<cole-h>
^ despite how simple it looks, this one carries a decent amount of baggage...
<cole-h>
It kinda leaves me wondering if it would even be desirable to auto-label package updates...
orivej has quit [Ping timeout: 272 seconds]
orivej_ has joined #nixos-borg
orivej_ has quit [Ping timeout: 272 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 240 seconds]
orivej_ has joined #nixos-borg
orivej_ has quit [Ping timeout: 272 seconds]
orivej has joined #nixos-borg
orivej_ has joined #nixos-borg
orivej has quit [Ping timeout: 264 seconds]
orivej_ has quit [Ping timeout: 240 seconds]
hmpffff has joined #nixos-borg
orivej has joined #nixos-borg
orivej_ has joined #nixos-borg
orivej has quit [Ping timeout: 258 seconds]
orivej_ has quit [Ping timeout: 264 seconds]
orivej has joined #nixos-borg
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 256 seconds]
orivej_ has joined #nixos-borg
hmpffff_ has joined #nixos-borg
hmpffff has quit [Ping timeout: 246 seconds]
orivej_ has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-borg
hmpffff_ has quit [Quit: nchrrrr…]
orivej has quit [Ping timeout: 265 seconds]
orivej has joined #nixos-borg
hmpffff has joined #nixos-borg
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos-borg
cole-h has quit [Quit: Goodbye]
hmpffff_ has joined #nixos-borg
hmpffff has quit [Ping timeout: 256 seconds]
WilliButz has quit [Quit: bye]
WilliButz has joined #nixos-borg
hmpffff_ has quit [Quit: nchrrrr…]
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos-borg
hmpffff has joined #nixos-borg
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej_ has joined #nixos-borg
orivej has quit [Ping timeout: 260 seconds]
orivej_ has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 264 seconds]
orivej has joined #nixos-borg
hmpffff has quit [Quit: nchrrrr…]
orivej has quit [Read error: Connection reset by peer]
orivej_ has joined #nixos-borg
orivej_ has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 272 seconds]
orivej_ has joined #nixos-borg
orivej_ has quit [Ping timeout: 265 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-borg
hmpffff has joined #nixos-borg
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 258 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 256 seconds]
orivej_ has joined #nixos-borg
orivej_ has quit [Ping timeout: 258 seconds]
orivej has joined #nixos-borg
orivej_ has joined #nixos-borg
orivej has quit [Ping timeout: 272 seconds]
cole-h has joined #nixos-borg
orivej has joined #nixos-borg
orivej_ has quit [Ping timeout: 256 seconds]
orivej has quit [Ping timeout: 256 seconds]
orivej has joined #nixos-borg
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-borg
<cole-h>
Is there a reason ofborg uses `nix-build ./nixos/release.nix -A tests.test` rather than `nix-build . -A nixosTests.test`?
<cole-h>
Using `-A nixosTests.test` would solve ofborg#504 AFAICT -- `nix-build ./nixos/release.nix -A tests.hostname` does jack, but `nix-build . -A nixosTests.hostname` runs both tests just fine.
<LnL>
nixos is already very tightly coupled to nixpkgs, but now the reverse is also true which is bad IMHO
<LnL>
while it's in the same repo for convenience nothing in nixpkgs should reference nixos
<LnL>
same with lib -> nixpkgs
* LnL
hopes this will become better again with flakes
orivej has quit [Ping timeout: 256 seconds]
orivej_ has joined #nixos-borg
<LnL>
that said, it's a thing now so we might as well use it
<MichaelRaskin>
Hmm, what about tests?
<MichaelRaskin>
Given we are effectively not allowed to have NixOS-independent GUI package tests…
<LnL>
there are some pure nixpkgs tests, like lib or the cc-wrapper stuff
<MichaelRaskin>
Some, but for Chromium one in principle wants to refer to NixOS Chromium test
<LnL>
that's fine but then it's a nixos test
<MichaelRaskin>
But if I ask «does Chromium still work after this change», I do want to have a reference to them
<LnL>
doesn't mean chromium needs to magically point to another project
<MichaelRaskin>
Are you willing to write an RFC about that?
<MichaelRaskin>
I would actually be very happy with the claim that NixOS tests must not be used for Nixpkgs so Nixpkgs is allowed to have its own tests even if there is some feature intersection in the supporting code.
<LnL>
what would that be in practice?
<MichaelRaskin>
NixOS tests not being an unreasonable pile of overhead would be nice anyway, but there is no a chance of it happenning
<MichaelRaskin>
there is now a chance
<MichaelRaskin>
Well, you need to launch some X and some D-Bus to test LibreOffice
<LnL>
one thing I've thought about is essentially splitting the tree up while evaluating
<LnL>
eg. evaluating lib without pkgs or nixos in the tree, pkgs without nixos, etc.
<LnL>
as for actual vm tests, I'm probably not really familiar enough with low level linux stuff to know how to deal with that
<MichaelRaskin>
Well, Eelco Dolstra has figured out what is needed to run containers inside the Nix builds
<LnL>
there's also _a lot_ you can do with just a regular build, the reason that the vm tests are easy/nice is because all of the setup/configuration logic is in the modules
<MichaelRaskin>
At that level modules are not worth it, you don't need much
<MichaelRaskin>
But if one can just run a NixOS container inside the build, then sure, why not just run tests there
<LnL>
yeah then we're basically back at nixos vs things like vim_configurable
<cole-h>
Also, tangentially related to the above issue: why are we using `restrict-eval` if the borg user isn't trusted? I think that message appears in every single log I've ever looked at...
<LnL>
that warning wrong, it's an evaluation time setting so only relevant to the client
<LnL>
is*
orivej has joined #nixos-borg
orivej_ has quit [Ping timeout: 264 seconds]
<cole-h>
Oh, OK.
<MichaelRaskin>
LnL: not sure how people not thinking that encapsulated conigurability is good relates to test
orivej has quit [Read error: Connection reset by peer]
<cole-h>
I don't think so -- it was merged ~1 PM my time, and last buildkite was 11 AM my time
<cole-h>
Yeah, last deploy was crate2nix
<LnL>
oh, maybe it was a bit late yesterday
<cole-h>
And we're off to the races
<LnL>
heh, you beat me to it :)
<cole-h>
>:)
<cole-h>
Payback for last time >:P
<cole-h>
Does the ofborg jobset build automagically, or does it need to be kicked off?
<LnL>
you mean hydra for the aarch builder?
<cole-h>
I think so, yeah
<LnL>
that runs periodically, but I can start an eval
<cole-h>
Buildkite deploys to aarch, right?
<LnL>
yeah, forgot how frequently that runs tho
<cole-h>
Oh it's not the ofborg deploy but the nix-community deploy
<cole-h>
I think Graham said it runs ~weekly. Already deployed this morning.
<cole-h>
I can't remember: does aarch *need* the hydra jobset to get the borg update, or does it just use that as a cache so it doesn't have to build it itself?
<LnL>
it uses the last successful build
<cole-h>
Right, OK. Thanks for kicking off the eval.
hmpffff has quit [Quit: nchrrrr…]
<cole-h>
Alright, real deploy begins.
<cole-h>
LnL: Something about the lapin bump doesn't look good.
<cole-h>
thread 'main' panicked at 'cannot run an executor inside another executor', <::std::macros::panic macros>:2:4
<cole-h>
Over and over again in ofborg-evaluator.service
<MichaelRaskin>
LnL: I mean, the stuff where tests are most beneficial is also the stuff where we do need something that can be criticised as feature duplication with NixOS tests
orivej_ has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-borg
<LnL>
true
<cole-h>
LnL: Probably want to kick off another hydra eval as soon as the revert merges, so aarch doesn't get the seemingly-busted lapin (if you hadn't thought about that already)
<LnL>
hrm, very good point
<LnL>
I wonder how we ended up with nested executors tho
<cole-h>
You saw the panic, and that version downgrade fixed it?
<LnL>
yeah panic on released and not with the branch
<cole-h>
If this works, we'll see `Revert "Revert "lapin: 1.0.0-beta4 -> 1.0.0-rc6""` :D
<cole-h>
(hopefully)
<cole-h>
(Would be nice to see that the lapin bump doesn't reintroduce that somehow)
<cole-h>
Not gonna lie, those SSL errors always spook me out. I was watching the evaluator logs saying "Yes! Yes! Yes! *sees SSL errors that happen every time* NO! *sees no backtraces and loads of heartbeats* YES!!!"
<cole-h>
When you have some time, it might be nice to post a little something on that related issue (or open a new one). I ask you because you did all the async stuff :P I haven't dipped my toes in to that ecosystem yet.