| *** xjuan has joined #buildstream | 00:08 | |
| *** xjuan has quit IRC | 01:08 | |
| *** xjuan has joined #buildstream | 01:27 | |
| *** alatiera has quit IRC | 01:33 | |
| *** xjuan has quit IRC | 01:35 | |
| *** nimish has joined #buildstream | 01:48 | |
| *** nimish has quit IRC | 03:19 | |
| *** nimish has joined #buildstream | 03:19 | |
| *** nimish has quit IRC | 03:25 | |
| *** juanalday has quit IRC | 03:33 | |
| *** juanalday has joined #buildstream | 04:36 | |
| *** juanalday has quit IRC | 04:58 | |
| *** juanalday has joined #buildstream | 05:04 | |
| *** mohan43u has quit IRC | 06:00 | |
| *** mohan43u has joined #buildstream | 06:03 | |
| *** juanalday has quit IRC | 06:56 | |
| *** tristan has joined #buildstream | 08:30 | |
| *** juanalday has joined #buildstream | 08:33 | |
| *** phildawson has joined #buildstream | 08:41 | |
| *** juanalday has quit IRC | 08:47 | |
| *** toscalix has joined #buildstream | 08:53 | |
| *** ChanServ sets mode: +o tristan | 09:09 | |
| tristan | Why is there now both tests/testutils and buildstream/plugintestutils ? And why do they *both* contain the CLI ? | 09:09 |
|---|---|---|
| tristan | or do they | 09:10 |
| phildawson | tristan, they don't both contain the cli | 09:11 |
| phildawson | we wanted to expose an api to make it easy for plugin authors to test their plugins without maintaining their own copies of a load of BuildStream's test utilities | 09:12 |
| tristan | No they don't, I was confused | 09:12 |
| tristan | phildawson, Yes that is a good plan, I was looking at adding a new test and was curious why I saw one test importing stuff from testutils and another importing from pluginutils | 09:12 |
| phildawson | tristan, you had me worried there that I'd left a second copy hanging around :P | 09:13 |
| tristan | So stuff we can standardize on ends up in plugintestutils, and the extra local stuff in testutils, this seems very sane yes | 09:14 |
| tristan | phildawson, Sorry for that :) | 09:14 |
| tristan | I had a bit of a jump :) | 09:14 |
| phildawson | :) | 09:14 |
| *** raoul has joined #buildstream | 09:50 | |
| *** jonathanmaw has joined #buildstream | 09:51 | |
| *** jennis_ has joined #buildstream | 09:58 | |
| *** jennis_ has quit IRC | 09:59 | |
| gitlab-br-bot | tristanvb opened issue #926 (Subproject overlap policy is inherited by including projects) on buildstream https://gitlab.com/BuildStream/buildstream/issues/926 | 10:15 |
| aevri | juergbi: I like the idea of a WIP bot :) It could WIP the things and then close them later too, while politely explaining the rules. | 10:20 |
| Kinnison | While I dislike the idea of having to write more code to do this, I do like that it'll apply the rules consistently and without the lingering taint of "a human said this, maybe they just don't like *me*" so I'd reluctantly +1 it too | 10:21 |
| * tristan shares Kinnison sentiment here fwiw | 10:23 | |
| juergbi | well, having a bot do it fully automated without a human first assigning the MR to the bot would be a bit of an issue, as it couldn't check whether the MR was stalled because of lack of review or because of the MR author not responding | 10:25 |
| juergbi | unless it could skip those non-WIP MRs that are CI green and have no open discussion points | 10:25 |
| Kinnison | CI green, no open discussion points, => needs huge rebase which will explode horribly | 10:26 |
| tristan | sounds pretty complex | 10:26 |
| Kinnison | Either the rules can be entirely applied by a bot, or they can't | 10:27 |
| Kinnison | if they can't, a bot isn't that helpful IMO | 10:27 |
| juergbi | possibly, yes | 10:28 |
| coldtom | hi, is there a way to specify an option in project.conf that can be any string, not just one of a restricted set? | 10:33 |
| *** bochecha has joined #buildstream | 10:33 | |
| tristan | coldtom, no, for the same reason there cannot be an unbounded integer | 10:34 |
| tristan | coldtom, a goal being that only a strictly limited, controllable set of outputs are possible for a given project | 10:35 |
| juergbi | tristan: overlaps, oh, very interesting case | 10:41 |
| tristan | It is :-S | 10:42 |
| tristan | And the plot thickened, it seems to almost work perfectly | 10:43 |
| tristan | But I have to get the test case to fail still ! | 10:43 |
| *** alatiera has joined #buildstream | 10:57 | |
| aevri | re: wip bot - I agree less code is better, perhaps if we manually apply the rules for a bit, we'll get a better idea of if we could automate that fully without annoying folks. | 11:03 |
| aevri | I do think that it would be slightly easier for me to assign something to a wip-bot than copy-paste something. The programming cost may not be worth it though. | 11:05 |
| *** tristan has quit IRC | 11:20 | |
| *** nimish has joined #buildstream | 11:31 | |
| juergbi | jmac: was your fdo-sdk build issue due to server issues of upstream sources (possibly alsa)? or did you encounter a possible incompatibility with buildstream master? | 11:34 |
| juergbi | I built fdo-sdk with buildstream master fairly recently | 11:34 |
| jmac | Yes, alsa sounds familiar | 11:35 |
| jmac | tar source at desktop/alsa-plugins.bst [line 25 column 2]: Error mirroring ftp://ftp.alsa-project.org/pub/plugins/alsa-plugins-1.1.6.tar.bz2: <urlopen error ftp error: OSError(113, 'No route to host')> | 11:35 |
| jmac | That was on the latest tag... | 11:35 |
| juergbi | ok, right. maybe I could give it a try, I might already have that in my local sources directory | 11:36 |
| jmac | I got a different error on current head of 18.08 but that's no longer in my scrollback | 11:37 |
| gitlab-br-bot | aevri opened issue #927 ('true' by itself is parsed as 'True' in .bst files) on buildstream https://gitlab.com/BuildStream/buildstream/issues/927 | 11:37 |
| juergbi | ah, latest head of 18.08 changed the alsa URLs to git | 11:39 |
| gitlab-br-bot | martinblanchard opened issue #928 (Remote Execution: Implement RequestMetadata) on buildstream https://gitlab.com/BuildStream/buildstream/issues/928 | 11:40 |
| *** sebastian has joined #buildstream | 11:43 | |
| *** juanalday has joined #buildstream | 11:44 | |
| *** juanalday has quit IRC | 11:47 | |
| *** raoul has quit IRC | 12:35 | |
| *** raoul has joined #buildstream | 13:10 | |
| *** flatmush has quit IRC | 13:30 | |
| *** flatmush has joined #buildstream | 13:31 | |
| *** flatmush has quit IRC | 13:48 | |
| *** flatmush has joined #buildstream | 13:49 | |
| *** lachlan has joined #buildstream | 14:14 | |
| *** lachlan has quit IRC | 14:33 | |
| *** lachlan has joined #buildstream | 14:49 | |
| raoul | Realised asked it last week but missed your answer juergbi, for the source cache clean up, what do you think should be done in the mean time? Quite a few tests fail as clearing an artifact doesn't necessarily free up that much space (especially if it's an import). I could fiddle around with cache sizes for these tests, but that doesn't seem like a good solution, would something that tries to clear sources not involved in the current build be | 15:03 |
| raoul | sufficient? | 15:03 |
| tpollard | what does 'xfailed' mean as a test result? | 15:24 |
| Kinnison | an xfail is traditionally a known-failure failing to fail | 15:25 |
| Kinnison | In past jobs, we had PASS - good. FAIL - bad. KFAIL - good (known failure, failing). XPASS/XFAIL - bad (known failure failing to fail) | 15:26 |
| tpollard | it seems to think xfail is fine | 15:27 |
| Kinnison | maybe it's spelling kfail and xfail the same | 15:29 |
| Kinnison | IYSWIM | 15:29 |
| tpollard | I'll try against master | 15:29 |
| juergbi | I marked one test as xfail because it is broken | 15:33 |
| tpollard | aha | 15:34 |
| juergbi | xfail is expected failure in pytest, afaik | 15:34 |
| tpollard | tests/integration/cachedfail.py::test_push_cached_fail[quit] ? | 15:34 |
| juergbi | yes, see #534 | 15:34 |
| gitlab-br-bot | Issue #534: Failed builds not pushing artifacts on quit https://gitlab.com/BuildStream/buildstream/issues/534 | 15:34 |
| juergbi | raoul: maybe ArtifactCache.clean() should actually be in CASQuota and use all refs until we do the reference storage service split (AaaP) | 15:38 |
| juergbi | may not be trivial to handle the required elements in that case | 15:38 |
| juergbi | although I suppose a list of required refs could be constructed | 15:39 |
| juergbi | the issue is a bit that there is no place where this code would fit | 15:39 |
| tpollard | cheers juergbi | 15:40 |
| raoul | Yeah I'll have to have a thing and look about where this might sensibly fit | 15:40 |
| raoul | s/thing/think | 15:41 |
| juergbi | raoul: maybe something like ArtifactCache and SourceCache registering a callback in CASQuota or so | 15:41 |
| juergbi | (for the required refs) | 15:42 |
| raoul | Not sure I quite follow, do you mean ArtifactCache and SourceCache keeping a list of required refs and have them tell CASQuota this? | 15:44 |
| *** nimish has quit IRC | 15:45 | |
| *** lachlan has quit IRC | 15:52 | |
| juergbi | raoul: ArtifactCache already has this internally, right? and something similar could be done for sources | 16:01 |
| juergbi | CASQuota would then collect those sets from the two cache classes | 16:02 |
| raoul | yeah it does, just looking through it atm to figure exactly what it's doing. I guess when you're talking about registering the callback that will be for updating the cache keys for artifacts? | 16:02 |
| raoul | As that still needs to be done if I've understood it correctly | 16:02 |
| juergbi | I'd just move the code that generates 'required_artifacts' to a separate method and pass that method to CASQuota, so that CASQuota can get 'required_artifacts' whenever it needs to | 16:03 |
| juergbi | (moving the rest of the clean() method to CASQuota) | 16:04 |
| raoul | Got ya, I fear this will end up throwing more problems somewhere else, but I'll have a go | 16:05 |
| *** lachlan has joined #buildstream | 16:08 | |
| *** raoul has quit IRC | 16:09 | |
| *** raoul has joined #buildstream | 16:11 | |
| juergbi | jmac: I'm wondering whether your freedesktop-sdk build issue is related to your host/platform. I assume this is on native Linux or is this in a VM? also, what CPU (as that affects /proc/cpuinfo)? | 16:14 |
| jmac | Native Ubuntu 18.04.1 on Intel(R) Core(TM) i7-4790 CPU @ 3.60GHz | 16:17 |
| juergbi | ok, thanks | 16:19 |
| * juergbi is confused | 16:19 | |
| jmac | I'll have more of a poke around later. The main problem is it takes about hours to get to the point where it crashes, so it's not an easy thing to fix | 16:20 |
| *** bochecha has quit IRC | 16:37 | |
| laurence | juergbi, did you receive a request to accept the marge bot into the BuildStream gitlab group? | 16:38 |
| juergbi | laurence: now I see it. last time I checked it wasn't there | 16:40 |
| juergbi | accepted | 16:41 |
| laurence | thanks | 16:42 |
| *** lachlan has quit IRC | 16:44 | |
| *** alatiera has quit IRC | 16:58 | |
| *** lachlan has joined #buildstream | 17:00 | |
| *** alatiera has joined #buildstream | 17:03 | |
| *** lachlan has quit IRC | 17:06 | |
| gitlab-br-bot | LaurenceUrhegyi opened (was WIP) MR !1171 (laurence/update-readme->master: Update CONTRIBUTING.rst to add paragraph on new MR policy) on buildstream https://gitlab.com/BuildStream/buildstream/merge_requests/1171 | 17:23 |
| *** lachlan has joined #buildstream | 17:23 | |
| *** Kinnison has quit IRC | 17:34 | |
| *** Kinnison has joined #buildstream | 17:35 | |
| *** lachlan has quit IRC | 17:40 | |
| *** toscalix has quit IRC | 17:48 | |
| *** juanalday has joined #buildstream | 17:54 | |
| *** lachlan has joined #buildstream | 17:58 | |
| *** phildawson has quit IRC | 18:02 | |
| *** phildawson has joined #buildstream | 18:02 | |
| WSalmon | test | 18:03 |
| jmac | Receiving you loud and clear | 18:04 |
| *** tpollard has quit IRC | 18:05 | |
| WSalmon | ta | 18:06 |
| *** jonathanmaw has quit IRC | 18:07 | |
| *** lachlan has quit IRC | 18:29 | |
| *** raoul has quit IRC | 18:35 | |
| *** slaf_ has joined #buildstream | 18:41 | |
| *** slaf_ has joined #buildstream | 18:41 | |
| *** slaf has quit IRC | 18:43 | |
| *** slaf_ is now known as slaf | 18:43 | |
| *** phildawson has quit IRC | 18:46 | |
| *** nimish has joined #buildstream | 18:47 | |
| *** nimish has quit IRC | 19:08 | |
| *** sebastian has quit IRC | 19:15 | |
| *** nimish has joined #buildstream | 19:18 | |
| *** tristan has joined #buildstream | 19:37 | |
| *** nimish has quit IRC | 21:15 | |
| *** tristan has quit IRC | 21:32 | |
| *** juanalday has quit IRC | 21:40 | |
| *** alatiera has quit IRC | 22:36 | |
| *** sebastian has joined #buildstream | 22:59 | |
| *** juanalday has joined #buildstream | 23:02 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!