<andi->
I am happy to take care of them since the discourse based workflow breaks patches
orivej has joined #nixos-dev
julm has joined #nixos-dev
julm has quit [Client Quit]
julm has joined #nixos-dev
<jtojnar>
andi-: would not it be more suitable for it to go to `defaultPackages` instead of `requiredPackages`
AlwaysLivid has quit [Ping timeout: 240 seconds]
MichaelRaskin has quit [Quit: MichaelRaskin]
LnL has quit [Ping timeout: 258 seconds]
LnL has joined #nixos-dev
LnL has joined #nixos-dev
cole-h has quit [Ping timeout: 256 seconds]
alp has joined #nixos-dev
lopsided98_ has quit [Ping timeout: 260 seconds]
<niksnut>
aminechikhaoui: well, you can already have flake jobsets
alp has quit [Ping timeout: 272 seconds]
saschagrunert has joined #nixos-dev
<rnhmjoj>
i'm trying to make an activation script run before another one, which i don't want to modify directly. is there a way to "override" its deps to specify mine?
<rnhmjoj>
the type of system.activationScripts is "attrsOf unspecified"
<rnhmjoj>
it looks like i'll have to make a PR to change this type
alp has joined #nixos-dev
sgrunert has joined #nixos-dev
saschagrunert has quit [Ping timeout: 256 seconds]
lopsided98 has joined #nixos-dev
<NinjaTrappeur>
niksnut: is there a --list-generations equivalent for nix profile?
<niksnut>
not quite, but 'nix profile diff-closures' does show the generations
<eyJhb>
Not entirely a dev thing, but I am trying to do a nix copy, but it just stucks at 2.1 mb each run. I have tried to verbose it as much as possible, but does not seem to give any meaningful - https://termbin.com/x0uaa . It is a ssh server which is behind a VPN, on a non standard port. I can scp to it without any issues
<eyJhb>
Not sure if this is known. It will just be stuch here forever. `nix copy -vvvvvvvvvvvvvvvv -s --to ssh://nix-deployer@srtoffee-build /nix/store/xj11my95snfjvm1g9c4g8w1s8s20a7dp-rsync-3.1.3 /nix/store/slh0pryq6blabhwvicr8azfsyxfv4hrc-nixos-system-toffee-20.09pre-git /nix/store/l99w581ii4z2ghkjdc9mjka3lbs730m8-switch`, engouh v's to ensure all the debug!
<NinjaTrappeur>
ack
<eyJhb>
ack ?
<NinjaTrappeur>
ack about the answer I got 2 messages above, it was not related to your message, sorry :(
<eyJhb>
NinjaTrappeur: Damn it :( Open for any good answer :p
<NinjaTrappeur>
Apart from setting up gdb, which might be pretty tricky provided all the moving parts (nix/ssh) here, I don't. Sorry :(
<NinjaTrappeur>
I'm sure there's a better way though
<eyJhb>
Hoping I can do it when I come home.. But weird that there is no more output :)
<regnat>
eyJhb: might not help at all, but you can try ssh-ng:// rather than ssh://. Maybe it'll work or at least give more infos
<eyJhb>
Somewhat more helpful, but I don't see why it should not have a valid signature
<eyJhb>
Tried where the to (ssh) is root user, same thing :(
<regnat>
eyJhb: Well unless you explicitely signed it, it won't have any. If the remote side is a trusted-user, you can use --no-require-sigs to bypass it
<eyJhb>
regnat: well, I would assume it to be OK, but I am using Nixus atm. to deploy - https://github.com/Infinisil/nixus . Signing is used each time one generates a package/derivation?
<regnat>
eyJhb: Nah, signing is a manual process. But actually it might be a bug to require signatures with nix copy, I'm not sure 🤔. Is this a stable Nix version or something more recent ?
<eyJhb>
Well, my OS is on the latest unstable channel https://status.nixos.org/ ( 5d0e2dedd559 ), so fairly new but the the newest. I was able to deploy to a machine yesterday on the previous unstable
<eyJhb>
Maybe a little rollback
<eyJhb>
Rollback does not seem to work that well from a nixos-rebuild switch --rollback
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
<regnat>
eyJhb: Nah, unless you installed nixUnstable explicitely you should be on a stable release of Nix. I'm afraid I don't really know how to help you more…
alp has quit [Ping timeout: 272 seconds]
<regnat>
Well you can always work around the problem by setting require-sigs = false in the remote machine's nix.conf file, but that's not really a proper solution
<eyJhb>
regnat: Maybe I will just do that once, and then hope for the best in the future. I haven't really had that issue a lot
rajivr has quit [Quit: Connection closed for inactivity]
alp has joined #nixos-dev
andi- has quit [Remote host closed the connection]
andi- has joined #nixos-dev
<aminechikhaoui>
niksnut right there is flake jobsets, but say I have a project with a default.nix that exposes a couple of arguments enableTests/enableDebug/.. and I used to have multiple jobsets to turn those on and off. With flakes my understanding since it's a pure eval that kind of arguments won't work and have to be defined in flake.nix instead.
<aminechikhaoui>
but why not define the whole hydra project through a flake in that case
rajivr has joined #nixos-dev
alp has quit [Remote host closed the connection]
<gchristensen>
aminechikhaoui: there are people doing pretty clever things with declarative jobsets that I'm not sure flakes would work for
alp has joined #nixos-dev
<gchristensen>
and probably not all declarative jobset users would want to switch to flakes for all their use cases
<aminechikhaoui>
gchristensen yeah perhaps both can co-exist like for jobsets (legacy and flakes)
alp has quit [Ping timeout: 272 seconds]
alp has joined #nixos-dev
<worldofpeace>
Yall today's the day
<worldofpeace>
lol, almost a month delayed.
<aminechikhaoui>
\o/ \o/
<stigo>
:)
<worldofpeace>
I do hope we could align these dates to be more convenient like jonringer suggested on discourse
Jackneill has quit [Ping timeout: 258 seconds]
Jackneill has joined #nixos-dev
<NinjaTrappeur>
nix 3.0 is breaking home-manager which depends on nix-env. I started to conditionally replace the nix-env calls with some nix profile ones, but I am now wondering: can we expect the daemon API to be stable?
<NinjaTrappeur>
IE. is writing external tooling directly talking to the daemon with the current protocol a good idea?
teto has joined #nixos-dev
<andi->
aminechikhaoui: What is the benefit of flakes in that case? Isn't hydra already enforcing pure eval model?
<regnat>
NinjaTrappeur: It should be, yes. But the protocol isn't really nice to work with, so I'd rather not go that way unless I have a really good reason to do it
<regnat>
(besides, the daemon is kind-of an implementation detail as it's not guaranted to be there)
<eyJhb>
worldofpeace: on the 46 of septomber! :D
<regnat>
NinjaTrappeur: wait… nix-env isn't removed in Nix 3.0, is it?
<eyJhb>
Is there a link to what Nix 3 brings to the table?
<regnat>
NinjaTrappeur: Ah, indeed, you can't mix nix-env and nix profile :)
<NinjaTrappeur>
^ once you migrated your profile, there's no way back
<NinjaTrappeur>
(my nix profile is empty)
<regnat>
I didn't know nix profile was usable at all :D
<aminechikhaoui>
andi- here is an example, I have a project repository `foo` that depends on repos bar, baz. I want to use flakes for better multi-repo management, but also `foo` exposes certain arguments { enableDebug, enableTracing, enableTests, variant ? "variant-1", (etc ..) } that we currently use to define multiple jobsets that generate different builds based
<aminechikhaoui>
on those arguments.
__monty__ has joined #nixos-dev
<aminechikhaoui>
so I'd like to both use flakes and also be able to manage those configs in hydra but since nix arguments afaik would prevent pure evaluations it wouldn't make sense in the case of flakes
<aminechikhaoui>
so what I thought of is perhaps a wrapper flake can be defined to declare the jobsets under a certain hydra project
<aminechikhaoui>
and that's where we inherit the foo flake and create various variants based on it
<V>
https://github.com/NixOS/nixpkgs/pull/101127 would anyone be able to take a look at this? it's been hanging around for a bit and I figure it's something a bunch of people would like
<{^_^}>
#101127 (by deviant, 6 days ago, open): nixos/defaults: init
alp has quit [Ping timeout: 272 seconds]
alp has joined #nixos-dev
sgrunert has quit [Remote host closed the connection]
saschagrunert has joined #nixos-dev
<V>
(ping adisbladis infinisil)
<qyliss>
> This covers every single text editor in nixpkgs
<{^_^}>
error: syntax error, unexpected IN, expecting ')', at (string):345:38
<qyliss>
wow!
zarel_ has quit [Ping timeout: 260 seconds]
<V>
if I missed any, they either weren't in GNOME and KDE, or weren't in pkgs/applications/editors
<V>
But I tested every single one I could find
zarel has joined #nixos-dev
Jackneill has quit [Remote host closed the connection]
Jackneill has joined #nixos-dev
alp has quit [Ping timeout: 272 seconds]
cole-h has joined #nixos-dev
alp has joined #nixos-dev
<cole-h>
worldofpeace: I'm planning to submit a PR fixing https://github.com/NixOS/nixos-homepage/issues/416 -- should the "recommended for most users" label go beside the Plasma ISO, or the Gnome ISO?
<{^_^}>
nixos-homepage#416 (by worldofpeace, 25 weeks ago, open): Download Page: Link to GNOME ISO
<supersandro2000>
cole-h: the default until now was plasma so my guess would be to point to this
<cole-h>
That's what I was planning on, but wanted input from an RM :P
rajivr has quit [Quit: Connection closed for inactivity]
<worldofpeace>
cole-h: ooh I think garbas already did in a bigger PR
<samueldr>
hi y'all, I think this should get somewhat high priority in reviewing and handling, considering it'll help us the next time Hydra is having a bad time with aggregate jobs https://github.com/NixOS/hydra/pull/825
<samueldr>
there is two open "TODOs" on it, but I'm also interested in feedback from Hydra maintainers and contributors before I tackle them, in case I tackle them wrong, or need to change a lot
alp has joined #nixos-dev
saschagrunert has quit [Remote host closed the connection]
justanotheruser has quit [Ping timeout: 272 seconds]
teto has joined #nixos-dev
saschagrunert has joined #nixos-dev
<worldofpeace>
+1 to that samueldr anyone with knowledge in those domains would be appreciated
<samueldr>
note that I don't pretend it's _the_ solution, but there are two problems that have been targeted and possibly resolved
saschagrunert has quit [Client Quit]
utsl has joined #nixos-dev
justanotheruser has joined #nixos-dev
justanotheruser has quit [Ping timeout: 260 seconds]
<gchristensen>
hmmm... per-machine sshPublicHostKey support was ported from hydra to nix, but not actually the implementation
justanotheruser has quit [Ping timeout: 260 seconds]
dhess has joined #nixos-dev
<dhess>
Anyone have a working example of how to use the GithubPulls.pm plugin with the new Hydra declarative job spec syntax?
<worldofpeace>
is it chaotic neutral that I love how peti moved the haskell docs to his own website in markdown? gchristensen I think it may be time soon to make docbook.sucks but it's the same content as docbook.rocks
<clever>
dhess: yes
<dhess>
clever: oh I figured you wouldn't as you're not using a new-ish Hydra, when the new syntax was introduced.
<clever>
the default.nix in not-os/ generates that
<dhess>
clever: yeah all of our old stuff is based on your configs like that one, but in the last month or so, it no longer works for new projects. (Projects that have existed in our Hydra since before whatever changed still work fine with PRs.)
<dhess>
it's just a minor change but it means you don't need to generate the JSON files using Nix anymore. (Unless you want to filter or things like that.)
<clever>
i think the new stuff, isnt dynamic in any way
<clever>
its just exactly whats in the json, and nothing more
<clever>
and i think its the exact same format my default.nix outputs
<clever>
it just skips the generation step
<dhess>
it is
<clever>
so you can still run my default.nix, and commit the output
<clever>
but its static, so it cant dynamically change when PR's get made
<clever>
dhess: that would pass you a list of all PR's, as a json file
<clever>
and the default.nix then has to parse it, and convert it into a list of jobsets
<dhess>
ahh I was hoping that part had been added to the new declarative job spec code.
<dhess>
well, in any case, as I said, your method of generating the jobsets for pull requests doesnt work anymore, so something is broken.
<clever>
can you link a hydra where its not working?
<dhess>
it *does* work for old Hydra jobsets, just not new ones. This is since... maybe early September? So around the time that declarative jobspec commit was made. (I'm not saying that commit is the cause for the breakage)