<srhb>
(I am proposing either disallowing the use of haskellPackages.override and haskellPackages.extend in-tree, or fundamentally making extend near-impossible to use for nixpkgs users.)
timokau has joined #nixos-dev
<Profpatsch>
srhb: What do you propose instead?
<Profpatsch>
I’m aware of .override not composing, but is there an alternative?
worldofpeace has quit [Remote host closed the connection]
vaibhavsagar has quit [Write error: Connection reset by peer]
rycee has quit [Remote host closed the connection]
philipp[m] has quit [Remote host closed the connection]
yegortimoshenko has quit [Remote host closed the connection]
Ericson2314 has quit [Remote host closed the connection]
sphalerit has quit [Read error: Connection reset by peer]
roberth has quit [Remote host closed the connection]
nocent has quit [Read error: Connection reset by peer]
bennofs[m] has quit [Remote host closed the connection]
matthewbauer[m] has quit [Read error: Connection reset by peer]
timokau[m] has quit [Remote host closed the connection]
dtz has quit [Remote host closed the connection]
codyopel has quit [Write error: Connection reset by peer]
thefloweringash has quit [Read error: Connection reset by peer]
<srhb>
Profpatsch: You can compose override. I'm OK with either of my suggestions there, really. I think we should remove extend if it's not going to be fixed, really. That's better than making in-tree changes that break the use of it.
<zimbatm>
the channel generation is taking 8G+ of memory, which triggered some OOM on our release box
drakonis_ has joined #nixos-dev
drakonis has quit [Ping timeout: 246 seconds]
ckauhaus has quit [Quit: WeeChat 2.2]
drakonis has joined #nixos-dev
lopsided98_ has joined #nixos-dev
lopsided98 has quit [Ping timeout: 252 seconds]
<infinisil>
srhb: So every time somebody wants to change something in haskellPackages, they have to use the boilerplate `.override (old: { overrides = lib.composeExtensions (old.overrides or (_: _: {})) (self: super: {` ?
<infinisil>
:/
<clever>
infinisil: which is why i dont use that method, i just use what was it called ...
<infinisil>
.extend?
<infinisil>
Apparently that's broken, though I don't know why
<clever>
yeah, extend
<clever>
extend works, but it strips off .overrride, so you cant change the ghc version
<clever>
you must stack things in the right order, (hspkgs.override { ghc = ..; }).extend ...
<infinisil>
I think we should just redesign haskellPackages such that extend can also change ghc and the other things then
<srhb>
infinisil: Yes, I agree it's really painful..
<srhb>
infinisil: I don't know what a good solution is right now though. Using extend internally may be Good Enough.
<srhb>
(As long as we don't export that extend-modified package set to the user)
<clever>
i was also caught by surprise by this a while back
<clever>
was helping a user extend the kernel with //
<clever>
but behind the scenes, nixos always does a .extend of the kernelPackages!
<clever>
which undid //
<srhb>
Yeah..
drakonis1 has joined #nixos-dev
drakonis_ has quit [Ping timeout: 252 seconds]
JosW has joined #nixos-dev
Melkor333 has joined #nixos-dev
<Profpatsch>
Damn the expression problem!
<Melkor333>
Can somebody help me with the 56940? The package pylast has been updated to version 3.0. But Version 3 of pylast doesn't support python 2.7 anymore, thus it breaks mopidy-iris, which builds with python2.7. I don't know if (and how) I should split the pylast package to build differently for python2.7 and python3.x
<Melkor333>
The simpler approach would be to revert the commit which updated pylast to version 3.0, but I don't know if that is something we want to do...
<Melkor333>
(as a sidenote, this would fix at least one Hydra failure for the release 19.3 :) )