** nimish has quit IRC | 06:28 | |
| *** nimish has joined #buildstream | 06:58 | |
| *** ChanServ sets mode: +o tristan | 07:47 | |
| tristan | Hrrrrrm :-/ | 07:47 |
|---|---|---|
| tristan | How come when I run `bst build --track-all sdk/gtk+-3.bst`... BuildStream master is pulling stuff ? | 07:47 |
| tristan | It cannot possibly know the cache keys until the elements get tracked | 07:47 |
| tristan | ahh, answer my own question | 07:48 |
| tristan | It's pulling freedesktop-sdk stuff | 07:48 |
| *** toscalix has joined #buildstream | 08:22 | |
| *** tristan has quit IRC | 08:26 | |
| *** alatiera has joined #buildstream | 08:36 | |
| *** finn has joined #buildstream | 08:41 | |
| Kinnison | :-) | 09:03 |
| *** nimish has quit IRC | 09:07 | |
| *** nimish has joined #buildstream | 09:08 | |
| *** solid_black has joined #buildstream | 09:14 | |
| *** toscalix has quit IRC | 09:17 | |
| jjardon | Hi buildstreamers, ok to merge https://gitlab.com/BuildStream/buildstream/merge_requests/972 for 1.4 then? The MR already has some reviews | 09:21 |
| *** tristan has joined #buildstream | 09:26 | |
| jmac | Hmm "remote: GitLab: API is not accessible" | 09:32 |
| jmac | Website seems fine but I can't push with git | 09:32 |
| Kinnison | pushing over https or ssh ? | 09:33 |
| jmac | ssh | 09:34 |
| * Kinnison nods, I just tested over https and have the same issue | 09:34 | |
| Kinnison | It seems to be that the hooks inside gitlab aren't running | 09:34 |
| * Kinnison imagines they'll be on it | 09:34 | |
| jmac | Looks like it's working again | 09:49 |
| *** jonathanmaw has joined #buildstream | 09:58 | |
| jmac | gitlab-br-bot doesn't seem to have returned after the disconnection on Monday | 10:00 |
| *** tpollard has joined #buildstream | 10:10 | |
| *** nimish has quit IRC | 10:33 | |
| valentind | tristan, I think the issue with ruamel was maybe it was missing on some older docker images. But I forgot to write down on the MR why. | 10:33 |
| *** nimish has joined #buildstream | 10:33 | |
| *** lachlan has joined #buildstream | 10:34 | |
| *** nimish has quit IRC | 10:38 | |
| *** nimish has joined #buildstream | 10:38 | |
| *** nimish has quit IRC | 10:48 | |
| *** nimish has joined #buildstream | 10:49 | |
| *** raoul_ is now known as raoul | 10:50 | |
| raoul | I've taken MR !969 out of WIP if anyone wants to take a look. Probably most relevant to juergbi and tristan who discussed it on the mailing list (and sander but I'm not sure if he's here or what his irc name is) | 10:52 |
| *** nimish_ has joined #buildstream | 10:54 | |
| *** nimish has quit IRC | 10:56 | |
| *** nimish_ has quit IRC | 10:57 | |
| *** lachlan has quit IRC | 10:59 | |
| *** ChanServ sets mode: +o tristan | 11:03 | |
| tristan | raoul, not diving into a full review right now... looks like a big one, just added one comment for juergbi there | 11:03 |
| tristan | valentind, Do we know that BuildStream works for the selected range of ruamel.yaml versions ? | 11:03 |
| tristan | just curious, in any case nailing it down to something a bit more specific will be good, and will hopefully resolve the new crasher | 11:04 |
| raoul | tristan, nw I'll keep an eye out for future comments | 11:04 |
| valentind | tristan, I remember I tested python 3 and found it works between 41 and 51. | 11:04 |
| valentind | But it did not work on .40 | 11:05 |
| tristan | Mkay, it is strange that I have had .30 for a long time and it has been "working" for me | 11:05 |
| *** lachlan has joined #buildstream | 11:05 | |
| tristan | valentind, Do you recall when python3 support was added ? | 11:05 |
| valentind | tristan, this I do not know. | 11:06 |
| tristan | valentind, Ummmmm | 11:07 |
| tristan | Okay but, it was clearly >= 0.15 ? | 11:07 |
| valentind | Python 3.5 was on 0.10.10 according to the changes. | 11:07 |
| valentind | Python 3.6 worked on 0.14. This I remember. | 11:07 |
| tristan | valentind, the "right thing" to do would be to follow the maintainers instructions, and say <= 0.15 | 11:07 |
| valentind | It was python 3.7 that needed .41. | 11:07 |
| valentind | tristan, that does not work with python 3.7. | 11:07 |
| tristan | Ohhhhh, there is a 3.7 already, and no ruamel.yaml ? | 11:08 |
| tristan | Do we support 3.7 ? | 11:08 |
| valentind | We need at least 0.15.41. | 11:08 |
| valentind | BuildStream works on Python 3.7. | 11:08 |
| tristan | In the above, s/and no ruamel.yaml/and no stable ruamel.yaml version which supports 3.7/ | 11:08 |
| valentind | I know juergbi always uses 3.7. | 11:08 |
| tristan | So, for BuildStream to support 3.7, we *must* use an unstable version of ruamel.yaml, that is not very ideal :-S | 11:09 |
| tristan | Probably worth a comment in setup.py beside the ruamel.yaml requirement, mentioning how dangerous and irresponsible this is :-S | 11:09 |
| valentind | tristan, I think it is irresponsible for ruamel not to maintain 0.14. | 11:11 |
| tristan | Well | 11:12 |
| tristan | valentind, That doesnt make us more responsible for depending on a library which does not | 11:12 |
| tristan | That reflects badly on ruamel.yaml maintainers, and on Python in general for bumping minor versions and introducing API breakages | 11:13 |
| tristan | This reflects badly on us :) | 11:13 |
| tristan | So Python maintainers, ruamel.yaml maintainers, and us... all irresponsible :) | 11:13 |
| tristan | we really need a C library for round tripping YAML anyway :-S | 11:14 |
| valentind | I can look at doing that. | 11:15 |
| tristan | I know ! | 11:18 |
| *** lachlan has quit IRC | 11:18 | |
| tristan | valentind, But it might take a while :) | 11:18 |
| tristan | hehe | 11:18 |
| tristan | I mean, we *also* need a python binding to it, which doesnt need to match `ruamel.yaml` exactly... | 11:19 |
| tristan | I guess not the hardest part; it needs to support round tripping, which I dare say you might have a better chance at achieving when writing from scratch | 11:20 |
| tristan | because ruamel.yaml is buggy in some cases, loses some existing whitespace when round tripping | 11:20 |
| tristan | and needs to remember the line/column for every loaded node, that aint hard | 11:20 |
| tristan | and then the whole parser | 11:20 |
| tristan | seems like... a lot ? | 11:20 |
| valentind | tristan, It is a difficult problem to keep indentation. | 11:21 |
| tristan | Well, I suppose it's much less of a difficult problem going off the assumption that anyway you *must* have the line/column information for every key/value/listelement/etc | 11:21 |
| valentind | The parser, we can probably use an LL(k) parser generator. | 11:21 |
| tristan | So that intendation *has* to be there anyway | 11:21 |
| *** lachlan has joined #buildstream | 11:25 | |
| *** lachlan has quit IRC | 11:30 | |
| *** lachlan has joined #buildstream | 11:31 | |
| *** solid_black has quit IRC | 11:32 | |
| *** solid_black has joined #buildstream | 11:32 | |
| *** lachlan has quit IRC | 11:35 | |
| laurence | Juerg is probably not able to be online much today, but i agreed with him yesterday that I'd chase people for reviews on this optimization for local junctions - https://gitlab.com/BuildStream/buildstream/merge_requests/290 | 11:40 |
| laurence | it's a relatively small patch if anyone has the brainspace | 11:40 |
| tristan | laurence, reviewed, was a no brainer really | 11:43 |
| *** nimish has joined #buildstream | 11:43 | |
| tristan | juergbi, has been very busy and the MR got lost under a mountain of other things I suspect, it was just a very easy change to his patch which he now fixed | 11:44 |
| jmac | Yep, I've just taken a look and it looks good to me too | 11:45 |
| laurence | tristan, that's exactly it, cheers for the review | 11:45 |
| *** lachlan has joined #buildstream | 11:50 | |
| *** nimish has quit IRC | 12:03 | |
| *** nimish has joined #buildstream | 12:04 | |
| juergbi | ta, will merge | 12:11 |
| juergbi | tristan: full build of freedesktop-sdk/gnome-meta via remote execution requires !915 would be great to merge this ;) | 12:17 |
| tristan | Aha ! | 12:17 |
| tristan | or, it will be slow without command batching | 12:18 |
| juergbi | no, that's not just a performance issue, it will fail | 12:18 |
| *** alatiera has quit IRC | 12:18 | |
| juergbi | because at least cairo has peculiar mtime requirements between configure and make steps | 12:18 |
| tristan | Ah that issue | 12:18 |
| tristan | Right | 12:18 |
| tristan | Ah I didnt know about contextlib.suppress() | 12:21 |
| tristan | I usually use ExitStack() for conditional context managers | 12:21 |
| *** alatiera has joined #buildstream | 12:21 | |
| tristan | juergbi, see: https://gitlab.com/BuildStream/buildstream/merge_requests/915#note_120392028 | 12:27 |
| *** lachlan has quit IRC | 12:28 | |
| tristan | jmac, since you spoke up in the unit test debate on the ML, you might be interested to comment on !973, either on the MR or ML... maybe particularly this: https://gitlab.com/BuildStream/buildstream/merge_requests/973#note_120329379 which I wrote up (amongst other comments this morning) | 12:41 |
| tristan | jmac, I totally accept that your opinion might not reinforce my own, and am willing to bend my own obstinate viewpoints... but for that I need other people to voice opinions :) | 12:42 |
| tristan | Kinnison, maybe you want to weigh in on this as well ? | 12:42 |
| tristan | it's not great when the whole argument is just my opinion vs Angelos's | 12:43 |
| jmac | I'll try and offer some comments on it. | 12:46 |
| jmac | To be honest, the volume of traffic on the mailing list and gitlab is too much for me to keep track of and make informed comments on. | 12:46 |
| *** nimish has quit IRC | 12:52 | |
| *** lachlan has joined #buildstream | 12:53 | |
| tristan | jmac, Indeed, do you follow the project wide notifications ? | 12:58 |
| * paulsherwood notices that folks on the ml don't use 'snip', which makes it seem as though there's way more to read | 12:59 | |
| paulsherwood | (and some folks just write very long emails :-) | 12:59 |
| tristan | jmac, I'm slowly planning on reducing spam there so that developers can subscribe the notifications stream and be included in the main feedback loop, one of the big steps towards this is to stop with all the premature WIP merge requests which bombard the stream with tons of irrelevant, premature information | 12:59 |
| *** lachlan has quit IRC | 13:00 | |
| *** lachlan has joined #buildstream | 13:01 | |
| Kinnison | tristan: I'll have a look at that MR, I'm broadly in favour of testing the right surfaces, which last we spoke matched both your ideas and Angelos', so I need to try and work out where the disconnect is | 13:19 |
| jmac | tristan: My GitLab notifications come through a completely separate account; they're not a problem | 13:21 |
| jmac | buildstream-list on its own is too high volume | 13:21 |
| *** lachlan has quit IRC | 13:22 | |
| *** tristan has quit IRC | 13:29 | |
| *** solid_black has quit IRC | 13:37 | |
| jmac | Wow, you really ran with the rivet analogy, huh | 13:39 |
| *** nimish has joined #buildstream | 13:59 | |
| *** tristan has joined #buildstream | 14:02 | |
| jjardon | tristan: any thoughts about https://gitlab.com/BuildStream/buildstream/merge_requests/972 ? I think is pretty important | 14:07 |
| *** nimish has quit IRC | 14:09 | |
| *** nimish has joined #buildstream | 14:09 | |
| tpollard | jjardon: interesting | 14:10 |
| *** cs-shadow has joined #buildstream | 14:12 | |
| juergbi | tristan: as mentioned in a mail last week, do you agree that #631 should be a 1.4 blocker? I think #780 should be a blocker as well | 14:20 |
| juergbi | #780 should be pretty easy to fix, if I'm not missing anything | 14:21 |
| *** lachlan has joined #buildstream | 14:22 | |
| *** ChanServ sets mode: +o tristan | 14:37 | |
| tristan | juergbi, I thought it was pretty ridiculous of a question and didnt answer | 14:38 |
| tristan | juergbi, i.e. it looks like "we've implemented remote execution, there is just no way to enable it !" | 14:38 |
| tristan | It is not good enough to say that we want to include the remote execution feature ? Or do we need to spell out every little detail in little issue reports ? | 14:38 |
| tristan | juergbi, regarding #780, I think that the question revolves mostly around "Will adding the encryption keys to the configuration cause a user configuration API break ?" if yes, then blocker | 14:40 |
| tristan | Otherwise, I'd rather lay off the blocker buzzword and expect that that might make sense to land in 1.4 anyway | 14:41 |
| tristan | Without adding undue stress to people | 14:41 |
| juergbi | I don't think it's an issue compatibility wise. I just think it's pretty bad to release 1.4 without those fixes. and marking them as blocker would show that | 14:49 |
| tristan | juergbi, How can remote execution possibly work at all if you cannot configure the URL for it ? | 14:54 |
| tristan | Since you are saying "fixes" | 14:55 |
| tristan | Is there some project data ? That would be wrong, a remote execution service is certainly a user configuration for which the project can make a recommendation of a default | 14:55 |
| tristan | but it is not "project data", it lives on Context for sure | 14:56 |
| tristan | juergbi, if that's the case -> please blocker | 14:56 |
| tristan | juergbi, Please blocker anyway; the only reason I didnt respond is because it sounded like such an absurd question :) | 14:56 |
| tristan | Like duh... of course | 14:56 |
| tristan | juergbi, take this comment I wrote today for context: https://gitlab.com/BuildStream/buildstream/issues/566#note_120400141 | 14:58 |
| tristan | There is a pattern already for what configurations are in what domain; remote execution should not just break this pattern | 14:58 |
| juergbi | tristan: yes, that's the case, which is why I think it's extremely important | 15:00 |
| juergbi | will mark it as blocker. wasn't sure about whether I can just go ahead :) | 15:00 |
| tristan | juergbi, yes then, please -> blocker for #631 at least | 15:00 |
| tristan | juergbi, for the other, I just feel that "blocker" is a big word, there is a difference between "really really should" and "must" | 15:01 |
| tristan | Let's try to be gentle with the "must" messaging, and presume that since it is very important, it will very highly likely get into 1.4 anyway | 15:01 |
| tpollard | tristan: a quick question on the recommendations thing, if a user has config that explicitly sets something to the default value, then the project cannot override it, even if it is set to the default value? | 15:01 |
| *** nimish has quit IRC | 15:01 | |
| tristan | tpollard, Correct, for user preferences, the project can never override the user's choice | 15:02 |
| juergbi | ok, marked the other as urgent instead (our priority level 3/4). I think that's warranted | 15:02 |
| tristan | tpollard, There is a way and pattern in which this is implemented which makes that always work | 15:02 |
| tristan | I.e. the composition order of Context dictionaries | 15:03 |
| * tpollard nods | 15:03 | |
| tpollard | just wanted to confirm | 15:03 |
| tristan | admittedly, I believe the way that the project recommends artifact servers is totally a mess and is an exception to that rule | 15:04 |
| tristan | as it goes ahead and parses things in the wrong places (ArtifactCache and ArtifactCacheSpec etc, goes and takes ownership of config parsing which it really shouldnt, imnsho) | 15:04 |
| juergbi | we also still have the outstanding issue about the current combination of artifact servers across junctions may not always be what the user wants | 15:07 |
| tpollard | yep | 15:08 |
| juergbi | but I don't think we have a full proposal yet how to better handle this | 15:08 |
| *** bochecha has joined #buildstream | 15:12 | |
| jonathanmaw | hmm, it seems `bst source-bundle` doesn't work properly if there's an open workspace | 15:17 |
| jonathanmaw | given there are discussions on retiring it (https://gitlab.com/BuildStream/buildstream/issues/672), I gather I should stop making source-bundle specific changes | 15:18 |
| *** toscalix has joined #buildstream | 15:28 | |
| *** ChanServ sets mode: +o ironfoot | 15:48 | |
| *** lachlan has quit IRC | 16:09 | |
| Kinnison | Nexus: looks like you have some lint failures on your MR, but otherwise the tests seem to be passing now | 16:18 |
| Kinnison | Oh no, one test failure too | 16:18 |
| Kinnison | :( | 16:18 |
| *** nimish has joined #buildstream | 16:18 | |
| Kinnison | May need to tweak error domains | 16:18 |
| *** xjuan has joined #buildstream | 16:30 | |
| *** lachlan has joined #buildstream | 16:41 | |
| jmac | juergbi: Thanks for the re-review on !946. I've fixed the bugs you found. | 16:46 |
| *** gitlab-br-bot has quit IRC | 16:51 | |
| *** gitlab-br-bot has joined #buildstream | 16:51 | |
| ironfoot | !946 | 16:53 |
| gitlab-br-bot | MR !946: Split remote execution from artifact cache https://gitlab.com/BuildStream/buildstream/merge_requests/946 | 16:53 |
| *** nimish has quit IRC | 17:03 | |
| *** nimish has joined #buildstream | 17:03 | |
| *** finn has quit IRC | 17:08 | |
| *** finn has joined #buildstream | 17:09 | |
| Nexus | Kinnison: yay it passed \o/ | 17:24 |
| Kinnison | Nexus: wonderful, have you dealt with all the discussions do you think? | 17:25 |
| Nexus | Kinnison: other than the one i opened for juergbi and tristan to reply to, i believe so. | 17:26 |
| Nexus | i've split it into more commits now, i'd apprecite your feedback on the structure | 17:26 |
| Kinnison | Okay, I'll have a look, though it might have to be in the morning since I need to skedaddle in a few mins | 17:27 |
| Nexus | no problem :) | 17:27 |
| * Kinnison skims the commit list now | 17:27 | |
| Kinnison | The only comment I have from the skim is that the first commit might be better worded as: "_context.py: Move `warn()` here from _plugin.py" | 17:28 |
| Nexus | yeah i wasn't certain about that one, thanks | 17:28 |
| *** nimish has quit IRC | 17:28 | |
| *** nimish has joined #buildstream | 17:29 | |
| Nexus | Done | 17:29 |
| Kinnison | I'll look deeper tomorrow, modulo any feedback from juergbi or tristan :-) | 17:29 |
| Nexus | kk c: | 17:30 |
| *** finn has quit IRC | 17:32 | |
| *** toscalix has quit IRC | 17:43 | |
| *** tpollard has quit IRC | 17:46 | |
| *** jonathanmaw has quit IRC | 17:58 | |
| *** nimish has quit IRC | 17:59 | |
| *** nimish has joined #buildstream | 18:00 | |
| *** xjuan has quit IRC | 18:04 | |
| gitlab-br-bot | juergbi approved MR !946 (jmac/remote_execution_split->master: Split remote execution from artifact cache) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/946 | 18:15 |
| gitlab-br-bot | jmacarthur closed issue #750 (Alter SandboxRemote to use separate CAS store) on buildstream https://gitlab.com/BuildStream/buildstream/issues/750 | 18:17 |
| gitlab-br-bot | jmacarthur merged MR !946 (jmac/remote_execution_split->master: Split remote execution from artifact cache) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/946 | 18:17 |
| *** bethw has quit IRC | 18:17 | |
| *** valentind has quit IRC | 18:17 | |
| *** johnward has quit IRC | 18:17 | |
| *** WSalmon has quit IRC | 18:17 | |
| *** laurence has quit IRC | 18:17 | |
| *** mablanch has quit IRC | 18:17 | |
| *** aiden has quit IRC | 18:17 | |
| *** aiden has joined #buildstream | 18:17 | |
| *** tiagogomes_ has quit IRC | 18:18 | |
| *** johnward has joined #buildstream | 18:18 | |
| *** nimish has quit IRC | 18:19 | |
| *** valentind has joined #buildstream | 18:19 | |
| *** nimish has joined #buildstream | 18:20 | |
| *** bethw has joined #buildstream | 18:20 | |
| *** raoul has quit IRC | 18:20 | |
| *** laurence has joined #buildstream | 18:21 | |
| *** WSalmon has joined #buildstream | 18:21 | |
| *** finn has joined #buildstream | 18:23 | |
| *** mablanch has joined #buildstream | 18:23 | |
| *** nimish has quit IRC | 18:30 | |
| *** nimish has joined #buildstream | 18:30 | |
| *** alatiera has quit IRC | 18:55 | |
| *** nimish_ has joined #buildstream | 19:00 | |
| *** nimish has quit IRC | 19:02 | |
| *** nimish_ is now known as nimish | 19:02 | |
| *** lachlan has quit IRC | 19:06 | |
| *** nimish has quit IRC | 19:22 | |
| *** nimish has joined #buildstream | 19:24 | |
| *** hergertme has joined #buildstream | 19:31 | |
| *** finn has quit IRC | 19:33 | |
| *** nimish has quit IRC | 19:48 | |
| *** nimish has joined #buildstream | 19:49 | |
| *** finn has joined #buildstream | 19:59 | |
| *** finn has joined #buildstream | 20:57 | |
| *** bochecha has quit IRC | 21:20 | |
| *** finn has joined #buildstream | 21:28 | |
| *** finn has quit IRC | 21:35 | |
| *** finn has joined #buildstream | 21:40 | |
| *** nimish has quit IRC | 22:17 | |
| *** tristan has quit IRC | 22:30 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!