<layus>
Any idea why builtins.fetchGit would not find a given commit? Could it be branch-related ?
__Sander__ has joined #nixos-dev
<sphalerite>
layus: iirc the git protocol doesn't have provisions for fetching a specific commit (as opposed to a named ref). Not sure what to do about that though. If you control the repo you could just tag it
<layus>
Seems to work with nix-shell, but not with nix-instantiate
<layus>
or maybe it just works through the daemon, but not with sudo...
<sphalerite>
ooooh, that would explain it
davidlt has joined #nixos-dev
davidlt_ has quit [Ping timeout: 240 seconds]
<sphalerite>
nix-instantiate won't add anything other than derivations to the store AFAIU
<sphalerite>
so IFD and builtins.fetch* won't work in it
<sphalerite>
I'm not sure though, I might be completely wrong :)
<sphalerite>
nvm, it works for me
Synthetica has joined #nixos-dev
<niksnut>
layus: if you want to fetch a commit from any branch other than master, you have to specify both "rev" and "ref" (where "ref" is a branch that contains the commit)
<layus>
working, thanks :-)
ashgillman has joined #nixos-dev
<ashgillman>
jkh
ashgillman has quit [Ping timeout: 248 seconds]
ashgillman has joined #nixos-dev
<fpletz>
niksnut: there are lots of hanging jobs on hydra again, have been buildig for ~12 hours
<ekleog>
(and I guess this has already been reported, but on https://hydra.nixos.org/jobset/nixos/release-18.03 even builds from 5 days ago haven't completed yet, despite new jobs having been triggered)
<fpletz>
ekleog: that's the problem, yeah :)
<fpletz>
I've restarted all failing builds on the current 18.03 evaluation because I was able to build them locally (mostly the qt5-stuff)
<ekleog>
oh, I thought that'd be a separate problem, where hydra shouldn't start new jobs before having finished the previous ones :)
<fpletz>
ekleog: the reason for that is that my restart restarted the same jobs in different evaluations, so if two jobs evaluate to the same package both will be restarted
<gchristensen>
:community management siren: can someone help me by reading a PR comment before I post it? :)
lopsided98 has quit [Ping timeout: 245 seconds]
<globin>
gchristensen: o/
s33se has quit [Ping timeout: 240 seconds]
lopsided98 has joined #nixos-dev
<angerman>
Sonarpulse: did you see my email? If you’ve a chance to point me somewhere further where I should look tomorrow that would be great!
<Sonarpulse>
angerman: no i did not!
<Sonarpulse>
checking now
Synthetica has quit [Quit: Connection closed for inactivity]
Synthetica has joined #nixos-dev
<Sonarpulse>
angerman: so you can trace how deps are gotten with NIX_DEBUG=7
<Sonarpulse>
in a nix-shell or nix-build
<Sonarpulse>
that will do bash -x on the propagation logic
<Sonarpulse>
angerman: I hope this isn't due to some horrible !isLinux somewhere
<angerman>
or !isDarwin...
ma27 has joined #nixos-dev
<angerman>
Sonarpulse: alright. That’s probably going to be like wading through a swamp of output...
<Sonarpulse>
angerman: yes
<angerman>
Sonarpulse: I’ll see how far I get tomorrow.
<Sonarpulse>
but hopefully only one thing introduces these
<Sonarpulse>
ok
<Sonarpulse>
sorry did not see sooner
<Sonarpulse>
angerman: is your mingw bump on master?
<angerman>
Sonarpulse: no worries.
<angerman>
Sonarpulse: no. Im still collecting what is needed.
<Sonarpulse>
angerman: if you want to do a WIP PR
<Sonarpulse>
I can probably help with that
<angerman>
Sonarpulse: a lot of local haskell package patches, some ghc patches and some minor additions to nixpkgs.
<Sonarpulse>
I'll also try to fix the overrride problems
<Sonarpulse>
yeah for the last i mean, the non-haskell-specific stuff
<angerman>
Sonarpulse: alright, I can make a WIP PR tomorrow I guess.
<Sonarpulse>
angerman: cool, thanks!
<angerman>
it’s not much though :-)
<angerman>
I could stick most of he customization into the config.nix
<angerman>
Too much for my liking but GHC Head is in a mess right now package wise. Will hopefully be sorted tomorrow.
<Sonarpulse>
angerman: yeah that's why I'm thinking some of the generic mingw stuff has less stuff in flux
<Sonarpulse>
there's something things like, should stack protection or whatever it was be disabled across the board
<angerman>
Right.
<globin>
ikwildrpepper: could you give Mic92 access to hydra?
jtojnar has quit [Quit: jtojnar]
<Mic92>
fpletz can then explain the low-level details of hydra.
jtojnar has joined #nixos-dev
<fpletz>
I can? :>
<Sonarpulse>
does --print-file-name work right with nix?
<Sonarpulse>
seems flaky AF
<Sonarpulse>
I tried gcc and clang, linux and darwin
<tilpner>
Any ideas on how to allow a user to set submodule defaults?
<tilpner>
I tried having an option (type = attrsOf thatSubmodule) for the instances, and another (type = thatSubmodule) to provide defaults, but that lead to infinite recursion when trying to set the defaults of thatSubmodule
FRidh has quit [Quit: Konversation terminated!]
JosW has quit [Quit: Konversation terminated!]
__Sander__ has quit [Quit: Konversation terminated!]
<gchristensen>
should Nix 1.x be unable to run NixOS tests from Unstable?
<gchristensen>
so to do a proper build of everything sure makes sense, but it is now impossible to build any nixos test on unstable without nix2 is all
<dtz>
that is a bit surprising
<gchristensen>
error: assertion failed at /home/grahamc/projects/nixpkgs/pkgs/build-support/closure-info.nix:11:1
<dtz>
oh, yeah the closure-info stuff...:(
<Sonarpulse>
gchristensen: so still no nix-pill deploy
<Sonarpulse>
ping eelco?
<gchristensen>
sure
<Sonarpulse>
niksnut: can you check the nix-pill deploy?
<Sonarpulse>
I temporarily switched it to unstable so my accientally-new-lib-function pill 20 would build (yay documentation)
<Sonarpulse>
(nixos-unstable)
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos-dev
<fpletz>
is somebody currently looking after staging? looks like some commits from there should be picked to 18.03 but it hasn't been merged to master for some time
<fpletz>
phew, still over 800 PRs…
MichaelRaskin has joined #nixos-dev
davidlt_ has joined #nixos-dev
ashgillman has quit [Ping timeout: 256 seconds]
davidlt has quit [Ping timeout: 260 seconds]
obadz has quit [Ping timeout: 264 seconds]
obadz has joined #nixos-dev
jtojnar has quit [Read error: Connection reset by peer]
<gchristensen>
I must have been dreaming. how about now?
<jtojnar>
gchristensen: better, though the repo is NixOS/nixpkgs
<gchristensen>
oh I see
<gchristensen>
how about now-now
<jtojnar>
great thanks
<gchristensen>
I set it to eval, too
<jtojnar>
cool
jtojnar has quit [Quit: jtojnar]
jtojnar has joined #nixos-dev
davidlt_ has quit [Remote host closed the connection]
ma27 has quit [Ping timeout: 252 seconds]
<shlevy>
dtz: Can you expand on the theoretical leak you alluded to in https://github.com/NixOS/nix/issues/1968#issuecomment-372723823 ? IIUC all that cache admins could see would be some hash they don't know about, which at best they could assume is *some* private path, right?
<dtz>
right
<shlevy>
I mean, that I've requested 03y0ca2js6mw2brcp3zfx2ff92nb5rxh.narinfo doesn't tell cache.nixos.org admins anything meaningful, even if they know that the request came from someone working at Target and they assume our internal cache later return successful
<shlevy>
An really that assumption is silly, since a huge portion of my builds have some small things that aren't cached for one reason or another, even if all the hard work is cached :)
<MichaelRaskin>
I guess the point was that even if the cache is open to the world but not enumerable, non-cache.nixos.org-admins cannot guess the hash
<shlevy>
Ah, so the risk would only be if *my* cache was public but I was hiding things behind the non-enumerability
<dtz>
lol right
<shlevy>
Gotcha
ma27 has joined #nixos-dev
jtojnar has quit [Read error: Connection reset by peer]