<Dirt[m]>
Linus: Good to know, I'll try to keep my messages trim then.
<Dirt[m]>
Yeah I was trying to use nixos' configs to manage everything (including user stuff), both of those approaches sound like they could be better. Editing root's files for user packages was definitely feeling screwy. Thank you!
* infinisil
casts another vote for home-manager
<sphalerite>
Dirt: if all you want is to install some packages you can have an expression looking like `let pkgs = import <nixpkgs> {}; in {inherit (pkgs) tmux vim nethack;}` in userenv.nix, and then do `nix-env -f userenv.nix -ir` to make your user profile contain exactly those packages (removing everything else)
jtojnar has quit [(Read error: Connection reset by peer)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] yegortimoshenko closed pull request #33191: melpaPackages: match version of magit-popup to magit (master...magit-popup-version) https://git.io/vbNzW
NixOS_GitHub has left #nixos []
<dhess>
Trying to figure out how to write the release.nix/jobset for that scenario
otherjrolfs has quit [(Ping timeout: 264 seconds)]
<LnL>
I used the toplevel attribute to hide everything except what's in the overlay
<dhess>
think of it.
<dhess>
err
<dhess>
weird, emacs cut off my message
<dhess>
LnL: thanks! I should have thought to check out the Mozilla Rust repo as well, now that I think of it.
otherjrolfs has joined #nixos
jb55 has quit [(Ping timeout: 272 seconds)]
<dhess>
LnL: That's interesting. I was trying to figure out how to do something similar with my set of overlays to expose them as a sort of top-level "all-packages.nix", as well.
<dhess>
LnL: hmm I don't understand how using 'toplevel' hides all but the overlay. How does that work, exactly?
<dhess>
nvm
<dhess>
got it
<dhess>
it's defined in all-packages.nix
otherjrolfs has quit [(Ping timeout: 268 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] yegortimoshenko pushed 2 new commits to master: https://git.io/vbNwk
<NixOS_GitHub>
nixpkgs/master 0004682 Vincent Laporte: gtksourceview: run checks on Linux only
<LnL>
import <nixpkgs> { ... } doesn't have a nixpkgs argument
<dhess>
oh I see
<dhess>
LnL: so can you just do 'overlays = [ (import ./nixpkgs-python) ]' or similar and that default.nix will work like a standard overlay file?
nuncanada has quit [(Quit: Leaving)]
<LnL>
yes, the all-packages.nix could be used like that
<dhess>
ok right
<dhess>
except that you'd also need to use the .toplevel attribute I think
<andromeda-galaxy>
I'm trying to import a nixpkgs checkout from inside a nix overlay, in a way that works both from inside configuration.nix and user ~/.config/nixpkgs/overlays. However, I can't figure out the right arguments to pass to 'import <nixpkgs-unstable>' to ensure that the nixpkgs configuration is the same as the configuration of the nixpkgs calling the overlay... is there any way to do that?
jrolfs has joined #nixos
<andromeda-galaxy>
self.config works for the 'config' argument of toplevel/impure, but I can't find equivalents for (local|cross)System, overlays, and stdenvStages
dieggsy has quit [(Remote host closed the connection)]
<andromeda-galaxy>
LnL: I saw that you were doing something fancy with toplevel, do you know how to do ^?
<LnL>
not sure how to do that
muji has joined #nixos
jrolfs has quit [(Ping timeout: 240 seconds)]
<muji>
Hi. guys... I am hoping you can help in an issue...
<muji>
I have installed nix-shell --version nix-shell (Nix) 1.11.16 on MACOS 10.3.2
dan_b has quit [(Ping timeout: 256 seconds)]
<muji>
it was having an issue error: build of ‘/nix/store/gq3cp25bl06cm45ys0hlg6c04h1fhym7-stage2.nix.drv’ failed
<yegortimoshenko>
andromeda-galaxy: do you have that overlay publicly accessible?
thc202 has quit [(Ping timeout: 240 seconds)]
<muji>
How do I update my nix to implement that change
<LnL>
muji: did you update your channel?
<muji>
umm... very new to this... can you indicate how I do it?
<andromeda-galaxy>
yegortimoshenko: my attempts to set pkgs.unstable = import <nixos-unstable> {the same config as before} is all that is in the overlay, and that doesn't work.
<yegortimoshenko>
andromeda-galaxy: actually, config *is* overridable, so taking back that it probably wouldn't work. now i think it probably would, or some variation of it.
<andromeda-galaxy>
yegortimoshenko: that or the equivalent that I was doing earlier does successfully override the config argument to impure.nix as far as I can tell. However, impure/default also take overlays, stdenvStages, crossSystem, and localSystem as arguments that are not part of the config dictionary, and it is these that I can't figure out how to override
<andromeda-galaxy>
in particular I want the new nixpkgs checkout to have the same set of overlays applied to it as the original, and I can't find the overlays anywhere
<LnL>
muji: don't understand why some people run into that, try sudo su -; nix-channel --update
<muji>
LnL i am building the app databrary which is at github.com/databrary/databrary
<andromeda-galaxy>
LnL, yegortimoshenko: unfortunately I unexpectedly have to go afk for a bit, and might not reply for a couple of hours... thanks for trying to help!
<LnL>
reflex-platform need to update their version of nixpkgs to to get the fix
<muji>
oh i see...
<LnL>
you can verify with nix-build '<nixpkgs>' -A darwin.Libsystem --check
infinisil has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
jrolfs has quit [(Ping timeout: 256 seconds)]
Tobba has quit [(Ping timeout: 240 seconds)]
infinisil has joined #nixos
<muji>
running
lambdamu has joined #nixos
lambdamu_ has quit [(Ping timeout: 248 seconds)]
<muji>
warning: rewriting hashes in ‘/nix/store/dp5dhzh55q3cxsaknclxqp3nj5nn86k3-Libsystem-osx-10.11.6’; cross fingers error: derivation ‘/nix/store/i4i0yffmfqmxgn1q5345hb83l7q2nl3p-Libsystem-osx-10.11.6.drv’ may not be deterministic: output ‘/nix/store/dp5dhzh55q3cxsaknclxqp3nj5nn86k3-Libsystem-osx-10.11.6’ differs
<LnL>
that's normal
<muji>
so basically the way my system will work is when reflex-platform updates their nixpkgs based on that one line change that was implemented to work on Macs?
justanotheruser has quit [(Quit: WeeChat 1.9.1)]
<LnL>
yep
justanotheruser has joined #nixos
<muji>
sigh
<palo>
yegortimoshenko: about vcvrack, I will create myself a ticket to improve the vcvrack like you mentioned it in the pr, wont to it now. Thanks for the quick response and the help.
<muji>
thanks for your help LnL
<yegortimoshenko>
palo: it's ok! don't worry, it is good enough, it doesn't particularly warrant further improvement :-)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] yegortimoshenko closed pull request #33155: fastpbkdf2: init at version 1.0.0 (master...fastpbkdf2) https://git.io/vbb8R
NixOS_GitHub has left #nixos []
<palo>
:D
Tobba has joined #nixos
mbrgm has quit [(Ping timeout: 252 seconds)]
otherjrolfs has joined #nixos
jrolfs has joined #nixos
<LnL>
muji: the rest should work if you can get a cache for that one build from somewhere
mbrgm has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Ericson2314 opened pull request #33196: doc: Breaking change release note for bintools-wrapper (master...bintools-release-note) https://git.io/vbNrx
NixOS_GitHub has left #nixos []
<muji>
LnL not sure how to do that... I have pinged reflex-platform owners to update?
jrolfs has quit [(Ping timeout: 240 seconds)]
otherjrolfs has quit [(Ping timeout: 248 seconds)]
<nix_newb>
is there a function to get the absolute path of current directory?
<clever>
muji: is there a nixpkgs.json in the reflex repo?
otherjrolfs has quit [(Ping timeout: 256 seconds)]
jrolfs has quit [(Ping timeout: 264 seconds)]
stepcut has quit [(Remote host closed the connection)]
jrolfs has joined #nixos
jrolfs has quit [(Ping timeout: 260 seconds)]
yegortimoshenko has quit [(Ping timeout: 256 seconds)]
otherjrolfs has joined #nixos
drakonis has quit [(Read error: Connection reset by peer)]
jrolfs has joined #nixos
otherjrolfs has quit [(Ping timeout: 272 seconds)]
jrolfs has quit [(Ping timeout: 248 seconds)]
schoppenhauer has quit [(Ping timeout: 260 seconds)]
otherjrolfs has joined #nixos
schoppenhauer has joined #nixos
tmaekawa has joined #nixos
tmaekawa has quit [(Client Quit)]
otherjrolfs has quit [(Ping timeout: 260 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] bhipple opened pull request #1769: Help? Working on Hydra release builds for master x64 Deb/RPM (master...fixing-hydra-builds) https://git.io/vbNi9
NixOS_GitHub has left #nixos []
justan0theruser has joined #nixos
justanotheruser has quit [(Ping timeout: 240 seconds)]
<gfixler>
any idea why nix-env -i <anything> would stall forever?
<gfixler>
oh, of course, the second I send that it finally installs :-/
<gfixler>
kept trying, and it kept sitting there for minutes
otherjrolfs has joined #nixos
<bhipple[m]>
Generally, it's much faster to use nix-env -iA <attribute-name> instead of nix-env -i <package-name>, since the latter requires that the parser run through and evaluate all of nixpkgs first, using a lot of CPU + RAM + time
pie_ has quit [(Ping timeout: 265 seconds)]
otherjrolfs has quit [(Ping timeout: 248 seconds)]
<gfixler>
Hmmm... don't grok that one yet.
otherjrolfs has joined #nixos
<samueldr>
simplified, -iA uses the attribute set in `<nixpkgs/pkgs/top-level/all-packages.nix>`, what is given is an attribute's name, that's fast
sary has quit [(Ping timeout: 240 seconds)]
<samueldr>
the other does *something* with the attribute set, mapping all packages that are also attributes deeper into the set, and then checks the `.name` attribute on each package
<samueldr>
that's a bit more work
<samueldr>
(the other meaning -i)
<gfixler>
I [sort of] see. It's all going in. More and more will click as I contiue.
sary has joined #nixos
otherjrolfs has quit [(Ping timeout: 268 seconds)]
<andromeda-galaxy>
self.config works for the 'config' argument of toplevel/impure, but I can't find equivalents for (local|cross)System, overlays, and stdenvStages, which aren't in config
otherjrolfs has joined #nixos
otherjrolfs has quit [(Ping timeout: 248 seconds)]
otherjrolfs has joined #nixos
<andromeda-galaxy>
oops, meant to say: I'm trying to import a nixpkgs checkout from inside a nix overlay, in a way that works both from inside configuration.nix and user ~/.config/nixpkgs/overlays. However, I can't figure out the right arguments to pass to 'import <nixpkgs-unstable>' to ensure that the nixpkgs configuration is the same as the configuration of the nixpkgs calling the overlay... is there any way to
<andromeda-galaxy>
do that?
otherjrolfs has quit [(Ping timeout: 240 seconds)]
pxc has joined #nixos
otherjrolfs has joined #nixos
otherjrolfs has quit [(Ping timeout: 248 seconds)]
cement has joined #nixos
otherjrolfs has joined #nixos
phreedom has joined #nixos
otherjrolfs has quit [(Ping timeout: 240 seconds)]
klntsky has quit [(Remote host closed the connection)]
vidbina has joined #nixos
p3a_ has joined #nixos
moet has joined #nixos
otherjrolfs has joined #nixos
vidbina has quit [(Ping timeout: 248 seconds)]
klntsky has joined #nixos
otherjrolfs has quit [(Ping timeout: 248 seconds)]
jb55 has quit [(Ping timeout: 264 seconds)]
otherjrolfs has joined #nixos
verite has joined #nixos
klntsky has quit [(Ping timeout: 272 seconds)]
otherjrolfs has quit [(Ping timeout: 248 seconds)]
rogue_koder has quit [(Ping timeout: 260 seconds)]
ma27 has joined #nixos
otherjrolfs has quit [(Ping timeout: 264 seconds)]
<hyper_ch>
sphalerite: also for you
romildo has joined #nixos
rogue_koder_ has quit [(Quit: Konversation terminated!)]
<romildo>
A Haskell application I am developing on NixOS unstable uses tagsoup-megaparasec, but currently tagsoup-megaparsec building fails with the message: Setup: Encountered missing dependencies: megaparsec >=5.0.1 && <6.
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
<romildo>
How can I override it so that it so that the newer megaparsec is used?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to python-unstable: https://git.io/vbNye
<NixOS_GitHub>
nixpkgs/python-unstable 4344831 Frederik Rietdijk: fixup pytest
<NixOS_GitHub>
nixpkgs/python-unstable 97850da Frederik Rietdijk: awscli: 1.14.6 -> 1.14.17
NixOS_GitHub has left #nixos []
otherjrolfs has joined #nixos
otherjrolfs has quit [(Ping timeout: 252 seconds)]
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to python-unstable: https://git.io/vbN77
<NixOS_GitHub>
nixpkgs/python-unstable f3a7e51 Frederik Rietdijk: python.pkgs.texttable: 0.8.4 -> 1.1.1
<NixOS_GitHub>
nixpkgs/python-unstable 5691049 Frederik Rietdijk: python.pkgs.docker_compose: fixup
NixOS_GitHub has left #nixos []
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 3 new commits to master: https://git.io/vbN7h
<NixOS_GitHub>
nixpkgs/master c3b98dc rnhmjoj: pythonpackages.libusb1: init at 1.6.4
<NixOS_GitHub>
nixpkgs/master aa6a18a rnhmjoj: steamcontroller: init at 2017-08-11
<NixOS_GitHub>
nixpkgs/master bb72931 Frederik Rietdijk: Merge pull request #33201 from rnhmjoj/steamcontroller...
NixOS_GitHub has left #nixos []
pukkamustard has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to python-unstable: https://git.io/vbN5J
<NixOS_GitHub>
nixpkgs/python-unstable 21baca8 Frederik Rietdijk: python.pkgs.attrs: disable tests to prevent infinite recursion
NixOS_GitHub has left #nixos []
Avaq has quit [(Ping timeout: 260 seconds)]
otherjrolfs has joined #nixos
moet has joined #nixos
ssmike1 has joined #nixos
ssmike has quit [(Read error: Connection reset by peer)]
ssmike1 is now known as ssmike
otherjrolfs has quit [(Ping timeout: 256 seconds)]
nallar has joined #nixos
Ross has quit [(Ping timeout: 240 seconds)]
nallar is now known as Ross
pukkamustard has quit [(Quit: Page closed)]
sary has quit [(Ping timeout: 272 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 21baca8 to 1bb7a2f: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 5b69494 Frederik Rietdijk: python: asgiref: 1.1.2 -> 2.0.1
<NixOS_GitHub>
nixpkgs/python-unstable b77f94f Frederik Rietdijk: python: asn1crypto: 0.23.0 -> 0.24.0
<NixOS_GitHub>
nixpkgs/python-unstable 8648ff4 Frederik Rietdijk: python: astor: 0.5 -> 0.6.2
NixOS_GitHub has left #nixos []
otherjrolfs has joined #nixos
sary has joined #nixos
pukkamustard has joined #nixos
Avaq has joined #nixos
Ross has quit [(Ping timeout: 240 seconds)]
otherjrolfs has quit [(Ping timeout: 264 seconds)]
Ross has joined #nixos
asuryawanshi has joined #nixos
Itkovian has joined #nixos
Avaq has quit [(Ping timeout: 250 seconds)]
jrolfs has joined #nixos
otherjrolfs has joined #nixos
Itkovian has quit [(Ping timeout: 240 seconds)]
<adisbladis>
No more 34c3 for me. Nice meeting everyone!
Fannar has quit [(Quit: Lost terminal)]
otherjrolfs has quit [(Ping timeout: 240 seconds)]
jrolfs has quit [(Ping timeout: 268 seconds)]
moet has quit [(Ping timeout: 252 seconds)]
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
ssmike has quit [(Ping timeout: 252 seconds)]
drakonis has joined #nixos
asuryawanshi has quit [(Remote host closed the connection)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 1bb7a2f to b06e13a: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 6ea7a3c Frederik Rietdijk: python: ldap3: 2.3 -> 2.4
<NixOS_GitHub>
nixpkgs/python-unstable c7731d2 Frederik Rietdijk: python: line_profiler: 2.0 -> 2.1.2
<NixOS_GitHub>
nixpkgs/python-unstable d13d464 Frederik Rietdijk: python: llfuse: 1.0 -> 1.3.2
NixOS_GitHub has left #nixos []
jrolfs has joined #nixos
sary has quit [(Ping timeout: 260 seconds)]
fendor has quit [(Read error: Connection reset by peer)]
stefanc_diff has quit [(Quit: bye)]
jrolfs has quit [(Ping timeout: 248 seconds)]
pukkamustard has quit [(Ping timeout: 260 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from b06e13a to 000a4e0: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 354a0c1 Frederik Rietdijk: python: mistune: 0.7.4 -> 0.8.3
<NixOS_GitHub>
nixpkgs/python-unstable 254ccc6 Frederik Rietdijk: python: moto: 0.4.31 -> 1.1.25
<NixOS_GitHub>
nixpkgs/python-unstable b8f311a Frederik Rietdijk: python: nbxmpp: 0.6.0 -> 0.6.2
NixOS_GitHub has left #nixos []
ssmike has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 8 new commits to master: https://git.io/vbNdA
<NixOS_GitHub>
nixpkgs/master af6f4f0 Peter Simons: ghc: drop obsolete version 7.8.3...
<NixOS_GitHub>
nixpkgs/master f5a5010 Peter Simons: Drop Haskell package set for GHC 7.8.4....
<NixOS_GitHub>
nixpkgs/master 0232282 Peter Simons: hackage2nix: disable broken Hydra builds...
NixOS_GitHub has left #nixos []
georges-duperon has quit [(Ping timeout: 272 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 1 new commit to master: https://git.io/vbNdx
<NixOS_GitHub>
nixpkgs/master 1b33f09 Peter Simons: uhc: disable broken build
NixOS_GitHub has left #nixos []
vidbina has joined #nixos
jrolfs has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 000a4e0 to b2cc11e: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable ab54a74 Frederik Rietdijk: python: textacy: 0.4.1 -> 0.5.0
<NixOS_GitHub>
nixpkgs/python-unstable a2ddb9a Frederik Rietdijk: python: Theano: 0.9.0 -> 1.0.1
<NixOS_GitHub>
nixpkgs/python-unstable 2068e35 Frederik Rietdijk: python: thespian: 3.8.3 -> 3.9.0
NixOS_GitHub has left #nixos []
jrolfs has quit [(Ping timeout: 248 seconds)]
sary has joined #nixos
JosW has joined #nixos
<JosW>
whooaa, don't know if this has been mentioned here before
<JosW>
"But every so often, I like to see something different—something that might remind me from where I came. That’s exactly what happened when I came into NixOS. To be quite honest, I had assumed this would be just another Linux distribution that offered the standard features, with the KDE Plasma 5 interface. "
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from b2cc11e to ffef903: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 37259a1 Frederik Rietdijk: python: widgetsnbextension: 3.0.8 -> 3.1.0
<NixOS_GitHub>
nixpkgs/python-unstable fc1f500 Frederik Rietdijk: python: ws4py: 0.4.2 -> 0.4.3
<NixOS_GitHub>
nixpkgs/python-unstable 3aeff31 Frederik Rietdijk: python: yapf: 0.19.0 -> 0.20.0
NixOS_GitHub has left #nixos []
<JosW>
the word is spreading :)
<gchristensen>
JosW: :D
<JosW>
find it nice to see others bumping into NixOS
<NickHu>
Can anyone check if they can install darcs?
<NickHu>
I think it's trying to use the wrong version of cabal and it's failing to build as a result
<gchristensen>
NickHu: I can on 17.09
<NickHu>
Oh, I'm on nixos-unstable
<JosW>
me trying to get my head around the turmoil caused by choosing to invest more time in Nixos/Haskell or containers/kubernetes
<NickHu>
What's the command that lets you go into a nix-shell which replicates the build environment of a package?
<gchristensen>
JosW: Nix can produce docker containers ... :)
<JosW>
but since i am writing this here, seems that I somehow always come back here the last few months...
georges-duperon has joined #nixos
georges-duperon has quit [(Remote host closed the connection)]
<JosW>
gchristensen: yeah i know, but i also read a few times that docker isn't the way to go
Mateon3 has joined #nixos
<NickHu>
Btw, how do you know what channel nix-* uses when you omit -A?
<JosW>
MoreTea: thats why I would still love to hear how your talk was at the conf and what the overall consensus has been about Nix(OS) versus the rest...
Mateon1 has quit [(Ping timeout: 256 seconds)]
Mateon3 is now known as Mateon1
<joepie91>
JosW: hmm. it's unfortunate that the article doesn't really go into the declarative nature at all, and rather presents it as a pre-install configuration mechanism :/ I've noticed this in a few more articles, and I'm wondering how this can be better communicated
<joepie91>
most every article so far that *isn't* written by an active NixOS user has not really gone into what makes NixOS NixOS at all
<joepie91>
so there's clearly something going wrong there :P
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] yegortimoshenko pushed 2 new commits to master: https://git.io/vbNbT
<JosW>
joepie91: true but at least NixOS has been mentioned in a more readable way, allmost every other writing assumes too much knowledge on the subject
<hyper_ch>
JosW: isn't NixOS just another LInux distribution that offers the standard features with KDE 5?
<joepie91>
JosW: sure, but that's the problem - there's basically two categories of article, 1) superficial articles that treat NixOS like an arbitrary FHS distro, and 2) in-depth articles by (semi-)experienced NixOS users that aren't accessible to the average reader :P
<JosW>
Yep but that is as far as it goes, installing packages is the first strumbling block and that is where you have to dive in the declarative stuff
<MoreTea>
Actually, you can use nix-env -i pkgname to install stuff in a more ad hoc way
<hyper_ch>
JosW: Remove the # sign at the beginning of the line to uncomment this option (making sure /dev/sda is the name of your newly created partition). --> "/dev/sda" is no partition
<joepie91>
MoreTea: then you're giving up half the declarative benefits though :P
<hyper_ch>
JosW: you can also configure default user in the configuration nix :)
<MoreTea>
Yes. I used to do this quite a lot. Now that I know more, I use the home manager
<JosW>
gchristensen: do you have any pointers to setups with docker and Nixos perhaps?
<JosW>
I persuaded a friend of mine to try NixOS, he did but after a week he quit. Talked about some preffered Terminator settings he could't get installed and I wasn't any help whith this...
<joepie91>
JosW: I don't think NixOS is quite usable enough yet to recommend it to the average Linux user; my advice so far has been "have a look into it and how it works, and give it a shot in a VM, but expect to need several weeks to get up and running as before"
<hyper_ch>
JosW: the one drawback on nixos is if you want to use software that's not packaged...
<JosW>
how and where would he be able to configure his liked Terminator settings?
<joepie91>
JosW: some of the people I've advised that to have decided to mess with it and consider an eventual switch, others have said "nope, don't have the time now"
<JosW>
joepie91: He's not a average Linux user, we both thinker a lot with stuff
<avn>
joepie91: but I can recommend it for "broken hands" users (like grandparents etc) with external administration. Is a bulletproof, because reboot fix everything ;)
<JosW>
But this is a bit more complex, but will pervail because I believe/feel Nix is something special
Kingsquee has quit [(Ping timeout: 265 seconds)]
<joepie91>
JosW: right; in that case the "try it in a VM" advice works well :P
ssmike has quit [(Ping timeout: 260 seconds)]
<JosW>
17:09 running on my test server under KVM with no problems:)
<adisbladis>
Its also super easy to just lift your config from that vm and nuke your other install once you are satisfied =)
ssmike has joined #nixos
luigy has joined #nixos
<avn>
I switch my work machine to nixos after 4 hours test in vm ;)
<avn>
as a second-boot option to be honest, but I never boot back ;)
ma27 has quit [(Ping timeout: 272 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from ffef903 to 741af0f: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 0d351db Frederik Rietdijk: python: Shapely: 1.6.2.post1 -> 1.6.3
<NixOS_GitHub>
nixpkgs/python-unstable 14a8691 Frederik Rietdijk: python: simplejson: 3.11.1 -> 3.13.2
<NixOS_GitHub>
nixpkgs/python-unstable aa28a2a Frederik Rietdijk: python: smart_open: 1.5.3 -> 1.5.6
NixOS_GitHub has left #nixos []
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] bendlas pushed 2 new commits to master: https://git.io/vbNNZ
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to python-unstable: https://git.io/vbNxc
<NixOS_GitHub>
nixpkgs/python-unstable 07d939a Frederik Rietdijk: python.pkgs.pyaml: move expression
<NixOS_GitHub>
nixpkgs/python-unstable c0e2cf4 Frederik Rietdijk: python: pyaml: 15.02.1 -> 17.12.1
NixOS_GitHub has left #nixos []
<wchresta>
Hey everyone. So I'm trying to configure vi to be an alias to nvim. I was able to get vim to be an alias with pkgs.neovim.override.vimAlias = true; but the same does not work with .viAlias. Any pointers?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from c0e2cf4 to 0604d9e: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 3257925 Frederik Rietdijk: python: pydot: 1.2.3 -> 1.2.4
<NixOS_GitHub>
nixpkgs/python-unstable dfea43c Frederik Rietdijk: python: pygit2: 0.26.0 -> 0.26.3
<NixOS_GitHub>
nixpkgs/python-unstable e699acd Frederik Rietdijk: python: pyglet: 1.2.4 -> 1.3.0
<infinisil>
wchresta: If you want an easy way, you can just use your .bashrc (or .zshrc, the config file for your shell) and put it there
<wchresta>
infinisil: I was hoping to be able to get it in my configuration.nix; mainly to understand how nix works - not so much to actually fix the problem. But yes, that would work but it's not what I'm after
<infinisil>
wchresta: I see, I also just noticed there is the viAlias overridable option
<infinisil>
wchresta: How did you try to set this?
<globin>
grahamc: have you tried the rust github libraries?
<globin>
gchristensen: ^ whatever nick you're using %)
<maurer>
globin: I'm curious what you're referring to (since I heavily use rust on nixos) - are you talking about mozilla's overlay? buildRustPackage? Or something new I don't know about yet?
<globin>
gchristensen: ah I see you're using hubcaps :)
<globin>
maurer: no refering to a github api library for rust
<maurer>
Oh, OK
<globin>
maurer: and I know he is interacting with it for his bot :)
stepcut has quit [(Remote host closed the connection)]
stepcut has joined #nixos
<maurer>
Ah. I just looked at it, and it looks a little underdocumented and is 2 years old and one API revision out of date
<maurer>
but that doesn't mean it won't work :P
<vidbina>
if anyone has a sec, can you try to build https://github.com/NixOS/nixpkgs/pull/33148. I'm trying to figure out why it "sometimes" fails to produce a working binary. Tested it on my local machine and in the VirtualBox demo but it seems to work just fine in there :'(
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 2 new commits to master: https://git.io/vbNpd
<FRidh>
vidbina: builds fine here. It's likely failing in your case because you do not use a sandbox.
<vidbina>
FRidh: cool. Besides build do the results run?
<vidbina>
my reviewer is having issues
<FRidh>
the following seems to run fine nix-build -A bitscope.chart && ./result/bin/bitscope-*
<vidbina>
FRidh: thank you. I'll need to look further to figure out how to reproduce the issue bjornfor is facing
<vidbina>
FRidh: are there any red flags you notice when you look at the package. This is my first addition so I may have done something noobishly stupid. Would appreciate any feedback.
<FRidh>
vidbina: nope, looks fine to me.
Turion has joined #nixos
<Turion>
Hi, I'm trying to help solve #32969 because I'm affected as well. Where do I find the place where all the haskell packages are defined in nixpkgs?
pie_ has joined #nixos
<Turion>
And, can I somehow define a package to be buildable only with an older version of cabal?
<FRidh>
Turion: pkgs/development/haskell-modules
simendsjo has quit [(Ping timeout: 240 seconds)]
<Turion>
FRidh: Thanks :)
<srhb>
Turion: You probably want one of the configuration x files
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh opened pull request #33204: Python: upgrade package set (master...python-unstable) https://git.io/vbNhV
NixOS_GitHub has left #nixos []
pukkamustard has quit [(Quit: WeeChat 1.9.1)]
pie_ has quit [(Ping timeout: 250 seconds)]
<Turion>
srhb: I was looking at hackage-packages.nix, is that what you mean?
<srhb>
Turion: No, that one is auto generated iirc and should not be edited ever.
<srhb>
Turion: If you're specifically altering the 8.2.x package set, you want pkgs/development/haskell-modules/configuration-ghc-8.2.x.nix
<Turion>
srhb: Ahh, makes a lot of sense. Is there a place where all this is somehow documented?
<srhb>
Turion: I don't recall if there's a lot about the actual infrastructure in the nixpkgs manual and not just how to use it.
<srhb>
Turion: I think most of it is inside petis head.
<srhb>
Which, admittedly, is hard to read given current technological deficits.
ma27 has quit [(Ping timeout: 265 seconds)]
pie_ has joined #nixos
<Turion>
srhb: In the manual I only found stuff about how to write derivations and so, but nothing about how it's organised in nixpkgs (in particular that stuff about automatically generating packages from hackage)
<Turion>
Haha, yes, though I heard they're working on a mind2nix tool, which will automatically create a derivation when you think about a program
asuryawanshi has joined #nixos
<srhb>
Turion: Yep, sorry, that's probably the extent of the documentation.
<srhb>
Oh, great!
<srhb>
:-)
asuryawanshi has quit [(Remote host closed the connection)]
Isorkin has quit [(Quit: Miranda IM! Smaller, Faster, Easier. http://miranda-im.org)]
<Turion>
srhb: Aha, so the .cabal file has a Custom-Setup block that restricts the Cabal (package, not executable) version to <1.25. But it seems like this is disregarded in the automatically generated package.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to staging: https://git.io/vbNjX
<NixOS_GitHub>
nixpkgs/staging 2d0bead Frederik Rietdijk: Merge remote-tracking branch 'upstream/master' into HEAD
NixOS_GitHub has left #nixos []
<Turion>
Do I have to somehow pass the package Cabal with a version restriction or so?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh force-pushed python-unstable from 0604d9e to bcedb4f: https://git.io/v5aIf
<NixOS_GitHub>
nixpkgs/python-unstable 2ec466e Frederik Rietdijk: python: absl-py: 0.1.5 -> 0.1.7
<NixOS_GitHub>
nixpkgs/python-unstable 15ceedc Frederik Rietdijk: python: aenum: 2.0.8 -> 2.0.9
<NixOS_GitHub>
nixpkgs/python-unstable 195a00e Frederik Rietdijk: python: aiohttp: 2.3.3 -> 2.3.7
NixOS_GitHub has left #nixos []
drakonis has quit [(Ping timeout: 252 seconds)]
<Turion>
On a more basic level, why is haskellPackages.cabal-helper not in the binary cache? I'm on nixos-unstable.
<srhb>
Turion: Presumably because it fails to build, which you're trying to remedy.
<srhb>
Turion: You want to override cabal-helper, passing in the older cabal version.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 1 new commit to staging: https://git.io/vbNjN
<NixOS_GitHub>
nixpkgs/staging 5920579 Frederik Rietdijk: Python: disable user site-packages in setup hook...
NixOS_GitHub has left #nixos []
<srhb>
Turion: (Or did I misunderstand your project?)
pukkamustard has joined #nixos
<Turion>
srhb: Yes, that's the plan. Or maybe find out why the correct Cabal version isn't selected by the automatic script
<srhb>
Turion: You mean the script that generates the hackage packages?
<Turion>
srhb: Yes, possibly
<srhb>
Turion: It never does any kind of version selection iirc. It's just the lts set.
<Turion>
srhb: Then something else in the later process incorrectly selects Cabal-2
<srhb>
There's no selection veer, it's just Cabal.
<srhb>
ever*
<srhb>
As far as I know.
<srhb>
So this is exactly the case for overrides.
<srhb>
The set is *supposed* to be internally consistent, and we override when it's not.
<srhb>
Turion: In order (perhaps) to further your understanding: cabal-helper should arguably not even be in the set, since it's not in lts 10, and thus there's no guarantee that it works with the lts 10 version of Cabal (and indeed it doesn't!)
<srhb>
Turion: haskellPackages is basically the lts set (except when it isn't, in this case because we have cabal-helper and lts 10 does not)
ma27 has joined #nixos
drakonis has joined #nixos
<srhb>
Turion: Not sure if that made any sense to you... Do ask if I can help, I think I have a decent grasp of the infrastructure by now :-)
sary has quit [(Ping timeout: 272 seconds)]
vidbina has quit [(Ping timeout: 260 seconds)]
leat has quit [(Ping timeout: 248 seconds)]
sary has joined #nixos
pukkamustard has quit [(Quit: WeeChat 1.9.1)]
zarel has joined #nixos
sary has quit [(Remote host closed the connection)]
<yegortimoshenko>
are tar archives deterministic or do they depend on e.g. gnu tar version?
<yegortimoshenko>
no compression
raynold has quit [(Quit: Connection closed for inactivity)]
<srhb>
yegortimoshenko: You probably need at least --sort
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vbAvw
<NixOS_GitHub>
nixpkgs/master 9fe81b5 Tuomas Tynkkynen: nvidia-x11: Attempt to fix library-only build...
NixOS_GitHub has left #nixos []
<Dezgeg>
check make-bootstrap-tools.nix for how to make it determinisstic
<srhb>
Oh, TIL.
<yegortimoshenko>
i see. so the answer is no by default? the problem is that our glibc currently has in-tree gzipped patches because upstream doesn't distribute them in tarballs
<Dezgeg>
it needs stuff like --hard-dereference --sort=name --numeric-owner --owner=0 --group=0 --mtime=@1
dan_b has joined #nixos
__monty__ has joined #nixos
<__monty__>
How do I use ghcWithHoogle's hoogle?
<__monty__>
And haddocks?
<yegortimoshenko>
and glibc is part of stdenv bootstrap so no fetchurl with postfetch, meaning that remote source should be deterministic, no stripping undeterministic parts in postfetch
<yegortimoshenko>
i see.
asuryawanshi has joined #nixos
Arcaelyx has joined #nixos
verite has joined #nixos
zarel has quit [(Ping timeout: 264 seconds)]
asuryawanshi has quit [(Ping timeout: 240 seconds)]
iyzsong has quit [(Ping timeout: 240 seconds)]
<Turion>
srhb: Ah, I see, thanks! So it would make more sense to maybe automatically get packages from stackage and not hackage. (This is becoming a limerick.) In the meanwhile, I'll try to fix it the way you said :)
willprice94 has joined #nixos
Tuba has joined #nixos
<Turion>
srhb: As soon as nixpkgs' cabal-helper is updated to 0.8, the fix is obsolete. How can I do something like "if the version is <0.8, apply the fix, else don't"?
<srhb>
Turion: Actually, I had second thoughts. I think it might be failing in the configure phase before the buildInputs at all
<srhb>
Turion: I would make an assertion on its own version so that we remember to remove the fix.
<srhb>
Whatever the fix might be
<aanderse_>
i'm having errors when i have a nixos client access a nixos server running subversion via svn+ssh
<aanderse_>
the subversion binary can't be found
asuryawanshi has joined #nixos
<srhb>
Turion: ie. cabal-helper = if super.cabal-helper.version is higher than... assert remove me from file... else fix;
willprice has quit [(Ping timeout: 264 seconds)]
<srhb>
Oh, that's not quite how assertions work. You get the point, I hope :-P
<srhb>
But I'm not sure how peti feels about that. Oh well.. Your suggestion may be better.
<aanderse_>
specific error is:
<aanderse_>
$ svn co svn+ssh://htpc@192.168.0.100/reponame
alexteves has quit [(Quit: My Mac Mini has gone to sleep. ZZZzzz…)]
<aanderse_>
/nix/store/hqi64wjn83nw4mnf9a5z9r4vmpl72j3r-bash-4.4-p12/bin/bash: svnserve -t: command not found
asuryawanshi has quit [(Remote host closed the connection)]
<aanderse_>
seems like the svn+ssh from client to server is making the client not know how to access the svnserve binary
<aanderse_>
because its not located in a usual location
<__monty__>
Turion: Update me on the conversation? I've wondered how to deal with nix's version of cabal hell. Would prefer to not depend on stackage though.
<tilpner>
aanderse_ - I assume you used services.svnserve.enable?
<aanderse_>
tilpner: yes
<tilpner>
aanderse_ - If so, I suggest disabling that, then copying the config portion of svnserve.nix into your local config, so you can experiment with it
jrolfs has joined #nixos
otherjrolfs has joined #nixos
<aanderse_>
ohhh.... i may have found the problem
alexteves has joined #nixos
<tilpner>
... oh, what was it?
<aanderse_>
i also have subversion package installed... but only for one user, not the whole system...
<aanderse_>
that *might* be it
<Turion>
srhb: I don't know how assertions work, but I hope I'll find out soon :)
* aanderse_
changes subversion to subversionClient
<tilpner>
aanderse_ - That would be a bug in nixos, services shouldn't require you to put things into systemPackages manually
orivej has joined #nixos
<aanderse_>
tilpner: i previously had subversion in there, but then just last night enabled the services.svnserve
<srhb>
Turion: I think we need to override setupHaskellDepends
jrolfs has quit [(Ping timeout: 248 seconds)]
otherjrolfs has quit [(Ping timeout: 248 seconds)]
<aanderse_>
no, that didn't fix it
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FRidh pushed 5 new commits to python-unstable: https://git.io/vbAfK
<NixOS_GitHub>
nixpkgs/python-unstable 543ba4d Frederik Rietdijk: python.pkgs.aws-xray-sdk: init at 0.95
<NixOS_GitHub>
nixpkgs/python-unstable 3f81679 Frederik Rietdijk: python.pkgs.boto3: move expression
<NixOS_GitHub>
nixpkgs/python-unstable a605f76 Frederik Rietdijk: python.pkgs.jsondiff: init at 1.1.1
NixOS_GitHub has left #nixos []
jrolfs has joined #nixos
otherjrolfs has joined #nixos
<Turion>
__monty__: I can understand we wouldn't want to depend on stackage. But something in this issue (#32969) is still wrong. Shouldn't the cabal build process just fail and say that the required version of Cabal (library) could not be found? (Instead of this weird type error?)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 2 new commits to master: https://git.io/vbAf1
<NixOS_GitHub>
nixpkgs/master 47c1711 Nick Hu: llvm-hs: fix llvm-config dependency
<NixOS_GitHub>
nixpkgs/master 2c557e1 Peter Simons: Merge pull request #33187 from NickHu/llvm-hs...
NixOS_GitHub has left #nixos []
<srhb>
Turion: I think I agree. The problem is that the error is actually occurring very early in the setup phase.
<Turion>
srhb: Yes. If it was just an ordinary dependency, it probably would just complain about missing dependencies
<srhb>
Indeed.
<srhb>
It's compileBuildDriverPhase, by the way.
<srhb>
Oh, well, right after. setupCompileFlags
<Turion>
srhb: Hmmm, I don't understand enough yet, it seems... :/
<srhb>
I'm not too clear on it myself. It's a bit deep.
<lenzj>
shell.nix file -> with import <nixpkgs> {}; stdenv.mkDerivation rec { name = "TestEnv"; env = buildEnv { name = name; paths = buildInputs; }; buildInputs = [ # revision control git # text editors vim # compilers gcc-unwrapped #(gcc-unwrapped.override { langAda=true; }) # libraries zlib ]; }
<lenzj>
Sorry for formatting
<lenzj>
The shell.nix file works when I use plain vanilla gcc-unwrapped, but when I try the override to set LangAda=true I get errors as follows.
alexteves has quit [(Quit: My Mac Mini has gone to sleep. ZZZzzz…)]
<lenzj>
Hunk #2 FAILED at 193. 2 out of 2 hunks FAILED -- saving rejects to file gcc/ada/gcc-interface/Makefile.in.rej builder for ‘/nix/store/s284rlwl1kqdzq7rsl9l97g10yml499g-gcc-6.4.0.drv’ failed with exit code 1 error: build of ‘/nix/store/s284rlwl1kqdzq7rsl9l97g10yml499g-gcc-6.4.0.drv’ failed /home/lenzj/.nix-profile/bin/nix-shell: failed to build all dependencies
<lenzj>
I'm not sure if I have a bad shell.nix configuration or if this is a bug?
<lenzj>
tilpner - Thanks for the reply! I think you are right. Looks like I'm not the only one with the issue. I somehow missed that when searching earlier.
<lenzj>
tilpner - I have limited experience with nix package development, but I will focus my energy on trying to help fix the package rather than my "shell.nix" setup. Thanks again.
Turion has quit [(Ping timeout: 256 seconds)]
<tilpner>
lenzj - Start by looking at why gnat-cflags.patch fails to apply. You can keep a failed build with -K, then look at the rejects
Turion has joined #nixos
<lenzj>
tilpner - Thanks for the tip. Will do!
<tilpner>
Running nix-build -KE 'with import <nixpkgs> {}; gcc-unwrapped.override { langAda = true; }' prints note: keeping build directory '/tmp/nix-build-gcc-6.4.0.drv-1' near the end, in which we find gcc/ada/gcc-interface/Makefile.in.rej
otherjrolfs has quit [(Ping timeout: 240 seconds)]
<sphalerite>
I'm having some difficulty getting a bluetooth keyboard to work on start-up — the logs complain about "No agent available for request type 0" and "device_request_pin: Operation not permitted". It works fine if I run bluetoothctl and power the bluetooth controller off and on again
<sphalerite>
anyone know how I can get it to work right away?
jb55 has joined #nixos
deech` has joined #nixos
<jeaye>
LnL: I'll try that additional-nix-code approach again.
<LnL>
to clarify, you need both otherwise the custom part will get removed next time we run the update script
<srhb>
Turion: The more I think about it, the more I don't think it makes sense to override this on a package level. Cabal-the-library is rather tied to ghc.
<srhb>
Turion: Which is probably why it's so difficult, while it's so easy to override for an entire set.
otherjrolfs has joined #nixos
jrolfs has joined #nixos
<deech`>
Hi all, how do I install X11 development headers? I'm using the Nix package manager only and have a GHC dev environment setup with "nix-shell '<nixpkgs>' -A haskell.compiler.ghcHEAD". I also need X11 headers so I did 'nix-env -i libX11' but that only downloaded the libraries.
nico202 has joined #nixos
dieggsy has joined #nixos
FRidh has quit [(Quit: Konversation terminated!)]
jb55 has quit [(Ping timeout: 252 seconds)]
<srhb>
deech`: To answer your direct question, it's libX11.dev (in xorg.libX11.dev) but you probably don't want to be mixing nix-env and nix-shell like that.
<srhb>
At least, I find that leads to lots of confusion.
<mpickering>
I think it is fine to override Cabal?
<srhb>
mpickering: It is, when it's a build input.
jrolfs has quit [(Ping timeout: 250 seconds)]
<deech`>
srhb: I'm a total noob. I got this far from your ( I think ) Reddit post this morning. :)
lenzj has quit [(Quit: Page closed)]
<srhb>
deech`: Ah, yes, that was me. :-)
<srhb>
deech`: Can I ask what you're actually trying to do?
<mpickering>
deech`: If you want to do anything more complicated then you should write a shell.nix
jrolfs has joined #nixos
<srhb>
deech`: I don't think GHC building requires X11 normally, does it?
<srhb>
deech`: (Or well, not my post, but my comment...)
<srhb>
mpickering: The problem here is that the compileBuildDriverPhase uses a ghc with a packagedb that is entirely unconfigurable
<srhb>
mpickering: Which... I *think* makes sense?
<deech`>
srhb: I'm trying to track down a compile time perf. issue in GHC and I have a test program that needs the X11 headers.
<srhb>
mpickering: I may be mistaken though!
<srhb>
deech`: In that case it would seem to me you would want a development that knows about both your program and your ghc sources.
<srhb>
Er, a development environment.
<srhb>
And your program knows about the X11 headers.
<mpickering>
do you actually want to build ghc for that deech`?
<mpickering>
The last few lines in this file is one way to augment the ghcHEAD derivation with some additional build inputs
alexteves has joined #nixos
<deech`>
srhb, mpickering: Yes I think I want that instead of what I have now.
<mpickering>
__monty__: What is "nix's version of cabal hell"?
jrolfs has quit [(Ping timeout: 248 seconds)]
otherjrolfs has quit [(Ping timeout: 268 seconds)]
<mpickering>
ah I see srhb, cabal-helper is a bit of a strange package anyway I think
<__monty__>
mpickering: Having packages not build. More of an issue when my haskell environment was on nixpkgs. Switched to nixos and only update the channel occasionally now.
rogue_koder has quit [(Ping timeout: 256 seconds)]
<srhb>
mpickering: Agreed. But I am surprised because it would seem that ghc *could* build that Setup.hs if only the right Cabal version were in the package db.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Lassulus opened pull request #33210: bitlbee-discord: init at 2017-12-27 (master...bitlbee-discord) https://git.io/vbAIs
NixOS_GitHub has left #nixos []
jtojnar has joined #nixos
<mpickering>
That's not what cabal hell ever was but anyway
<mpickering>
I'm not sure what the solution is for cabal-like solver workflows
<srhb>
cabal2nix all the things?
<mpickering>
because you lose package caching if you are generating custom derivations for each dep
<srhb>
True.
<__monty__>
mpickering: I know, hence why I called it nix's version of cabal hell. It's equally annoying to me.
<srhb>
Cabal hell cannot occur in nix though.
<srhb>
But whatever. :-P
<__monty__>
It's not cabal hell, it's nix's version of cabal hell. It has little to do with cabal.
* srhb
thinks the term has been watered down by people having access to such a wealth of package management solutions
<srhb>
:-)
jrolfs has joined #nixos
otherjrolfs has joined #nixos
lfish has joined #nixos
<srhb>
mpickering: It's possiable to work around it by manipulating PATH in nasty ways during preCompileBuildDriverPhase but eeek.
<srhb>
Clearly a lot of effort has been put into making it hard for me to shoot myself in the foot here...
<__monty__>
The problem is I can't update the channel my haskell environment uses willy nilly because there's always a possibility that it won't work. (rollbacks sure, but what about updating being fine and then adding a package to your environment that doesn't build? Now you're screwed because you have to rollback one or more generations until you find one where all the packages you need build.)
<srhb>
Right, broken packages is annoying.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jtojnar pushed 1 new commit to master: https://git.io/vbAIu
<NixOS_GitHub>
nixpkgs/master 9db8e22 Sébastien Maccagnoni: gnomeExtensions.mediaplayer: init at 3.5
NixOS_GitHub has left #nixos []
nico202 has quit [(Ping timeout: 268 seconds)]
<__monty__>
That's what I refer to as nix's version of cabal hell because it's annoying technical details relating to a package manager/build tool that prevent me from building them sweet sweet haskell programs.
jrolfs has quit [(Ping timeout: 240 seconds)]
<srhb>
I think that's misleading, but I understand the frustration. :)
<__monty__>
Still not clear is 821 being removed from the channels or just the caches?
<__monty__>
And why?
<srhb>
__monty__: That's the same thing.
<jeaye>
LnL: I move the buildInputs and buildPhase into additional-nix-code and any rebuilds I do, even with changes to the buildPhase, don't rebuild anything.
<srhb>
__monty__: Once no channel is built that contains it it will eventually be garbage collected.
<jeaye>
LnL: On the contrary, the same changes within default.nix will cause a rebuild.
<__monty__>
Does this mean eventually we won't be able to install 821 with nix?
<srhb>
__monty__: Not unless you define your own 821 package set.
<srhb>
Er, yes unless*
otherjrolfs has joined #nixos
jrolfs has joined #nixos
Mic92 has quit [(Ping timeout: 250 seconds)]
<jeaye>
LnL: Furthermore, if I move only the buildInputs to additional-nix-code, for example, and leave the buildPhase in default.nix, the build fails, since the build inputs aren't considered.
<__monty__>
srhb: But nixpkgs still contains 6.10.4, 7.0.4, 7.2.2, etc.?
<Turion>
srhb: Yes, it's hard overriding the library Cabal. I tried overriding setupHaskellDepends, same error as the original bug.
jrolfs has quit [(Ping timeout: 252 seconds)]
otherjrolfs has quit [(Ping timeout: 260 seconds)]
<srhb>
Turion: Yes, it's even earlier.
<srhb>
Turion: It's the bootstrapping of Setup.hs which is in no way configurable.
<__monty__>
Oh, those can generate a package set based on older LTSs? Nice.
Tuba has quit [(Ping timeout: 276 seconds)]
<andromeda-galaxy>
afaik stackage2nix will build an entire lts packages sets
<andromeda-galaxy>
*set from an older lts release, and stack2nix will build only 1 derivation that includes all the necessary overrides for a particular project to use a particular version
<srhb>
Yep.
<__monty__>
Ok, good to know. Don't need it right now but having the option soothes the mind : )
<srhb>
But I mean, getting older package sets is never really an issue as you can just go back in history with git. :)
<srhb>
Not that I think it's a bad tool at all.
<srhb>
It's just that there's automatic updates and stability: Pick one. :-P
<srhb>
(Hence why people who choose the latter tend to pin nixpkgs a lot in their projects..)
otherjrolfs has quit [(Ping timeout: 268 seconds)]
jrolfs has quit [(Ping timeout: 272 seconds)]
<srhb>
Ugh, I'm butchering english today. Sorry. :-)
<__monty__>
So I ran hoogle generate in my ghc821 environment and this is part of the output: http://ix.io/Dy2
<__monty__>
Why is it downloading ghc 8.0.1 docs?
pukkamustard has quit [(Quit: WeeChat 1.9.1)]
wchresta has joined #nixos
<srhb>
__monty__: I've no idea how to use hoogle manually, but is there any reason to not use ghcWithHoogle?
<__monty__>
But I am using ghcWithHoogle.
<srhb>
Why did you run hoogle generate then?
<srhb>
I didn't think that was necessary.
<__monty__>
I want to serve the hoogle page locally so I can search hoogle in the browser.
<__monty__>
And hoogle server -p 8080 said no database, run hoogle generate.
<srhb>
That sounds wrong for sure.
<srhb>
How did you use ghcWithHoogle?
otherjrolfs has joined #nixos
jrolfs has joined #nixos
<__monty__>
This is my config.nix, the ghc82Env is the relevant part: http://ix.io/Dy6
<__monty__>
I'm not in a nix-shell, I'm in a myEnvFun environment.
<mpickering>
In this incremental package builds talk, he says that he doesn't want .o files in the nix store, why is that? seems like the correct approach to me
<srhb>
mpickering: I was confused about that, too.
<srhb>
WEll, am.
<srhb>
__monty__: Strange. I don't see why that shouldn't work.
nico202 has quit [(Read error: Connection reset by peer)]
<__monty__>
srhb: Could the removal of ghc821 have anything to do with it?
<srhb>
__monty__: Nope.
<mpickering>
his solution is instead to have a separate store with different semantics and require every build system to deal with the caching themselves hmm
<tilpner>
mpickering - I have played with a very thin Nix wrapper for a small project, putting every .o file into the store. It's worked fine so far, but then again it's not very big. IIRC some filesystems degrade in performance drastically once you put a few million additional files into a single directory
<srhb>
__monty__: So, you're nix-env -iA'ing your environment, then load-env-ing it and hoogle serve does not work?
<tilpner>
mpickering - The separate store also solves concerns regarding project-level control of the GC. If every project shared the store, one would need additional mechanisms to collect build artifacts of just one project (though this sounds quite easy, now that I say it)
moet has joined #nixos
<srhb>
__monty__: Works for me!
<__monty__>
srhb: Exactly.
<srhb>
__monty__: What commit are you on?
<mpickering>
tilpner: consider building Haskell libraries, all the .o files are stored in the store anyway but in a directory
<__monty__>
srhb: How do I check? .nix-channels just has nixos-unstable.
<srhb>
__monty__: Are you on NixOS?
<__monty__>
Nope.
<srhb>
Oh, hang on, there's gchristensen s tool
<mpickering>
he is talking a bit about multiple stores as well
<srhb>
nix-shell -p nix-info --run nix-info
<mpickering>
which makes sense
<tilpner>
mpickering - I'm not familiar with how Nix handles Haskell libraries. Are they stored directly as /nix/store/*.o, or like /nix/store/*/**.o?
<__monty__>
Based my environment on nixos channel because it broke less than nixpkgs.
<srhb>
__monty__: Yep, the channels there are more thoroughly tested.
<mpickering>
the latter, you have one entry into the store for each package and then the .o files are in some subdirectory
<__monty__>
Is that still a good idea or should nixpkgs be fine to base the environment on now?
<srhb>
__monty__: The tests are still more extensive on nixos channels
<__monty__>
And the only drawback is they move slower?
jrolfs has quit [(Ping timeout: 248 seconds)]
<srhb>
__monty__: Well, yes. The tests are mostly nixos vms :)
otherjrolfs has quit [(Ping timeout: 256 seconds)]
<__monty__>
nixos-18.03pre120540.b8f7027360
<tilpner>
mpickering - In my project, every .o files is stored separately (/nix/store/*.o), which I assume is what layus was trying to avoid
<srhb>
__monty__: Lemme try that.
moet has quit [(Ping timeout: 265 seconds)]
randomstrangerb has quit [(Ping timeout: 268 seconds)]
jb55 has joined #nixos
jrolfs has joined #nixos
otherjrolfs has joined #nixos
randomstrangerb has joined #nixos
<mpickering>
yes, I'm trying to say that you won't get many more files in the store, they are already there but in subdirs for some derivations
<srhb>
__monty__: Oh, I notice right now that 17.09 is on 8.0.2, not 8.2.1
<mpickering>
ah you were talking about a single directory
<srhb>
__monty__: ie. the caches are gone.
<dhess>
Is Darwin sandboxing safe to turn on again?
<dhess>
In Nix/Nixpkgs master?
<mpickering>
the fact the store is currently in a single directory is an implementation details surely?
zzamboni has joined #nixos
<__monty__>
srhb: I didn't have caches to begin with because the ghc version didn't match the default. At least I think so.
rihards has joined #nixos
JosW has quit [(Quit: Konversation terminated!)]
* srhb
nods
<srhb>
Takes me a while. :-P
<__monty__>
Caches aren't a dealbreaker for me since I set up a local build server to do all the nix building.
<srhb>
Yeah, same.
otherjrolfs has quit [(Ping timeout: 248 seconds)]
jrolfs has quit [(Ping timeout: 264 seconds)]
<srhb>
Wait, what's the default haskell set for you then?
<srhb>
__monty__: Ah, I see
<srhb>
__monty__: You're using haskell.packages.ghc802.ghcWithHoogle
<LnL>
jeaye: yes, the additional-nix-code part is only used by the update script
<srhb>
Or.. not
<srhb>
No, sorry.
<__monty__>
Yeah, I should be using 821, right?
<srhb>
Yes, indeed. Sorry.
jrolfs has joined #nixos
otherjrolfs has joined #nixos
<srhb>
Building just the hoogle dependencies is like time travelling to before I even used Nix. x_x
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] gebner pushed 1 new commit to master: https://git.io/vbAmK
<NixOS_GitHub>
nixpkgs/master a9f1189 Gabriel Ebner: electrum: enable support for qr codes
NixOS_GitHub has left #nixos []
taktoa has joined #nixos
jb55 has quit [(Ping timeout: 252 seconds)]
otherjrolfs has quit [(Ping timeout: 268 seconds)]
jrolfs has quit [(Ping timeout: 260 seconds)]
<srhb>
__monty__: Works for me with (pkgs.haskell.packages.ghc821.ghcWithHoogle (ps: with ps; [ warp ]))
<srhb>
__monty__: Do you perchance have more than one hoogle in scope?
<jeaye>
LnL: I don't quite follow. What am I meant to do?
<__monty__>
Ah, yes. I have a hoogle in ~/.local. Can't remember where that came from though.
<srhb>
__monty__: Ah, okay. :)
_alice_ has joined #nixos
<LnL>
jeaye: what you did looks good, but don't remove it from vim-plugins
<jeaye>
LnL: I see other packages using llvmPackages.clang.cc without any optional for darwin. Do you have a suggested approach for fixing the darwin build error?
<srhb>
__monty__: It's glorious! :-)
<jeaye>
Perhaps just taking llvmPackages as a build input might fix it, instead of llvmPackages.clang.cc?
jrolfs has quit [(Ping timeout: 248 seconds)]
<Turion>
srhb: So the problem is that we have no way to handle `Custom-Setup` sections? And which part is the inflexible one? The automatically generated derivation? Or nix?
<Turion>
Because if it's just the automatically generated derivation, we can copy that single one and modify it
<__monty__>
srhb: What's your solution to not being able to navigate from http: to file: though?
<srhb>
Turion: We can handle custom setup in general. The problem is that adding an older Cabal version to the setup packages will not make ghc pick it.
<srhb>
__monty__: Not understood?
<srhb>
Turion: If you could inject an argument to the ghc Setup.hs invocation (--package Cabal-1...) I think it would work
<__monty__>
srhb: I can't click on the links on the local hoogle page because firefox doesn't allow navigating from an http:// page to a file:// page.
<srhb>
Turion: But as far as I can see there's no nice way of doing that other than hacking the internals of the generic builder in very ugly ways.
otherjrolfs has joined #nixos
jrolfs has joined #nixos
<Turion>
srhb: If I just build the package by hand using cabal build, it just works. Does nix not invoke cabal when building haskell packages?
fendor has joined #nixos
<srhb>
Turion: I think it uses Cabal directly, not cabal-install.
<srhb>
__monty__: Oh, right, chrome has the same issue.
<srhb>
__monty__: Doesn't hoogle serve them if you specify --local ?
<__monty__>
srhb: Ah yes, thanks.
srdqty has joined #nixos
<LnL>
jeaye: it's failing to find headers because of the unwrapped clang, is there some cmake flag to configure the libclang path?
<jeaye>
LnL: There is; right now, I'm leaving it to CMake to find where things are, since that has been the most reliable, in my testing.
jrolfs has quit [(Ping timeout: 272 seconds)]
otherjrolfs has quit [(Ping timeout: 272 seconds)]
<jeaye>
LnL: How do you see the full errors?
phreedom has quit [(Remote host closed the connection)]
<jeaye>
Thank you. Looks like it's failing before it even looks for clang.
<LnL>
yeah that's because of the unwrapped clang
<LnL>
that shadows the one from the stdenv
<jeaye>
Ok, so not related to other libs using libpthreadstubs?
<jeaye>
I know very, very little about the nix/nixpkgs infrastructure, so I appreciate the patience.
<LnL>
if we had a separate lib output for clang we could only include that and cmake would find the libraries correctly
<jeaye>
So this may require a change to the way clang libs are packaged?
<LnL>
yeah, so replacing the input with an explicit cmake flag is an easier solution for now
<jeaye>
I see.
<velovix>
Hi, I'm trying to create a package for an application that depends on docbook-xml. A call to "nix-env -qa | grep "docbook-xml"" shows that docbook-xml is a package, but if I add it as an argument to my default.nix, it isn't passed automatically and I get a "called without required argument" error. I'm not sure what to put in the all-packages.nix to fix this
__monty__ has joined #nixos
<velovix>
I get the idea that docbook-xml isn't a standalone package in the same way other kinds of packages are, sinc eit seems to be created in the "docbook" directory instead of its own "docbook-xml" directory. Not sure if that matters.
<LnL>
the attribute name might be different, nix-env -qaP will include that
<hyper_ch>
I switched from unstable to unstable small... it gets updated quicker but sometimes results in building stuff on your own like libreoffice or firefox
<samueldr>
(if that was actual data you needed)
stanibanani has joined #nixos
<flokli>
__monty__: just file a pr, sure :-)
stepcut has quit [(Read error: Connection reset by peer)]
<hyper_ch>
makefu: awake?
<makefu>
hyper_ch: just saw you are also lurking in #wireguard
<hyper_ch>
makefu: yes, why?
<hyper_ch>
btw, we need to update vpn docs for ip forwarding and masquerade
<makefu>
i also wanted to set up wireguard for my router
<hyper_ch>
this is for openvpn and wireguard
<hyper_ch>
so not sure where to set this in the wiki
<hyper_ch>
and then link back from the individual articles?
jrolfs has quit [(Ping timeout: 272 seconds)]
<Turion>
srhb: I see. "cabal get cabal-helper && cd cabal-helper-0.7.3.0 && ghc Setup.hs" doesn't work with ghc 8.2.1
<hyper_ch>
ip forwarding and masquerading is needed by both.... openvpn and wireguard... so question is where to add it best...
jrolfs has joined #nixos
otherjrolfs has joined #nixos
<hyper_ch>
adding it to both articles seems kinde meeeehe
<velovix>
Is anybody familiar with how xmlcatalog is dealt with in Nix? There's this findXMLCatalogs tool but I'm not entirely sure what it does or how I should use it
raynold has joined #nixos
<hyper_ch>
makefu: oh, forgot to name the invidivudal coonections
<makefu>
hyper_ch: maybe in some kind of "network routing" or "network setup"
<makefu>
& article
aanderse has joined #nixos
<hyper_ch>
ah secing in wg and openvpn article about routing/setup and then just have it in the main vpn article?
<hyper_ch>
s/secing/section/
<makefu>
yes that is a good idea
dywedir has quit [(Remote host closed the connection)]
<makefu>
nixos-on-your-router seems to be missing though
<hyper_ch>
makefu: added it to wireguard
<flokli>
it's pretty new :-D
<hyper_ch>
makefu: wireguard is pretty easy to setup and performs better than openvpn as far as I can tell
<flokli>
makefu: added the channel
<hyper_ch>
drawback is on an old centos the kernel is too old; windows and os x, ios currently not supported;
<hyper_ch>
but on linux and android it works fine
<hyper_ch>
if it's security-wise au pair with openvpn - I can't tell
otherjrolfs has quit [(Ping timeout: 256 seconds)]
jrolfs has quit [(Ping timeout: 264 seconds)]
wchresta has joined #nixos
<hyper_ch>
makefu: already adjusted your cofniguration nix?
blym has quit [(Read error: Connection reset by peer)]
hiratara has quit [(Ping timeout: 265 seconds)]
hiratara has joined #nixos
<hyper_ch>
makefu: instead of wg0 etc... you can name the interface device as you want... I use wg_home and wg_office
<jeaye>
LnL: I don't think I'm able to tell the bot to rebuild, but I've made the suggested changes, as much as possible (the cmake stuff can't change until your PR goes through) -- darwin may build now.
_alice_ has quit [(Remote host closed the connection)]
vidbina has quit [(Ping timeout: 265 seconds)]
<makefu>
hyper_ch: there was a talk at the 34c3 and it seems to be using quite recent crypto
<LnL>
jeaye: ah yes
<hyper_ch>
makefu: I only watched the car charging hacking one
<jeaye>
LnL: Thanks.
<hyper_ch>
it's so simple that basically also non-tesla drivers can recharge for free :)
otherjrolfs has joined #nixos
jrolfs has joined #nixos
<makefu>
it was a self-hosted session actually, so no recording anyway
<makefu>
i talked with a friend with a tesla and he said that charging is free currently anyway because charging is heavily supported by the state
<makefu>
(germany)
<makefu>
however it will change at some point of course
<hyper_ch>
makefu: well, the charging points and their protocols are so simple
<hyper_ch>
e.g. they use a nfc card and all that is checked is the plaintext card id
<makefu>
i have this talk on my backlog actually
<makefu>
professionals!
<hyper_ch>
also at some charging stationgs you can simple insert a usb key and it gives all necessary data
<hyper_ch>
and root shell
<hyper_ch>
and you can get like the last 100 valid card id stored on your usb
<makefu>
i just finished the talk about the apollo guidance computer, which was fantastic
<hyper_ch>
so just fake the card and use one of those ids and charge for free :)
<makefu>
they speak a bastardisation of german there, yes
<makefu>
(schwäbisch)
<hyper_ch>
and when will BaWü file request to join the Helvetic Republic?
<hyper_ch>
s/Republic/Confederation
jrolfs has quit [(Ping timeout: 240 seconds)]
otherjrolfs has quit [(Ping timeout: 240 seconds)]
<makefu>
not so soon it seems
<makefu>
;)
<hyper_ch>
we urgently need a Kanton Ausland :)
<Turion>
Different question. I'm using xterm in xmonad these days. It somehow swallows every "W" I type. Does anyone else have that problem?
<jeaye>
Are any folks here using a startx/xlaunch solution? I really want to boot into tty, but I don't want to log in twice (first into tty, then into a display manager).
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] yegortimoshenko created yegortimoshenko-patch-1 (+1 new commit): https://git.io/vbAG3
<Turion>
Another question. How do I find out which haskell package in my systemPackages pulls in cabal-helper?
<Turion>
(Since that one fails to build)
orivej has quit [(Ping timeout: 256 seconds)]
drakonis has quit [(Remote host closed the connection)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] yegortimoshenko opened pull request #33215: libinput: disableWhileTyping = false by default (master...yegortimoshenko-patch-1) https://git.io/vbAGR
NixOS_GitHub has left #nixos []
drakonis has joined #nixos
<makefu>
Turion: only in xterm the W is swallowed?
jrolfs has joined #nixos
otherjrolfs has joined #nixos
LexLeoGryfon has joined #nixos
shlevy has quit [(Ping timeout: 255 seconds)]
wchresta has quit [(Ping timeout: 252 seconds)]
dywedir has quit [(Remote host closed the connection)]
dywedir has joined #nixos
shlevy has joined #nixos
aanderse_ has joined #nixos
jrolfs has quit [(Ping timeout: 265 seconds)]
otherjrolfs has quit [(Ping timeout: 256 seconds)]
aanderse has quit [(Ping timeout: 272 seconds)]
<velovix>
Is there some kind of generally accepted way to create a temporary file at build time for a Nix package?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg closed pull request #32641: release-lib: evaluate nixpkgs on armv6l and armv7l (master...arm-release-lib) https://git.io/vb2Dp
NixOS_GitHub has left #nixos []
jrolfs has joined #nixos
otherjrolfs has joined #nixos
aarvar has joined #nixos
<srhb>
velovix: I think people just tend to create it in the build environment and just don't copy it over.
aanderse has joined #nixos
<srhb>
Turion: Honestly, if you're bent on fixing this, you could either override the source of the package to the github version that works or patch Setup.hs
<srhb>
Turion: We have mechanics in place to easily do that.
<velovix>
srhb: Is there a variable I can use that represents the build directory? I assume you're referring to the /tmp/<my-sha-hash>-<package>-<version> directory, right?
jrolfs has quit [(Ping timeout: 248 seconds)]
otherjrolfs has quit [(Ping timeout: 248 seconds)]
aanderse_ has quit [(Ping timeout: 268 seconds)]
<srhb>
velovix: I never thought about it! I tend to just litter cwd, wherever that is.
<srhb>
I'm a bad person.
<srhb>
:-)
<velovix>
srhb: hey, if there's precident for it, that's good enough for me :)
<Turion>
srhb, the issue is medium important to me, it just blocks me from updating my system
<Turion>
I was just hoping for a first easy issue for me to resolve :/
<srhb>
Turion: You can, give me a sec to grab a coffee
<srhb>
and sorry I misunderstood your intention :/
<srhb>
I thought you wanted that particular fix.
hiratara has quit [(Remote host closed the connection)]
<srhb>
(Which is difficult)
hiratara has joined #nixos
<Turion>
srhb: Haha no worries, and thanks for your help so far :) The two things I want are: 1. An updated system 2. Writing a nice solution, submitting it as my first PR to nixpkgs and learning something about nixos in the meantime
<srhb>
Turion: You said it worked with the latest version from github, right?
<srhb>
If so, we can just override the src attribute
aanderse_ has joined #nixos
aanderse has quit [(Ping timeout: 248 seconds)]
otherjrolfs has joined #nixos
<srhb>
Turion: You will need something like... cabal-helper = overrideCabal super.cabal-helper (drv: { src = pkgs.fetchFromGitHub { ... }; buildTools = [ self.Cabal_1_24_2_0 ]; })
b has quit [(Quit: leaving)]
b has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] toonn opened pull request #33217: haskell docs: add info about hoogle's --local flag to section 9.5.2.3 (master...master) https://git.io/vbAZd
NixOS_GitHub has left #nixos []
<__monty__>
: o That's me!
<srhb>
Woot!
<srhb>
Wow, that was very zeroes of me..
otherjrolfs has quit [(Ping timeout: 240 seconds)]
<jeaye>
When I do a nix-shell into an environment with a src from fetchgit/path/url, is it fetched for me somewhere, or do I need to manually do it?
<srhb>
jeaye: unpackPhase will do it :)
<srhb>
jeaye: That is, if you invoke it.
<jeaye>
srhb: Awesome. Thank you.
jrolfs has joined #nixos
otherjrolfs has joined #nixos
<srhb>
__monty__: Nitpick on your PR, I hope you don't mind... :)
<velovix>
Is there a way to make sure a setup hook runs after another setup hook? I tried adding the hook as a dep, but I don't think it had the desired effect
<__monty__>
Not at all.
<clever>
jeaye: nix-shell will fetch it, and point $src to the path that was fetched, unpackPhase copies&unpacks it to the cwd
<jeaye>
Looks like it can't unpack my tar.bz2. Might need another input.
b has quit [(Ping timeout: 256 seconds)]
<jeaye>
Hm, even with the bzip2 buildInput, it can't.
<clever>
what args are you running nix-shell with?
<jeaye>
Ah, I just get a bunch of "tar: foo/COPYING: Cannot change ownership to uid 500, gid 500: Permission denied"
<jeaye>
clever: `nix-shell my-dv.nix`
<clever>
ah, thats safe to ignore
<jeaye>
Ok, then I see "do not know how to unpack source archive /nix/store/...my.tar.bz2"
jrolfs has quit [(Ping timeout: 256 seconds)]
<jeaye>
When I wget & tar xvf it manually, it unpacks just fine.
<clever>
can you gist my-dv.nix?
<jeaye>
Yep.
otherjrolfs has quit [(Ping timeout: 256 seconds)]
<jeaye>
clever: Hm, seems like it was only in my nix-shell. With nix-build, the bz2 unpacks just fine. I think it's the directory I was in, trying to unpack.
<dhess>
it's just a set of overlays
<dhess>
oh weird, it's coming from license.boost
<dhess>
that is bizarre
<srhb>
Uh, what? This one? mathics only supports django-1.8.x
<dhess>
yep
<dhess>
no worries, it's just a weird aliasing thing going on
<srhb>
Interesting. :-)
<dhess>
because of multiple "with" statements and the 'boost' argument being passed into the derivation
<srhb>
Ah.
<dhess>
all resolved now. Wow that was a weird one.
<srhb>
with can do that. :-P
<jeaye>
clever: Why would I see "no configure script, doing nothing" when there's very clearly a configure script?
<clever>
jeaye: did you cd into the directory with the configure script?
<jeaye>
clever: i.e. I can `mkdir $out; cd $out; unpackPhase; ./configure && make` with no problem, but nix-build doesn't see it.
<clever>
jeaye: you have to cd into the output directory after unpackPhase
dywedir has quit [(Ping timeout: 268 seconds)]
<jeaye>
clever: So, I need a preconfigurePhase to just cd somewhere?
otherjrolfs has quit [(Ping timeout: 256 seconds)]
<jeaye>
I haven't seen other derivations doing that.
<clever>
jeaye: when the stdenv has finished unpackPhase, it will cd for you
<clever>
but if you manually run the unpackPhase, it wont
<clever>
jeaye: try instead just runnign genericBuilder, with nothing else, dont even mkdir or cd $out
<jeaye>
My error is with nix-build, so it's not manually being run.
rihards has quit [(Quit: rihards)]
<jeaye>
I think my wording has been unclear, so we're not on the same page.
<clever>
jeaye: can you gist the output of "tar -tvf $src" ?
<jeaye>
Hold on and hear me out, please.
stepcut has quit [(Remote host closed the connection)]
<jeaye>
The issue here, I'm quite certain, is my (lack of) understanding of what nix should be doing for me when building, not the archive I'm downloading and unpacking.
stepcut has joined #nixos
<clever>
jeaye: i want to confirm what path configure is at
<jeaye>
clever: When I have a stdenv.mkDerivation with some configureFlags set, nix should download my bz2, unpack it, look for a configure file, run it with the flags if it exists, look for a make file, and run make if there is one. Right?
<clever>
yeah
Mic92 has joined #nixos
vidbina has joined #nixos
<jeaye>
clever: I shouldn't need to do any build phase or unpack phase or any other phase work, if it's in the standard layout, right?
<jeaye>
Assuming I'm just doing `nix-build my-drv.nix`
<clever>
yeah
<jeaye>
Ok, that was my expectation as well. The configure script is top-level when extracted, adjacent to the make file that's generated.
<jeaye>
clever: Also, now that I interpret what you're saying a different way, there is a top-level directory within the tarball, so te contents aren't extracted to ./ they go to xinit-version/
<clever>
jeaye: yeah
<jeaye>
So I'll need to cd into that in some phase?
<clever>
after the unpackPhase, and the stdenv goes it for you
<clever>
does*
<jeaye>
Right, you linked that. Ok, well hopefully you can reproduce.
jrolfs has quit [(Ping timeout: 248 seconds)]
otherjrolfs has quit [(Ping timeout: 240 seconds)]
stepcut has joined #nixos
stepcut has quit [(Remote host closed the connection)]
stepcut has joined #nixos
<__monty__>
srhb: How much explanation should I add? Just added a sentence saying the file links are served over http.
<__monty__>
srhb: Also, maybe I should just update the command and then explain older versions don't have it and there are workarounds in the text? Since people are more likely to use the recent version and most don't bother reading carefully.
taktoa has quit [(Quit: No Ping reply in 180 seconds.)]
<srhb>
__monty__: That's a good idea. I still think it's good to let people know they're serving files from the nix store. They might have secrets in there :)
<srhb>
And I don't know how well sandboxed it is.
<__monty__>
The current solution isn't _more_ secure, surely?
<srhb>
__monty__: The current solution does not serve files from the nix store at all.
taktoa has joined #nixos
<srhb>
(as far as I know)
jb55 has quit [(Ping timeout: 240 seconds)]
<__monty__>
Oh, because someone wouldn't be able to use the file:// uris if they didn't have access to the machine in the first place?
<srhb>
__monty__: Right :)
<srhb>
__monty__: Maybe I'm being silly. I just think it's a no-cost to mention it briefly. :-P
jb55 has joined #nixos
lassulus has quit [(Quit: WeeChat 1.9.1)]
lassulus has joined #nixos
vidbina has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vbAWB
<NixOS_GitHub>
nixpkgs/master b81de99 Jörg Thalheim: pythonix: mark it as a python module...
NixOS_GitHub has left #nixos []
<jeaye>
clever: Gah, figured it out.
thc202 has quit [(Ping timeout: 268 seconds)]
<jeaye>
My /tmp was mounted noexec, which had never been an issue until building this.
<clever>
i would expect that to break building of anything
<__monty__>
srhb: It's mentioned. Waiting on a reply from jb55 to find out if/how I should mention the hoogle module.
<srhb>
__monty__: Great! :)
<jeaye>
clever: Most everything I get comes precompiled, it'd seem.