<gchristensen>
interesting, it seems the hydra queue runner can't actually keep all the different archs busy at once
<gchristensen>
am I misunderstanding?
sir_guy_carleton has joined #nixos-dev
pie___ has quit [Ping timeout: 260 seconds]
<dtz>
lol why can't I run any NixOS test >:|
* dtz
checks hydra
<dtz>
oh, is everything on fire? lol :P
<gchristensen>
is it? I don't think it is
<gchristensen>
I dont know it to be*
<dtz>
lol oh okay
<dtz>
when went to see hydra the first one I looked at said hydra-eval-jobs was killed with signal 9
<dtz>
so I thought I'd take the hint :3
<dtz>
and FWIW tests not "Working" is I'm seeing invalid reference from extra-utils, this is on staging. Will poke a bit more....
* dtz
submits all the mass-rebuild PR's to staging muahaha
<dtz>
lmao but seriously: bind, utillinux, busybox, lua, curl
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 240 seconds]
lassulus_ is now known as lassulus
<dtz>
well maybe hydra can't keep everything busy because everything is failing :P :P
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-dev
sir_guy_carleton has quit [Quit: WeeChat 2.0]
orivej has quit [Ping timeout: 244 seconds]
MichaelRaskin has quit [Quit: MichaelRaskin]
Lisanna has quit [Remote host closed the connection]
ciil has quit [Quit: Lost terminal]
ckauhaus has joined #nixos-dev
FRidh has joined #nixos-dev
ciil has joined #nixos-dev
goibhniu has joined #nixos-dev
<peti>
domenkozar: We should change the generic Haskell builder so that no Haskell packages ever show up in buildInputs. When I wrote that code originally, adding Haskell libraries there was a convenient hack to make them propagate to other builds that depend on the package, but in fact that choice makes zero sense since the generic builder cannot use any of those packages for anything. All it does it create huge
<peti>
search paths that exceed ARG_MAX. Instead of doing the propagation at run-time, like the generic builder does, we should do it at evaluation time using lib.closePropagation. Then our build script just knows all the paths it needs to know, but the enviroment remains empty (except for build-tools, maybe).
<FRidh>
peti: would it make sense to use another attribute for these type of inputs than `buildInputs`? With Python I'm looking at creating a runtimeDependencies attribute to replace `propagatedBuildInputs` which we now abuse.
<FRidh>
in this case runtimeDependencies is only for Python packages that the package depends on
<FRidh>
(could also use pythonPath again though...)
capisce has quit [Read error: No route to host]
capisce has joined #nixos-dev
<LnL>
what about a more consistent name like pythonInputs? :)
<FRidh>
fine as well
<peti>
FRidh: Well, the thing is that the generic Nix builder has no use for that information. The generic Haskell builder does, but Nix has no clue what to do with a Haskell package. So I'm not sure what we'd gain by introducing a special attribute in the generic Nix builder? What should be in its purpose?
<FRidh>
peti: I did not mean the generic Nix builder, just the Haskell builder. But my point was to distinguish between any build-time dependency `buildInputs` and specifically haskell packages
<FRidh>
In Python we used lib.closePropagation, but that wasn't working always so well because it could not distinguish between the packages of interest (Python packages in propagatedBuildInputs) and other packages.
<FRidh>
python.buildEnv specifically
<peti>
FRidh: Yes, Haskell packages should be moved into a seperate place in the generic Haskell builder -- away from buildInputs. Distinguishing between run-time and build-time dependencies is not a problem in Haskell, though, because (a) it's a compiled language and (b) Cabal files declare those kind of dependencies very accurately.
goibhniu has quit [Ping timeout: 264 seconds]
Lisanna has joined #nixos-dev
goibhniu has joined #nixos-dev
<teto1>
I am trying to use parts of the module backend to improve the kernel config system. I am calling mkMerge on 2 sets { anser= "y"; } and { optional = true; }. This returns: { _type = "merge"; contents = [ { answer = "y"; } { optional = true; } ]; } but can I convert it to { answer="y"; optional=true; } ? (I have introduced a kernelConfigItem = types.submodule { answer/optional } submodule. )
<LnL>
reference config.foo.bar to get the actual merged value
<teto1>
LnL: I don't define any config={}, it's a kind of non-standard use of the modules system. I wanted to reuse just parts of the machinery like mkMerge instead of recreating my machinery.
<LnL>
I'm confused, lib/kernel.nix isn't used by the module system
<teto1>
LnL: To sum up how to get the merged value of 2 types.submodule without relying on the traditional config.foo.bar ?
<domenkozar>
peti: I agree in principle, but I'd really like to unblock macos builds since it's broken for a month and fix this properly with calm discussion
<teto1>
LnL: think I found it, I have to call type.merge.
<LnL>
I'm still confused what you're trying to do, lib/kernel.nix isn't a nixos module
woffs has quit [Quit: Gateway shutdown]
woffs has joined #nixos-dev
orivej has joined #nixos-dev
goibhniu has quit [Ping timeout: 256 seconds]
orivej has quit [Ping timeout: 264 seconds]
<peti>
domenkozar: I am worried about the notion of hacks being committed to the code that we'll clean up "later", especially if those hacks don't really fix the issue they're supposed to fix while causing new issues on top of the old ones.
__Sander__ has joined #nixos-dev
genesis has quit [Ping timeout: 256 seconds]
genesis has joined #nixos-dev
goibhniu has joined #nixos-dev
Mic92 has quit [Ping timeout: 245 seconds]
Mic92 has joined #nixos-dev
ckauhaus has joined #nixos-dev
pie_ has joined #nixos-dev
Sonarpulse has joined #nixos-dev
sir_guy_carleton has joined #nixos-dev
<LnL>
Sonarpulse: ping
__Sander__ has quit [Quit: Konversation terminated!]
goibhniu has quit [Ping timeout: 268 seconds]
pie_ has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-dev
Sonarpulse has quit [Ping timeout: 276 seconds]
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-dev
<domenkozar>
peti: I agree about strictDeps as a potential hack, but removing $out/lib is not a hack since we'll have it as separate output anyway hopefully soon
<domenkozar>
and it finally allows us to build haskell packages for macos again
sir_guy_carleton has quit [Quit: WeeChat 2.0]
Sonarpulse has joined #nixos-dev
FRidh has quit [Quit: Konversation terminated!]
orivej has quit [Ping timeout: 240 seconds]
ckauhaus has quit [Quit: Leaving.]
infinisil has quit [Quit: Configuring ZNC, sorry for the joins/quits!]