gchristensen changed the topic of #nixos-borg to: https://www.patreon.com/ofborg https://monitoring.nix.ci/dashboard/db/ofborg?refresh=10s&orgId=1&from=now-1h&to=now "I get to skip reviewing the PHP code and just wait until it is rewritten in something sane, like POSIX shell. || https://logs.nix.samueldr.com/nixos-borg
disasm has joined #nixos-borg
orivej has joined #nixos-borg
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-borg
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-borg
<LnL> huh, is something wrong?
<gchristensen> dunno
<gchristensen> looking
<LnL> there's an eval that just started
<gchristensen> In the original comment there two spaces between the username and the buildcommand, which is apparently a bug in ofborg. However I deleted the extra space and it seems to be working: https://logs.nix.ci/?attempt_id=3560f106-d074-4a29-85fa-8fcca91f6a4c&key=nixos%2Fnixpkgs.45967
<LnL> heh
<LnL> "syntax error"?
<LnL> that gives me an idea...
<gchristensen> ooh?
<LnL> what if we implement a ping command, people could use that as a sanity check to see if something is wrong
<gchristensen> that'd be cool
<gchristensen> also reacting to posts would be very nice
<gchristensen> a "confused" to indicate a syntax error for example
<LnL> hah, we could do that :)
<gchristensen> also two spaces should be supported I think
<LnL> yeah, but there are other cases like a typo in build and maybe also no valid attributes
<gchristensen> totally
<gchristensen> LnL: I had an idea for a fairly novel ofborg checker, which is it just tries to build the manual for every PR
<gchristensen> on a really big system, to handle mass rebuilds
<gchristensen> and it only posts status updates if it finishes
<gchristensen> (finish: failure or success)
<LnL> hmm
<LnL> what about --option timeout 10m (or whatever) and only report if it's not a timeout?
<samueldr> could the manual be compiled from a known-good nixpkgs, where the proper directories are transplanted? may be important since the site builds using a channel IIRC
<gchristensen> no
<samueldr> (yeah, it wouldn't represent the reality)
<gchristensen> and may just not work, if the structure changes or packages change
<gchristensen> I very badly wanted that to work :) but I also want it to do the nixos manual
<LnL> I think timeout + re-queue makes much more sense
<gchristensen> ooh!
<gchristensen> I could apply TAGS
<LnL> wasn't it you that sent me that variable rebuild scheduling post/paper?
<gchristensen> hiyeah!
<gchristensen> a queue for unknown duration with a low timeout -- most jobs, timeout -> pushed to a longer queue
<LnL> exactly, and this also works for regular builds
<gchristensen> have 3 buckets, one for <1min, one for <5min, one for >5min
<gchristensen> it can even be on the same machine, with appropriate -j settings
<gchristensen> LnL: that is brilliant!
<LnL> lol, it's not my idea
<gchristensen> applying it is!
<LnL> and we could do something like if the "large" bucket has more than x items cancel something and post a timeout
<gchristensen> yea
<LnL> that way you utilise idle resources
<gchristensen> I need to focus on the task at hand first, but then this is definitely a thing I want to do
<LnL> if the _only_ thing building is a mass rebuild, let it do it's thing
<gchristensen> task at hand: cleaning up massrebuild.rs X)
<gchristensen> yes that is so cool
<LnL> oh! :o
<LnL> I know that file
<gchristensen> all 855 lines of that file
<LnL> nope, most definitively not
<gchristensen> hah
<gchristensen> it is so long I get lost in it :(
<LnL> nah, that's nothing
<gchristensen> lol
<LnL> that work project I've been working has a 3k file
<gchristensen> no
<gchristensen> lol
<gchristensen> I'm making a generic EvaluationTask which has a before_merge() and an after_merge()
<gchristensen> and from there I think it'd be nice / feasible to break out each of these things in to a cli you can call locally
* LnL wants
zimbatm has quit [*.net *.split]
<LnL> let me know if there's something I can help with
<LnL> btw, the me message tagging is still open I think
<gchristensen> me message tagging?
<gchristensen> help: yes! definitely! would you be open to implementing my maintainer ping, currently in nix + a shell script, in to massrebuilder.rs?
<{^_^}> ofborg#218 (by LnL7, 3 weeks ago, open): buildresult: replace serialization with an enum to simplify compatibility
<gchristensen> oohh
<gchristensen> yeah
<gchristensen> ack
<gchristensen> mind resolving the conflicts and we'll merge?
<LnL> nothing depends on it, but might be nice to get the tagged messages in place for when we do
<gchristensen> yeah
<LnL> oh sure
<gchristensen> this code is complicated
<gchristensen> this code looks like it was written in PHP and ported to Rust at a hackathon in a city far away from here
<gchristensen> and then expanded over a year
<LnL> is tasks::githubcommentposter::tests::test_failing_build and friends new?
<gchristensen> don't ... think so ..?
<gchristensen> one sec
<gchristensen> don't think so
<gchristensen> wait
<gchristensen> yes
<gchristensen> let me clarify
<gchristensen> I'm not familiar with that test being broken
<gchristensen> and I am surprised that it is
<LnL> oh, request ids in the comments
<gchristensen> I ran the tests though?
<gchristensen> in the ofborg shell env, I don't get color coded compiler failures
<LnL> nah, all good
<gchristensen> do you get colorful output?
<gchristensen> oh locale stuff probably
<LnL> pushed
<LnL> the last commit is the most dangerous one, but I think it's fine if builders don't update before a deploy
<gchristensen> ok
<gchristensen> well I know how to make sure we deploy quickly
<LnL> ah, no the inputs are the dangerous part
<LnL> maybe we could do something similar there and throw stuff back into the queue if it's an unknown format?
<gchristensen> so what is the dangerous part?
<gchristensen> like we'll be emitting stuff older builders can't parse?
<{^_^}> [ofborg] @grahamc pushed 3 commits to move-mass-rebuilder: https://git.io/fA4sa
<LnL> yeah, so this versions the results to make changing them more clear
<LnL> but the build inputs could also change
<gchristensen> ahh
<{^_^}> [ofborg] @grahamc opened pull request #234 → Move mass rebuilder → https://git.io/fA4sr
<LnL> and the dangerous part about the last commit is that it enables the newly introduced format
<LnL> but I think legacy and v1 are compatible
<gchristensen> right
<LnL> yeah, v1 is just stricter and adds the tag which gets ignored
<LnL> so v1 -> current should be fine
<gchristensen> ok
<LnL> types are nice...
<gchristensen> yeah :)
<{^_^}> [ofborg] @grahamc pushed to move-mass-rebuilder « Fixup: add mod »: https://git.io/fA4sd
<gchristensen> wtf why does this build locally
<gchristensen> oooh cargo test
<LnL> what else?
<{^_^}> [ofborg] @grahamc pushed to move-mass-rebuilder « fixup tests »:
<gchristensen> I like that due to type checking this will definitely work when deployed https://github.com/NixOS/ofborg/pull/234
<{^_^}> ofborg#234 (by grahamc, 16 minutes ago, open): Move mass rebuilder
<gchristensen> I guess thaht is more of a feature of being compiled
<gchristensen> LnL: ok I'm going to deploy your PR, k?
<LnL> yeah, unless you want to finish your thing first
<LnL> might be conflicts
<gchristensen> ah
<gchristensen> I'll merge & deploy it in about 10min
<gchristensen> brb
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-borg
<gchristensen> this is very very annoying
<{^_^}> [ofborg] @grahamc pushed to move-mass-rebuilder « fixup test impurity »: https://git.io/fA4GC
<gchristensen> aoeuaoeuoeau
<LnL> :p
<LnL> travis not happy?
<gchristensen> yeah
<gchristensen> locale issues
<LnL> huh, how did you end up there
<LnL> oh, glibc
<LnL> is that env variable still required?
<gchristensen> guess so
<gchristensen> I need to take a break and get off the copmuter / get away from this for a few hours
<LnL> :(
<gchristensen> back later :)
<LnL> btw, I'm fine with replacing travis if it's annoying
<LnL> was just the easiest to setup at the time
<{^_^}> [ofborg] @grahamc pushed to move-mass-rebuilder « travis-ci: use nix 2.1 »: https://git.io/fA4GN
<gchristensen> ok I'm out, see you :)
<LnL> :)
<gchristensen> (I didn't delete it)
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-borg
orivej has quit [Ping timeout: 244 seconds]