supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-dev
rajivr has joined #nixos-dev
<MichaelRaskin>
ekleog: nobody cares about the speed of running a working test, but debugging overhead of NixOS tests is suboptimal
<MichaelRaskin>
VM tests have most of the features, I believe. Except not being a pain to write.
<MichaelRaskin>
I am not sure if that NixOS-container-inside-Nix-build featuer is ever going to be merged… It could help a lot, as its overhead is indeed small.
<ekleog>
MichaelRaskin: yup, I'm mostly concerned about “not being a pain to write” — I kind of wonder whether they're a pain to write only due to the fact that they're awfully slow to boot, or whether there are other things that could be done to make them faster to iterate on
<ekleog>
(and I must plead guiltly to having only written nixos tests for network services yet, and not having even tried to write one for a GUI application)
<MichaelRaskin>
When starting LibreOffice and clicking around a bit and checking at least something happens takes less time than booting the VM to begin with, the operating word is probably not «guilty» but «reasonable»
<MichaelRaskin>
With network tests you at least have some vague idea what you are suffering for…
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
<MichaelRaskin>
Maybe staged tests where VMs are prepared, booted, suspended to storage and then resumed before tbe real work could be better.
<MichaelRaskin>
Of course, writing VM tests only got more annoying since my PR, because some people think that the current mode of formatting checks is acceptable at all (and somehow good)
ajs124 has quit [*.net *.split]
garbas has quit [*.net *.split]
johnny101 has quit [*.net *.split]
jpo has quit [*.net *.split]
grw1 has quit [*.net *.split]
ekleog has quit [*.net *.split]
catern has quit [*.net *.split]
pbogdan has quit [*.net *.split]
Valodim has quit [*.net *.split]
harrow has quit [*.net *.split]
<MichaelRaskin>
(asserting bit-perfect formatting with strict rules about line breaks — neither too few nor too many — that are applied to post-parameter-expansion sources, just to make sure you cannot use the formatter automatically at the edit time)
ekleog has joined #nixos-dev
scott has quit [Quit: Ping timeout (120 seconds)]
scott has joined #nixos-dev
pbogdan has joined #nixos-dev
Valodim has joined #nixos-dev
garbas has joined #nixos-dev
johnny101 has joined #nixos-dev
jpo has joined #nixos-dev
harrow has joined #nixos-dev
ajs124 has joined #nixos-dev
grw1 has joined #nixos-dev
jpo has quit [Ping timeout: 249 seconds]
jpo has joined #nixos-dev
catern has joined #nixos-dev
tokudan has quit [Remote host closed the connection]
tokudan has joined #nixos-dev
janneke_ has joined #nixos-dev
Valodim- has joined #nixos-dev
jlotoski-znc has joined #nixos-dev
harrow` has joined #nixos-dev
energizer_ has joined #nixos-dev
AtnNn_ has joined #nixos-dev
pbogdan_ has joined #nixos-dev
garbas_ has joined #nixos-dev
abathur has quit [*.net *.split]
MichaelRaskin has quit [*.net *.split]
AtnNn has quit [*.net *.split]
evanjs has quit [*.net *.split]
bridge[evilred] has quit [*.net *.split]
janneke has quit [*.net *.split]
energizer has quit [*.net *.split]
delroth has quit [*.net *.split]
ajs124 has quit [*.net *.split]
garbas has quit [*.net *.split]
johnny101 has quit [*.net *.split]
grw1 has quit [*.net *.split]
pbogdan has quit [*.net *.split]
Valodim has quit [*.net *.split]
harrow has quit [*.net *.split]
Valodim- is now known as Valodim
ajs124 has joined #nixos-dev
delroth1 has joined #nixos-dev
evanjs has joined #nixos-dev
abathur has joined #nixos-dev
grw1 has joined #nixos-dev
energizer_ is now known as energizer
mkaito has quit [Quit: WeeChat 3.0]
<jtojnar>
symphorien: it would probably be feasible to construct such hook but not sure how practical it would be
<jtojnar>
closure-size wise (cocci is written in ocaml, IIRC)
hexa- has quit [Quit: WeeChat 2.9]
hexa- has joined #nixos-dev
<supersandro2000>
Would it be insane to exit with 1 if pytest/setuptools executed 0 tests?
<supersandro2000>
I mean half of the python packages would break...
<energizer>
supersandro2000: pytest exits 5 if it executed zero tests
<supersandro2000>
🤔 this means we could modify the setupHook to exit 1 if it exists 5
<supersandro2000>
something to do in the future I guess
<lovesegfault>
It's more motivated by upset at having navigated the source of bs1770gain
<lovesegfault>
now I just want it gone
<lovesegfault>
saw your suggestion just now; I thought it was recommended to justify why a package was yanked?
<supersandro2000>
well... you are right but I want to try to find something over which you can't argue
<supersandro2000>
maybe just add that it was unused after beets dropped it
<supersandro2000>
or it is no longer reqired
<supersandro2000>
you can't argue against that 😉
<lovesegfault>
fair enough, on it
<lovesegfault>
bs1770gain = throw "bs1770gain has been removed from nixpkgs, as it had no maintainer or reverse dependencies."; # added 2021-01-02
<supersandro2000>
yeah I think that works
janneke_ is now known as janneke
orivej has joined #nixos-dev
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
FRidh has joined #nixos-dev
<ekleog>
Did anyone see a recent ofborg darwin build happening? I've got a build that's been stuck in the queue since yesterday, and with monitoring.nix.ci down I can't check whether the queue is just really full or whether there's no darwin builder live these days
kalbasit has quit [Ping timeout: 240 seconds]
kalbasit has joined #nixos-dev
<zowoq[m]>
no darwin builders for ofborg for a while now
<lukegb>
which changed the kdeApplications mkDerivation to require pname, which the new derivation didn't have, so... [sigh]
<lukegb>
We need BORS or something :p
<supersandro2000>
that can always happen and we have no way to check other than checking eval if merging older PRs
<supersandro2000>
BORS?
<lukegb>
Effectively a merge train, so instead of having +w people merge directly the bot adds it to a queue, does some tests with the merged state and then pushes
<supersandro2000>
that sounds like something we want
<supersandro2000>
we should probably prioritize that more
<ekleog>
yessssssssss bors
<ekleog>
the problem being, compute power — it'd probably be quite resource-intensive, and integrating it with hydra would probably be hard
<samueldr>
for a good "merge train" approach for Nixpkgs, we probably want it to build through the hydra infra, including sending binaries to the cache, so that once the train is in master, it's already built the `tested` jobset
<samueldr>
instead of having things build needlessly in isolation, again and again
<samueldr>
"but what about $badthing"?
<samueldr>
well, if it's in the merge train, it's slated to go into master, $badthing would have gotten into master anyway
cole-h has joined #nixos-dev
orivej has quit [Ping timeout: 260 seconds]
kalbasit has quit [Ping timeout: 260 seconds]
kalbasit has joined #nixos-dev
__monty__ has quit [Quit: leaving]
<supersandro2000>
I wouldn't build the packages
<supersandro2000>
just evaling should be fine to catch those nasty eval errors that occur when two PRs are merged which are not compatible