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
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 246 seconds]
jtojnar has joined #nixos-borg
orivej has joined #nixos-borg
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-borg
<{^_^}> [ofborg] @grahamc pushed to eval-darwin « Test darwin-tested during evaluation phase »: https://git.io/fjJ9e
<{^_^}> [ofborg] @grahamc opened pull request #339 → Test darwin-tested during evaluation phase → https://git.io/fjJ9f
<{^_^}> [ofborg] @grahamc pushed 2 commits to released: https://git.io/fjJ9s
<{^_^}> [ofborg] @grahamc merged pull request #339 → Test darwin-tested during evaluation phase → https://git.io/fjJ9f
<{^_^}> [ofborg] @grahamc pushed 0 commits to eval-darwin: https://git.io/fjJ9G
<samueldr> gchristensen: lazy-me asks: does ofborg catch tarball build failures?
* samueldr checks meanwhile
<samueldr> it nix-instantiates it, so I'd assume yes
<gchristensen> it should
<gchristensen> eval failures, anyway
<gchristensen> it doesn't build it
* samueldr is writing a "how can we make sure this won't happen" comment for something
<samueldr> won't happen again*
<gchristensen> cool
<samueldr> though I think it might require tarball to be built :/
<samueldr> any reasons it would be problematic to build in addition to eval'ing?
<samueldr> (it didn't fail in evaluation locally, currently building, and the hydra logs are basically screaming it was during build)
<gchristensen> we don't build anything by default
<gchristensen> I'd _like_ to, but ofborg's deployment model doesn't really support that
<samueldr> yeah, then "if it would have been opened as a PR, waited for ofborg, then self-merged we wouldn't be having this issue" wouldn't :/
<gchristensen> right
<gchristensen> wait
<gchristensen> I don't understand
<samueldr> (sorry if it feels negative, not against you, just not helping this case)
<samueldr> was pushed directly to our stable branch
<gchristensen> ah
<gchristensen> and ofborg would/would-not have caught it?
<samueldr> and here, if it was opened as a PR and self-merged, it wouldn't have failed since instantiate wouldn't have picked up on it
<gchristensen> ah
<gchristensen> yeah, it is tough :/
<samueldr> would we want to rely on an additional out-of-ofborg step to make tarball builds for every push?
<samueldr> don't know if github with actions would like it to see the amount of work on nixpkgs on their infra :)
<gchristensen> right
<gchristensen> I think a better bet is make ofborg able to build every PR
<samueldr> sure, thinking in a "until we can" way
<gchristensen> I wouldn't want to do that, no
<gchristensen> those green boxes are sacred
<gchristensen> a trouble with building anything is mass rebuilds cause red boxes
<samueldr> I'm thinking only tarball, but then I'm now understanding that tarball builds need stdenv
<gchristensen> it happens to be ofborg-building-untrusted-stuff has been my side project for a bit now
* samueldr just realised he was instantiating on the local machine instead of the kitchen workstation
<gchristensen> "kitchen workstation" :D
<samueldr> fan noise away from my desk!
<samueldr> dining room were the words I was thinking of
<MichaelRaskin> I think for self-merged PRs (and for other stuff) conditional/delayed merges by ofborg would be great
<gchristensen> +1
<MichaelRaskin> Unlike untrusted builds, I probably won't be able to find a hole in that…
<MichaelRaskin> (Is ofBorg testing being easier with FOSS repo a good enough reason to just move to Gitlab?)
<gchristensen> I'm counting on you to!
<gchristensen> what do you mean?
<MichaelRaskin> Well, you cannot test merge functionality without a test repo on GitHub or something like that
<gchristensen> ah
<MichaelRaskin> Yeah, this can still be fixed-output derivation for nix-managed tests, but still annoying
<MichaelRaskin> (because credentials, if nothing else)
* samueldr wonders if that specific failure could be caught simply with eval
<gchristensen> there are other things we can do, too
<gchristensen> like part of the problem here is probably surprise
<gchristensen> "oh, it's been 1 week why is 18.09 busted... oh..."
<samueldr> that's part of it
<samueldr> but just found it!
<samueldr> the part that's failing in the tarball build is `nix-instantiate --eval --strict --show-trace ./maintainers/scripts/eval-release.nix`
<gchristensen> oh cool
<samueldr> I'm guessing this isn't onerous to add as a step for ofborg?
<gchristensen> what does it eval?
<gchristensen> (probably not no)
<samueldr> sounds like the main difference from simple nix-instantiate of release.nix is recursing?
<gchristensen> we don't do this already?
<samueldr> not sure
<samueldr> from outside of ofborg, what's the best way to eval like that?
<samueldr> (want to see if ofborg would have caught the issue)
<samueldr> <3 documentation
<gchristensen> :)
<samueldr> gchristensen: right, this would have caught it <3
<gchristensen> nice =)
<LnL> I'd expect so, the eval checks should be equivalent to what's run inside the tarball build
<samueldr> I looked at the list of instantiate commands in the output, while skipping the ^.^ check since it wasn't obvious how to run it
<samueldr> I was thinking of opening a PR to rename ^.^ but didn't know what to
<gchristensen> so I've given up on trying to make a very minimal kernel for ofborg
<gchristensen> builds
<gchristensen> but if anyone has favorite options to turn off, pass them along :P
<gchristensen> I'm planning on going through and doing a mild turning-off through menuconfig and starting from there.
<{^_^}> [ofborg] @veprbl opened pull request #340 → Enable tagging for TeX PRs → https://git.io/fjJHq
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-borg
<{^_^}> [ofborg] @grahamc merged pull request #337 → config.extra-known-users.json: Add myself (balsoft) → https://git.io/fjJGV
<{^_^}> [ofborg] @grahamc pushed 2 commits to released: https://git.io/fjJHO
<gchristensen> ok so maybe having evaluation processes would be next, so evaluations are parallelized
jtojnar has quit [Remote host closed the connection]
jtojnar has joined #nixos-borg
jtojnar has quit [Ping timeout: 250 seconds]
<samueldr> (I was mainly curious about if it was possible, not if I should)
<samueldr> running the step could be as simple as ~/Projects/ofborg/src/outpaths.nix while CWD is nixpkgs
<gchristensen> it isn't clear to me that I should hate it at all :)
<samueldr> running a .nix thing directly!
jtojnar has joined #nixos-borg
<gchristensen> oh!
<samueldr> I've never seen that done
<gchristensen> LOL
<gchristensen> that is hilarious!
<gchristensen> shipit
<{^_^}> [ofborg] @samueldr opened pull request #341 → outpaths.nix: Makes callable → https://git.io/fjJHr
jtojnar has quit [Read error: Connection reset by peer]
orivej has quit [Ping timeout: 246 seconds]
orivej_ has joined #nixos-borg
jtojnar has joined #nixos-borg
tilpner has quit [Quit: WeeChat 2.4]
tilpner has joined #nixos-borg