| *** Prince781 has joined #buildstream | 02:49 | |
| *** tristan has quit IRC | 02:49 | |
| *** tristan has joined #buildstream | 02:54 | |
| *** ruben has quit IRC | 03:25 | |
| *** Prince781 has quit IRC | 04:05 | |
| *** tristan has quit IRC | 04:23 | |
| *** tristan has joined #buildstream | 04:41 | |
| *** mcatanzaro has quit IRC | 05:00 | |
| *** lantw44 has quit IRC | 06:27 | |
| *** lantw44 has joined #buildstream | 06:28 | |
| *** toscalix has joined #buildstream | 08:39 | |
| *** aday has joined #buildstream | 08:39 | |
| toscalix | This project might be interesting for buildstream http://qmstr.org/ | 09:25 |
|---|---|---|
| *** noisecell has quit IRC | 09:45 | |
| *** noisecell has joined #buildstream | 09:46 | |
| *** jonathanmaw has joined #buildstream | 09:59 | |
| *** ssam2 has joined #buildstream | 10:00 | |
| *** aday has quit IRC | 10:10 | |
| *** aday has joined #buildstream | 10:56 | |
| tlater | So, this is the output of pylint in CI: https://gitlab.com/BuildStream/buildstream/-/jobs/51742595 | 10:56 |
| *** ironfoot_ is now known as ironfoot | 10:57 | |
| tlater | I've disabled most messages that cause issues in our codebase, just left a handful for show here | 10:57 |
| *** ernestask has joined #buildstream | 10:57 | |
| *** ChanServ sets mode: +o ironfoot | 10:57 | |
| tlater | The list of disabled messages starts here: https://gitlab.com/BuildStream/buildstream/blob/239-use-pylint-for-linting/.pylintrc#L67 | 10:58 |
| tlater | I've annotated them, as per persia 's suggestion yesterday, is there anyone spinning up who'd like to have a look at fixing those errors? | 10:59 |
| jennis | o/ | 11:15 |
| jennis | I will look into it | 11:16 |
| tlater | Cool, shoot if my comments are unclear jennis :) | 11:16 |
| jennis | will do | 11:16 |
| *** tristan has quit IRC | 11:43 | |
| *** aday has quit IRC | 11:51 | |
| tlater | Ack, I re-discovered an MR that needs merging before I can merge my image documentation, does anyone have time to look at https://gitlab.com/BuildStream/bst-external/merge_requests/13/diffs ? | 11:54 |
| tlater | It's a three-liner... | 11:54 |
| ssam2 | i'd like to test that it doesn't break image builds of the Freedesktop SDK | 11:56 |
| tlater | Right | 11:56 |
| ssam2 | which does use the GNU Tools, and they sometimes lack flags that Busybox tools have | 11:56 |
| ssam2 | i suspect the mkfs.ext4 invocation won't work , actually | 11:57 |
| tlater | Hm, perhaps the x86image plugin should have an option to use busybox/gnu tools? | 11:57 |
| ssam2 | that sounds terrible, but maybe unavoidable | 11:57 |
| ssam2 | the stat change should be fine | 11:59 |
| tlater | It's probably best to test it anyway | 11:59 |
| tlater | Think you could link me to the repo for this, ssam2? | 12:02 |
| ssam2 | https://gitlab.com/freedesktop-sdk/freedesktop-sdk/ | 12:02 |
| ssam2 | building a VM isn't documented yet I think | 12:03 |
| tlater | Ack | 12:03 |
| tlater | Well, I'll see if I can figure it out | 12:03 |
| ssam2 | actually the build process is a bit painful for that project right now | 12:03 |
| ssam2 | see the .gitlab-ci.yml for how to do it | 12:03 |
| ssam2 | or you can rebase https://gitlab.com/freedesktop-sdk/freedesktop-sdk/merge_requests/81 and do it the easy way using the shiny new recursive pipelines feature :-) | 12:04 |
| ssam2 | the vm element is sdk/elements/vm/minimal-vm-image-x86_64.bst | 12:04 |
| tlater | ta ssam2 | 12:05 |
| *** aday has joined #buildstream | 12:19 | |
| ssam2 | the benchmark script lives: https://pastebin.com/pExAquHn | 12:28 |
| ssam2 | and, it shows that we have almost halved startup time in 'master' and saved 10MB of memory usage on startup :-) | 12:29 |
| ssam2 | i presume that's by avoiding pkg_resources | 12:29 |
| tlater | \o/ | 12:29 |
| ssam2 | i suppose if we run a test 3 times it should return the results for all 3, as well as the average | 12:32 |
| ssam2 | otherwise it's hard to know how big the variance is | 12:32 |
| ssam2 | hmm, unless it also calculates the variance | 12:32 |
| tlater | ssam2: Are there any hooks for completion performance? | 12:33 |
| ssam2 | there is nothing | 12:33 |
| ssam2 | nothing, nothing, nothing | 12:33 |
| ssam2 | all it does is run `time bst --help` | 12:33 |
| ssam2 | actual tests are a separate task :-) | 12:33 |
| tlater | Heh, gotcha | 12:33 |
| ssam2 | each test is a script that can do anything, and has to output some JSON | 12:33 |
| ssam2 | so everything should be possible | 12:33 |
| * tlater is very curious about how much !256 affected a standardized completion performance step | 12:35 | |
| tlater | This would also be very useful if someone ever sat down to get snappy completion | 12:35 |
| ssam2 | cool, we can add a test | 12:36 |
| ssam2 | i hope to get the automation finished early next week | 12:36 |
| ssam2 | most of the useful stuff will need to wait until the log parser is done | 12:36 |
| ssam2 | but completion could be tested just with with /usr/bin/time | 12:37 |
| ssam2 | -with | 12:37 |
| tlater | Sweet :) | 12:37 |
| tlater | Oooh, this project has an artifact cache | 12:48 |
| ssam2 | i've sent an MR for the benchmarking framework : https://gitlab.com/BuildStream/benchmarks/merge_requests/1 | 12:55 |
| ssam2 | i didn't want to just merge to master of brand new repo and sidestep review altogether | 12:55 |
| ssam2 | however, since it's the first pass and nothing uses it yet, i don't think anyone needs to spend significant effort reviewing it | 12:56 |
| ssam2 | from here i will look at setting up some automation so the benchmark tests can be run as part of our CI pipeline | 12:56 |
| tlater | ssam2: I thought recursive pipelines have landed? I'm getting 'Unexpected key: junction' | 12:57 |
| ssam2 | i thought they landed too, but i haven't tried using the feature from master yet | 12:58 |
| tlater | Hrm, maybe I should just use the commit you checked out | 12:58 |
| ssam2 | something is wrong if it doesn't work from master | 12:58 |
| tlater | ssam2: nvm, I think I just forgot to install the version of buildstream I checked out -.- | 13:06 |
| ssam2 | that's good news | 13:16 |
| jmac | Python's getpass.getuser() returns `tomjon` in my build sandbox. Does that name mean anything to anyone? | 13:19 |
| ssam2 | it's a placeholder | 13:19 |
| ssam2 | dating back at least 3 build tools: http://git.baserock.org/cgit/baserock/baserock/morph.git/tree/morphlib/buildenvironment.py#n49 | 13:19 |
| jmac | Right, just wondered if it was a BuildStream thing or something getpass was doing | 13:20 |
| ssam2 | it comes from here in buildstream: https://gitlab.com/BuildStream/buildstream/blob/master/buildstream/data/projectconfig.yaml#L82 | 13:21 |
| *** skullman has joined #buildstream | 13:26 | |
| gitlab-br-bot | buildstream: merge request (image-authoring->master: WIP: Image authoring documentation) #262 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/262 | 13:58 |
| gitlab-br-bot | buildstream: merge request (image-authoring->master: WIP: Image authoring documentation) #262 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/262 | 13:59 |
| gitlab-br-bot | buildstream: merge request (image-authoring->master: WIP: Image authoring documentation) #262 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/262 | 13:59 |
| *** dominic has joined #buildstream | 14:40 | |
| ssam2 | hi dominic | 14:42 |
| dominic | o/ | 14:42 |
| ssam2 | i'm going to set up the benchmarks automation in the benchmarks.git repo for now instead of buildstream.git | 14:49 |
| ssam2 | partly to avoid complicating the already crowded buildstream/.gitlab-ci.uml | 14:49 |
| ssam2 | *yml | 14:49 |
| ssam2 | and partly because the benchmarks test multiple versions of buildstream, so it's kinda of confusing having the CI think it's testing a specific version | 14:49 |
| ssam2 | we could later set up a job which tests every candidate branch of buildstream against older versions to check for regressions | 14:50 |
| ssam2 | but i don't think we'll have the resources to do that on just a single laptop | 14:50 |
| ssam2 | so i'll just have it run the benchmarks overnight every night or something | 14:50 |
| nexus | o/ hi dominic | 14:50 |
| ssam2 | we will be able to spot regressions in master that way anyway | 14:50 |
| *** mcatanzaro has joined #buildstream | 14:58 | |
| *** csoriano has quit IRC | 15:21 | |
| *** csoriano has joined #buildstream | 15:22 | |
| *** ruben has joined #buildstream | 16:05 | |
| ssam2 | a successful benchmark run! https://gitlab.com/BuildStream/benchmarks/-/jobs/51900502 | 16:08 |
| ssam2 | it's a bit of a pain to view the results, you have to click the artifacts 'Download' button, which gives you a .zip, then you can see the .json file | 16:08 |
| ssam2 | postprocessing them neatly is a separate work item though | 16:08 |
| ssam2 | i guess we could dump them into the build log for now | 16:09 |
| ssam2 | there we go, the results are a bit easier to view for now https://gitlab.com/BuildStream/benchmarks/-/jobs/51901526 | 16:13 |
| persia | Nice, and clear gains over the recent timeframe. | 16:14 |
| *** aday has quit IRC | 16:16 | |
| *** aday has joined #buildstream | 16:17 | |
| *** aday has quit IRC | 16:21 | |
| tlater | ssam2: It seems that the freedesktop-sdk image still builds | 16:21 |
| ssam2 | ok, great | 16:21 |
| *** aday has joined #buildstream | 16:22 | |
| ssam2 | can you note that on the gitlab issue ? | 16:22 |
| tlater | Sure :) | 16:22 |
| ssam2 | i will then give my approval :-) | 16:23 |
| * tlater just makes sure to document exactly what he did | 16:27 | |
| *** toscalix has quit IRC | 16:28 | |
| *** toscalix has joined #buildstream | 16:33 | |
| gitlab-br-bot | buildstream: merge request (image-authoring->master: WIP: Image authoring documentation) #262 changed state ("opened"): https://gitlab.com/BuildStream/buildstream/merge_requests/262 | 16:33 |
| tlater | ssam2: I also simplified the busybox image | 16:34 |
| tlater | ta for that review btw, I completely missed it yesterday :) | 16:34 |
| tlater | *element | 16:34 |
| ssam2 | ah great | 16:39 |
| ssam2 | in return, maybe you could review https://gitlab.com/BuildStream/bst-external/merge_requests/16 ? :-) | 16:39 |
| *** adds68 has quit IRC | 17:02 | |
| *** adds68 has joined #buildstream | 17:02 | |
| *** ssam2 has quit IRC | 17:15 | |
| jjardon[m] | juergbi: tristan FYI, trying to use linux32 in aarch64 returns an error: "linux32: failed to set personality to linux32: Invalid argument" | 17:15 |
| persia | That's not consistently true over all hardware (but I have seen reports of that issue in environments known to work in other channels in the last week, so it seems that a kernel with all the necessary personality support still can't do linux32 on some hardware) | 17:18 |
| juergbi | jjardon[m]: CONFIG_COMPAT is enabled in the kernel? | 17:21 |
| juergbi | and the hardware actually supports aarch32? | 17:21 |
| juergbi | if this is in a VM, not sure whether aarch32 needs to be also enabled in the hypervisor | 17:23 |
| *** ruben has quit IRC | 17:26 | |
| *** dominic has quit IRC | 17:29 | |
| jjardon[m] | ITs an instance from http://scaleway.com/ | 17:30 |
| jjardon[m] | juergbi: cant check that as /proc/config.gz doesnt exist :/ Is there any other way to check the kernel options? | 17:31 |
| juergbi | SYSCALL_DEFINE1(arm64_personality, unsigned int, personality) | 17:34 |
| juergbi | { | 17:34 |
| juergbi | if (personality(personality) == PER_LINUX32 && | 17:34 |
| juergbi | !system_supports_32bit_el0()) | 17:34 |
| juergbi | return -EINVAL; | 17:34 |
| juergbi | as you get EINVAL, i guess system_supports_32bit_el0 returns false | 17:34 |
| juergbi | system_supports_32bit_el0 appears to directly check the hardware capabilities | 17:35 |
| juergbi | so my guess is the CPU doesn't support it | 17:35 |
| persia | Or, more likely, the hypervisor capabilities from a cloud provider | 17:35 |
| juergbi | indeed. not sure how wide-spread CPUs without physical 32-bit EL0 support are | 17:37 |
| *** toscalix has quit IRC | 17:44 | |
| jjardon[m] | ok, thanks. I think we have to accelerate our plan to switch to other aarch64 servers: https://gitlab.com/freedesktop-sdk/freedesktop-sdk/issues/43 | 17:44 |
| persia | Yeah. Either Moonshot or WorksonARM is the right place to run those. | 17:50 |
| *** tiago has quit IRC | 18:08 | |
| *** valentind has joined #buildstream | 18:50 | |
| *** ernestask has quit IRC | 19:52 | |
| *** jonathanmaw has quit IRC | 21:16 | |
| *** cs_shadow has quit IRC | 22:01 | |
| *** aday has quit IRC | 22:01 | |
| *** valentind has quit IRC | 23:29 | |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!