<nkaretnikov>
elvishjerricco: i've tried adding haskellPackages.ghcjs-dom to executableHaskellDepends, but it complains about missing socket-io attribute for some reason. any idea how to fix this?
<clever>
Lisanna: first, try calling it something other then $out, and then echo both placeholder and $out
<nkaretnikov>
elvishjerricco: the error itself is: `error: attribute 'socket-io' missing, at nixpkgs/pkgs/development/haskell-modules/with-packages-wrapper.nix:74:34`
<nkaretnikov>
elvishjerricco: ^ the code in that file
<Lisanna>
clever okay, so if I want to do it in a runCommand I just need to make sure I reference it through an attribute
<clever>
Lisanna: yeah
<Lisanna>
that's a bit limiting unfortunately, but it's okay, I found a different way to do the actual thing I was trying to do
stepcut has joined #nixos
<clever>
Lisanna: what are you wanting to do?
woodson has quit [Ping timeout: 260 seconds]
<Lisanna>
clever running simics (it's like qemu) in runCommand, trying to get $out into the shell of the virtual host
<clever>
Lisanna: ah
<Lisanna>
I found an easier way to pass the value tho
drakonis has quit [Ping timeout: 245 seconds]
oltoAltn has joined #nixos
olto has quit [Ping timeout: 240 seconds]
camsbury has quit [Quit: WeeChat 2.0]
kerrhau has left #nixos ["WeeChat 2.1"]
thc202 has quit [Ping timeout: 260 seconds]
lassulus_ has joined #nixos
woodson has joined #nixos
blankhart has quit [Quit: WeeChat 1.9.1]
lassulus has quit [Ping timeout: 252 seconds]
lassulus_ is now known as lassulus
<elvishjerricco>
nkaretnikov: Looks like we need to add socket-io to the GHCJS derivation
<nkaretnikov>
elvishjerricco: i've also tried reflex-dom and it complains similarly about oldtime
<nkaretnikov>
elvishjerricco: i'm now trying the reflex-platform. perhaps it's better
<elvishjerricco>
nkaretnikov: reflex-platform is at least stable at the moment
<elvishjerricco>
your problems with ghcjs80 and ghcjs710 were because I was dumb and forgot to change the compilers in their package sets from the old names
<elvishjerricco>
Will push fixes for both shortly
<nkaretnikov>
elvishjerricco: i don't have any library preference and don't really know what's considered best practices in the haskell js world. i just need something with which i can draw shapes on the screen and accept mouse/keyboard events. iiuc, the reflex platform gives me all this in a nice package thanks to the bundled libraries, correct?
blankhart has joined #nixos
<elvishjerricco>
nkaretnikov: Yea. reflex-platform is going to pretty much always be in a known-working state. GHCJS in nixpkgs gets much less attention (I don't get notifications for any failures, so I have to anticipate any possible failures or hope to be cc'd on any issues)
fragamus has joined #nixos
<nkaretnikov>
elvishjerricco: why is the platform in a separate repo, tho? was there some issue with the upstream?
asuryawanshi has joined #nixos
<nkaretnikov>
elvishjerricco: or is it just a way to have some stable environment?
<elvishjerricco>
nkaretnikov: reflex-platform is a collection of things that we know are needed for reflex development (such as a checkout of nixpkgs with a working ghcjs). The hope for pretty much everything in there is to be upstreamed eventually, but there's always things that aren't there yet. Plus a number of decisions have to be made about versions of stuff like reflex, so it takes care of pinning those for you
<nkaretnikov>
elvishjerricco: the other question i have is can i still use cabal build when inside the reflex shell? will it automagically call ghcjs? the readme page suggests to just use ghcjs --make
<elvishjerricco>
nkaretnikov: Ugh. The README in reflex-platform is annoying because people always see that and assume that's how you're supposed to use reflex-platform. I'd rather it just never said that, because later in the README there's a link to "project-development.md", which is the right way to do a realworld project. Yes, it uses cabal in the nix-shell
* nkaretnikov
checks
<nkaretnikov>
elvishjerricco: okay, this is awesome
<elvishjerricco>
nkaretnikov: Yea the `project` stuff is another example of something we need in reflex-platform, but it's so nice that I really want to upstream it somewhere
<elvishjerricco>
`project` really simplifies haskell dev with Nix
atondwal has quit [Quit: atondwal]
<nkaretnikov>
elvishjerricco: on the development page, there are some references for building mobile apps. does it work by bundling webkit or something?
<{^_^}>
→ 681826c0 by R. RyanTM: calibre: 3.23.0 -> 3.24.1
<{^_^}>
→ 3c98015e by @grahamc: Merge pull request #41122 from r-ryantm/auto-update/calibre
<elvishjerricco>
nkaretnikov: It actually cross compiles your Haskell to a native executable for the mobile device that links to the device's native webview library for rendering. The Haskell code uses the `jsaddle`library to shim JS calls over a bridge to the webview
<elvishjerricco>
Turns out that bridge is way faster than GHCJS, because GHCJS is slow and native Haskell is very fast :P
<gchristensen>
samueldr: ^
<elvishjerricco>
It's quite important for mobile Haskell
<nkaretnikov>
elvishjerricco: sweet
emmanuelrosa has joined #nixos
JasonGrossman has quit [Remote host closed the connection]
<samueldr>
gchristensen: this looks like the previous revision, right?
atondwal has joined #nixos
<gchristensen>
yea
<samueldr>
could you tell me the deployed commit ID?
<elvishjerricco>
nkaretnikov: I have to rebuild GHCJS from scratch to test (I let it get GC'd away :P), but once I've confirmed it's working I'll push my fixes (though I kinda recommend reflex-platform anyway :P)
<samueldr>
gchristensen: ah, this could explain weirdness :)
sanscoeur has joined #nixos
<nkaretnikov>
elvishjerricco: yeah, we'll see. thanks for your help!
<gchristensen>
samueldr: mind applying those patches and I can deploy again?
<samueldr>
hmm, what's < IRC_QUEUE = "amq.direct"
<samueldr>
for?
<samueldr>
(I first want to understand the changes :) )
sanscoeur has quit [Ping timeout: 264 seconds]
Fare has joined #nixos
<gchristensen>
heh no idea
<samueldr>
and i should use `channel.direct("", ...)` instead of `channel.IRC_EXCHANGE("", ...)` if I follow you, right?
<samueldr>
(not used to that diff format)
<samueldr>
(and I'm not 100% sure which side is which!)
<JasonGrossman>
Long shot, but does anybody know an easy way to use stumpish on NixOS? It's meant to be installed as part of a stumpwm installation, but the current Nix derivation for stumpwm doesn't seem to install it.
stepcut has quit [Remote host closed the connection]
semilattice has joined #nixos
<semilattice>
anyone have experience with ruby gems?
xy2_ has quit [Ping timeout: 268 seconds]
ray_ has joined #nixos
<ray_>
Hello, I'm having a bug where in grub, if I choose "default configuration", the grub background image stays up and covers up the prompt to enter my password to decrypt my ssd. I'm still able to type the password, but it's annoying since I can't see the prompt. But the weird thing is that if I choose "all configurations" then pick any one, this weird behavior doesn't happen and I see the prompt as normal.
<ray_>
Another crucial detail is that I'm running libreboot, so I don't have grub installed to the ssd, instead grub.cfg is loaded from libreboot's grub
<mog>
i have clearlooks installed and it is working for gtk2 apps but i cant seem to get it working for my gtk3 one
<ray_>
I've tried changing the grub configuration so it doesn't have a splash screen, but then libreboot's background image just stays to cover the password prompt
endformationage has quit [Quit: WeeChat 1.9.1]
michas has quit [Ping timeout: 248 seconds]
sanscoeur has joined #nixos
sanscoeur has quit [Ping timeout: 264 seconds]
<JasonGrossman>
semilattice: Not me, and ruby generally doesn't seem to get mentioned much here. Sorry.
<semilattice>
@jasongrossman thanks regardless. I'm just tryiing to sneak an extra step into one of the dependency gems without negating auto generation
<JasonGrossman>
ray_: You've accidentally achieved an extra level of security! Harder for anyone else to type your password. :-)
furon has quit [Quit: Connection closed for inactivity]
ray_ has quit [Quit: leaving]
seafood has quit [Quit: seafood]
liori has quit [Remote host closed the connection]
newhoggy has joined #nixos
sanscoeur has joined #nixos
newhoggy has quit [Remote host closed the connection]
newhoggy has joined #nixos
sanscoeur has quit [Ping timeout: 264 seconds]
jasongro` has quit [Remote host closed the connection]
JasonGrossman has quit [Remote host closed the connection]
JasonGrossman has joined #nixos
newhoggy has quit [Remote host closed the connection]
JasonGrossman has quit [Remote host closed the connection]
JasonGrossman has joined #nixos
sbdchd has quit [Remote host closed the connection]
<cocreature>
Hey, if I run “nix-shell -p "llvm_6.overrideAttrs(oldArgs: {debugVersion = true;})" --command "llvm-config --build-mode"” (and wait until the build finishes) I still get “Release” instead of “Debug”. am I doing something wrong here?
Lisanna has quit [Remote host closed the connection]
<cocreature>
I also have the LLVMExports-release.cmake file rather than the LLVMExports-debug.cmake file in the store directory pointed to by running `which llvm-config` inside the shell
JasonGrossman has quit [Ping timeout: 245 seconds]
<cocreature>
it seems like the debug flag is not being applied but given that it rebuild llvm something™ seems to have been applied
Mateon3 has joined #nixos
JasonGrossman has joined #nixos
andymandias has quit [Quit: ZNC 1.6.5 - http://znc.in]
Mateon1 has quit [Ping timeout: 252 seconds]
Mateon3 is now known as Mateon1
<{^_^}>
[nixpkgs] @dtzWill opened pull request #41128 → Revert "sqlite{,-analyzer}: use the same src (#40945)" → https://git.io/vhqbr
sary has quit [Ping timeout: 260 seconds]
andymandias has joined #nixos
<dtz>
^ didn't take all day to chase down why Nix was being super crashing
<dtz>
cocreature: ithink debug flag is outdated, I think it's not tied to cmake build type anymore
<dtz>
lemme check
dvim has joined #nixos
<dtz>
okay well two "problems" -- I got a short bit into the build (using command you provided) and already I'm seeing CMAKE_BUILD_TYPE=Release -- it actually appears twice! :)
<dtz>
the other problem is more of a possible problem-- i'm not sure how neatly "override" works given the way we add the manpages output
<dtz>
oh you're using overrideAttrs
<dtz>
if you use ".override{debugVersion=true;}" that's invoking it right
<dtz>
not sure if anyone's tested/maintained that--bug me/file issue if not
<cocreature>
dtz: hm if I try nix-shell -p "llvm_6.override(oldArgs: {debugVersion = true;})", I’m still seeing “cmake flags: -DCMAKE_BUILD_TYPE=Release …” in the buld log
<dtz>
err
<cocreature>
ah it’s overwritten later in the same line
<dtz>
oh you probably had it right in your shell haha
<dtz>
yeah it still has the first one because cmake setup-hook sets that automatically
<cocreature>
alright, time to build llvm again :)
<dtz>
we should probably tell it we're doing a debug build, not sure. This might work O:).
<cocreature>
so I think get why I need override instead of overrideAttrs but I don’t get why using overrideAttrs causes a rebuild. shouldn’t that just be a useless parameter? do those influence the hash?
aszlig has quit [Quit: Kerneling down for reboot NOW!]
<dtz>
yes, changed attrs changes hash
<dtz>
they're also set as env variables in the build
<dtz>
".override" doesn't change the attrs but changes the values of the function in that file
<cocreature>
ah ok, so they’re not unused. that makes sense, thanks!
aszlig has joined #nixos
<dtz>
which is something like { stdenv, ...., debugVersion, ... } : stdenv.mkDerivation { <<go go gadget LLVM build as a function of (stdenv, ...., debugVersion, ....)>>};
<dtz>
lol
Fare has quit [Ping timeout: 260 seconds]
joehh has joined #nixos
chimay has joined #nixos
asuryawanshi has quit [Remote host closed the connection]
tzemanov_ has joined #nixos
asuryawanshi has joined #nixos
tzemanovic has quit [Ping timeout: 252 seconds]
asuryawanshi has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
sanscoeur has joined #nixos
grp has quit [Ping timeout: 245 seconds]
joehh has quit [Ping timeout: 260 seconds]
sanscoeur has quit [Ping timeout: 268 seconds]
asuryawanshi has quit [Ping timeout: 268 seconds]
dbe has joined #nixos
grp has joined #nixos
semilattice_ has joined #nixos
semilattice has quit [Ping timeout: 260 seconds]
dottedmag has joined #nixos
dottedmag has quit [Changing host]
dottedmag has joined #nixos
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<sphalerite>
tazjin: jD91mZM2_: The thing you suggested yesterday, with creating my channel by pointing it to a github archive, didn't work because the github tarball doesn't contain the symlink nixpkgs -> .
jasongro` has joined #nixos
jD91mZM2 has joined #nixos
newhoggy has joined #nixos
<sphalerite>
jD91mZM2: connection problems? :p
<jD91mZM2>
sphalerite: You saw! Randomly went out and had to restart networkmanager
<sphalerite>
idk if you saw my message to you just a minute ago then :p
<sphalerite>
tazjin: jD91mZM2_: The thing you suggested yesterday, with creating my channel by pointing it to a github archive, didn't work because the github tarball doesn't contain the symlink nixpkgs -> .
jD91mZM2_ has quit [Ping timeout: 264 seconds]
<jD91mZM2>
Oh yeah I didn't, thanks for resending :)
jasongro` has quit [Remote host closed the connection]
tmaekawa has quit [Quit: tmaekawa]
lostman has joined #nixos
<jD91mZM2>
That's actually a really weird symlink
<jD91mZM2>
`ls nixpkgs/nixpkgs/nixpkgs/`
<sphalerite>
yep
<jD91mZM2>
yay recursion
<jD91mZM2>
I wonder what happens if I ask find to follow symlinks
<sphalerite>
I think it has logic for detecting loops. Not 100% sure though.
<taktoa>
I wonder if symlink resolution is turing complete
<jD91mZM2>
find: File system loop detected; ‘./nixpkgs’ is part of the same file system loop as ‘.’.
<sphalerite>
taktoa: how would it be?
<jD91mZM2>
It's also missin programs.sqlite and svn-revision
<jD91mZM2>
missing*
tzemanovic has joined #nixos
sanscoeur has joined #nixos
<jD91mZM2>
That said, programs.sqlite is only used by command-not-found I think. So maybe you could recreate an old nixpkgs version by just downloading a tarball and adding the symlink
<ldlework>
if I installed a package by way of fetchTarball how do I get nixos-rebuild to refetch it?
JasonGrossman has quit [Remote host closed the connection]
JasonGrossman has joined #nixos
JasonGrossman has quit [Read error: Connection reset by peer]
<sphalerite>
jD91mZM2: yeah. Hell of a fuss though for just setting the channel to a specific version (that it was at one point at, no less!)
<jD91mZM2>
I mean programs where you need old versions you probably want to create a custom expression for either way
<jD91mZM2>
Like autoconf264
tmaekawa has joined #nixos
tmaekawa has quit [Client Quit]
sanscoeur has quit [Ping timeout: 276 seconds]
joehh has joined #nixos
nD5Xjz has quit [Ping timeout: 252 seconds]
<sphalerite>
I don't want old versions of anything per se, I just want the initial state of the channels on the servers that I deploy to be the same as on the machine I deployed it from :/
<jD91mZM2>
I don't see why exactly. Anyway, wouldn't `nix copy` do this?
<sphalerite>
what I'm trying to achieve is to have this look like a regular nixos installation as much as possible
<sphalerite>
that is, I want to be able to deploy, then ssh in, make some config change, and nixos-rebuild switch
<sphalerite>
without having it rebuild a bunch of stuff ebcause the channel state doesn't match
<sphalerite>
and also be able to do nixos-rebuild switch --upgrade and have that update stuff
<jD91mZM2>
Could you upload some channel files from the host computer?
<sphalerite>
hm
<taktoa>
sphalerite: naively I expect it wouldn't be but there might be some kind of trick, maybe using broken symlinks or something
<taktoa>
unless the kernel has cycle detection
<sphalerite>
it doesn't have cycle detection AFAIK, but it does have a hard limit on how many levels of links it will resolve before giving up
<taktoa>
I guess the question is how you would ever branch
gonz_ has joined #nixos
seafood has joined #nixos
<taktoa>
you'd have to maybe figure out some way to get the kernel to create a file during symlink resolution
<jD91mZM2>
There is programs.sqlite and everything
<jD91mZM2>
it's like christmas
<sphalerite>
lol
<sphalerite>
thing is, that's not necessarily the nixpkgs that the deployment was made from ;/
<sphalerite>
but that might
<sphalerite>
but that might be too much to ask for, as that's not guaranteed to be a channel version
<jD91mZM2>
No?
<sphalerite>
no. Because the user can do -I nixpkgs=/path/to/their/nixpkgs
<jD91mZM2>
Oh okay. Can't you just get the value of <nixpkgs>?
<sphalerite>
yes, but then I still have the same problem as with the github archive
<jD91mZM2>
> <nixpkgs>
<{^_^}>
/var/lib/nixbot/state/nixpkgs
<sphalerite>
I think what I'm trying to do is just not possible
<sphalerite>
at least in the very general case that I envision. If we have a guarantee that it's in a channel we're better off
<jD91mZM2>
For me <nixpkgs> would work. It returns /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs, which is the right link
<sphalerite>
yeah, but with -I nixpkgs=/path/to/checkout/of/nixpkgs it won't
<jD91mZM2>
Then it would return /path/to/checkout/of/nixpkgs wouldn't it?
<jD91mZM2>
And that's correct, no?
<sphalerite>
yes, but that's not something we can use as the channel state
<jD91mZM2>
Why not?
<jD91mZM2>
(Forgive me if I'm stupid)
<sphalerite>
because it doesn't contain the nixpkgs symlink
<jD91mZM2>
oh, right
<jD91mZM2>
Is it possible to create it?
<sphalerite>
also, why does htop default to showing threads…
justanotheruser has quit [Ping timeout: 264 seconds]
thblt has joined #nixos
<sphalerite>
if you make a derivation based on what the user has, but ugh. I'll just keep it simple and not bother
thblt has quit [Changing host]
thblt has joined #nixos
coot has joined #nixos
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
semilattice_ has quit [Ping timeout: 240 seconds]
<tilpner>
jD91mZM2 - sphalerites "very general case" might even include the case where <nixpkgs> is a file, so I agree that it's not easily possible to support all setups
<sphalerite>
jD91mZM2: no, I definitely wasn't plannign to go that far, just for valid nixpkgss
<sphalerite>
err tilpner ^
woodson has quit [Ping timeout: 240 seconds]
<tilpner>
Hey, my <nixpkgs> is definitely valid, just uncommon!
<tilpner>
I partly solved [the problem I think you're having] by automatically symlinking source nixpkgs into NIX_PATH on activation
<tilpner>
But that doesn't transfer overlays, or nixpkgs config, nor does it allow updating the nixpkgs version from the server (this last part I'm fine with)
<sphalerite>
well overlays and nixpkgs config aren't a problem since they're part of the nixos config which I'm copying over
<tilpner>
Are you copying it over automatically?
<Cypi>
So, I'm still unable to run teamspeak (cf https://github.com/NixOS/nixpkgs/issues/33274 and https://github.com/NixOS/nixpkgs/issues/36793 for references to the same problem). I have tried to basically upgrade/reinstall basically everything in my systemwide configuration and my user nix-env (at least I think so), and still getting the exact same error. What can I try?
<tilpner>
I'm managing all that via git now, but I don't always remember to pull before doing local changes
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
<tilpner>
Cypi - Do you want to fix the issue upstream, or do you just want to run teamspeak by yourself?
<tilpner>
One comment says nixos-18.03 (still?) has a working teamspeak_client
xy2_ has joined #nixos
<Cypi>
I would be happy with the latter, as I am very much a novice
<Cypi>
Hmm, maybe it's just a matter of upgrading my system actually, it seems my current channel is 17.09
<lostman>
Hi folks. I've been trying to find out why libclang pulls in gcc in addition to gcc-lib as dependency (when building docker image). Running strings on libclang.so.6.0 shows /nix/store/cm5znbhvylrh702hhsy9hnad7wk3v7xv-gcc-7.3.0 which as far as I understand has something to do with how nix computes runtime dependencies. Is there anything I can do about it?
newhoggy has quit [Remote host closed the connection]
<tilpner>
nix why-depends --all nixpkgs.llvmPackages_6.libclang nixpkgs.gcc-unwrapped confirms your issue :/
<lostman>
tilpner: is there some way to override runtime dependencies?
sanscoeur has quit [Ping timeout: 252 seconds]
<tilpner>
Sort of. You can set disallowedReferences = [ gcc ]; to make Nix complain about gcc still being referenced, and you can use remove-references to break those references
<tilpner>
No idea what will happen after you broke them though. If libclang actually used that path for anything, it will no longer find anything there
<Huw>
I've reached the point where installation says it has succeeded, I set the root password and reboot. And stage 1 works, but stage 2 can't find the root pool
nD5Xjz has joined #nixos
<Huw>
(Actually I think stage one says it can't find it which causes stage 2 to panic immediately).
<tilpner>
You could wait around a while (and repeat your question every 2 hours), there are some people who use ZFS in here
<Huw>
I'm in Australia, but yeah, I can try for a few hours more.
<jD91mZM2>
(Pinging adisbladis[m])
hlolli_ has joined #nixos
<Cypi>
I'm not sure upgrading my system is working. I get a few "warning: unknown setting 'signed-binary-caches'", then it seems to hang indefinitely after "building the system configuration...". (maybe it's just very long, but it doesn't seem to be doing anything)
thblt has quit [Remote host closed the connection]
<tilpner>
Cypi - That warning is expected and Fine™. It'll go away soon
taktoa has quit [Remote host closed the connection]
<Cypi>
Oh hey, I just noticed I don't have any program anymore in my nix-env, how did that happen...
<Cypi>
I'm such a noob.
hlolli_ has quit [Ping timeout: 276 seconds]
simukis has joined #nixos
<jD91mZM2>
Cypi: What do you mean?
<Cypi>
I mean that I had stuff installed with `nix-env -i`, and now I have nothing when I do `nix-env -q`. I was meddling to try and get teamspeak working, and somehow I must have...uninstalled everything?
<Cypi>
I ran `nix-collect-garbage -d` earlier, thinking (probably wrongly) it might cleanup any old version of Qt I had laying around. That command cannot uninstall anything, right?
<{^_^}>
[nixpkgs] @peti pushed 0 commits to revert-41123-auto-update/bind: https://git.io/vhqp6
<tilpner>
samueldr - Should ^ be posted here? Or only master/staging?
<lostman>
tilpner: thanks for the pointers. I think the issue is that clang derivation sets GCC_INSTALL_PREFIX. I wonder if anything there is actually required for libclang though (I'm not using whole clang, just libclang)
<adisbladis[m]>
jD91mZM2: Pong
<sphalerite>
Cypi: it won't remove anything you have installed in a current version of any profile
<Cypi>
Ok. Then I have literally no idea how I managed to do that :/
<tilpner>
adisbladis[m] - Huw was having problems with ZFS
<Cypi>
(like I said, I wouldn't know how to do it even if I meant to)
<sphalerite>
Cypi: it deletes old generations of all profiles, removing the ability to roll back, then deletes anything that nothing depends on anymore. This might mean removing an old version of qt for instance.
<tilpner>
(Only deletes from *all* profiles when root)
Lisanna has joined #nixos
* tilpner
is being nitpicky today
<sphalerite>
all profiles it has permission to modify :)
* sphalerite
one-ups nitpickyness
<tilpner>
:)
<yurb>
What's the licensing policy for packages in nixpkgs? I see Adobe Reader given as an example. Does Nixpkgs allow closed-source software?
* jD91mZM2
always two-ups
<Cypi>
I guess I'm done for then. It looks like the numerous ".default-profile-*-link" symlinks in my home are all dead
<Cypi>
(why aren't they garbage-collected? :o )
<tilpner>
yurb - Yes, there's a bunch of nonfree software in nixpkgs
JasonGrossman has joined #nixos
<MichaelRaskin>
jasongrossman: you can just checkout the upstream stumpwm-contrib repository
<tilpner>
yurb - But users need to opt-in to unfree software in general (or per-package)
<sphalerite>
Cypi: that seems like an odd situation, usually the numbered links should only be in /nix/var/nix/profiles/
<iqubic>
Is it possible to run the garbage collect command, but keep the last N generations?
<Huw>
Yep I'm back
<sphalerite>
Cypi: and there should be a link ~/.nix-profile pointing to /nix/var/nix/profiles/per-user/$username/profile
<MichaelRaskin>
iqubic: you can even manually remove exactly the generations you want to delete in /nix/var/nix/profiles
<yurb>
tilpner: thanks.
chimay has quit [Ping timeout: 264 seconds]
<sphalerite>
iqubic: there isn't a single command for it, but you can use nix-env --delete-generations, or you can use --delete-older-than which allows you to select by age (rather than number of gens)
<Cypi>
Hmm, in my home, ~/.nix-profile is pointing to ~/.default-profile, which is pointing to ~/.default-profile-68-link, which is pointing to some /nix/store/blurb-user-environment/
<tilpner>
yurb - FWIW, adobe-reader is in nixpkgs already
<Huw>
adisbladis[m] I'm getting issues with the root pool not being found on boot
<Cypi>
I don't think I've done anything special
<yurb>
tilpner: yes, I've noticed it linked in the manual. That's why I was curious.
<adisbladis[m]>
Ahh, then a simple `zfs import rpool` should suffice
<adisbladis[m]>
zpool import I mean
<Huw>
That says "no pools available to import"
<Huw>
Is this conversation better on the public channel?
joehh has quit [Ping timeout: 245 seconds]
<Huw>
I can now mount the root
toby1851 has joined #nixos
<Huw>
So should I reboot?
Have-Quick has joined #nixos
b has quit [Quit: Lost terminal]
romildo has joined #nixos
<romildo>
I am running NixOS from a VM built with nixos-rebuild build-vm and I want to share the clipboard with the host and the virtual machine. How can it be done?
thblt has joined #nixos
JasonGrossman has quit [Remote host closed the connection]
JasonGrossman has joined #nixos
b has joined #nixos
simukis has quit [Quit: simukis]
newhoggy has joined #nixos
simukis has joined #nixos
joehh has joined #nixos
freusque has joined #nixos
ixxie has quit [Ping timeout: 248 seconds]
seafood has quit [Quit: seafood]
seafood has joined #nixos
aarvar has quit [Ping timeout: 256 seconds]
sanscoeur has joined #nixos
newhoggy has quit [Remote host closed the connection]
seafood has quit [Ping timeout: 268 seconds]
Lisanna has quit [Remote host closed the connection]
sanscoeur has quit [Ping timeout: 264 seconds]
coot has quit [Quit: coot]
<adisbladis[m]>
Huw: So what was the issue?
<{^_^}>
[nixpkgs] @yegortimoshenko merged pull request #41128 → Revert "sqlite{,-analyzer}: use the same src (#40945)" → https://git.io/vhqbr
<{^_^}>
[nixpkgs] @yegortimoshenko pushed 2 commits to staging: https://git.io/vhqhA
<Huw>
From the live usb I can mount import the pool. But the system doesn't find it at stage on.
<adisbladis[m]>
Huw: Does your `boot.supportedFilesystems` contain zfs?
<Huw>
^ import and mount
<Huw>
Yeah
<adisbladis[m]>
Huw: And your hardware-configuration.nix contains what?
jtojnar has quit [Read error: Connection reset by peer]
jtojnar has joined #nixos
thc202 has joined #nixos
tmaekawa has quit [Quit: tmaekawa]
<Huw>
Sorry, I'll have to transcribe it. Which part are you interested in?
<sphalerite>
actualy you know what's probably easiest? Try booting the installed system with debug1devices, and run zpool list and friends there to see if it can find the pool
<sphalerite>
and the filesystems
Have-Quick has quit [Quit: Have-Quick]
robstr has quit [Quit: WeeChat 1.9.1]
<sphalerite>
I don't know if maybe rpool having mountpoint / might confuse it, but you said it's not finding the pool at all?
robstr has joined #nixos
<sphalerite>
If that's the case then the mountpoints shouldn't be the problem AFIAK
<Huw>
My boot loader doesn't really have any options apart from nixos, efi default loader, and reboot into firmware interface.
__monty__ has joined #nixos
<sphalerite>
if you're using systemd-boot or grub (which is probably the case), you can press e to edit a boot entry
jasongro` has joined #nixos
<sphalerite>
so select the entry, press e, put boot.debug1devices on the kernel command line, and then boot (enter in systemd-boot, F10 in grub)
<Huw>
Got it
JasonGrossman has quit [Ping timeout: 256 seconds]
<Huw>
Ok
<Huw>
zpool import rpool says it was used by a previous system
<nikivi>
What do I do in case where a Python script requires some module to run?
<Huw>
But that I can force it
<nikivi>
I get this error: ImportError: No module named docopt
<nikivi>
So I tried to install `python27Packages.docopt` from nix packages but still can't run the script
<nikivi>
I read that you should avoid using pip with Nix
<__monty__>
nikivi: How much of the manual have you read? Nix definitely needs more upfront effort than most package/dependency managers.
<nikivi>
I was reading the PhD theses on Nix now, and only the begging of the manual
<nikivi>
probably a bad idea to jump head first into Nix and delete my other packager manager
<symphorien>
nikivi: the build.nix file is just a way to have nix-env accept a "complicated command line" . Once you install the content of the file, you can remove it
<symphorien>
the content of the file is a python with docopt built-in
<symphorien>
so you can't have several
<symphorien>
But you can have one python with all the modules you want
<symphorien>
if some of your scripts have conflicting deps, then you must either use nix-shell or package your scripts properly as __monty__ said
<nikivi>
You mean have python and some module installed globally?
<symphorien>
yes the nix-env stuff is to have a global python batteries included
<symphorien>
isn't that what you want ?
<nikivi>
I want to run this script globally without errors
<Huw>
sphalerite I can force import, then ctrl D and get a boot.
<nikivi>
Assuming I use nix-shell how can I run it inside nix-shell?
<Huw>
But by default it's not booting, still saying can't find the pool
<symphorien>
just run exactly the commands of the manual in a regular shell, and then python3 -c "import numpy"
<symphorien>
you'll see
<nikivi>
python3 -c "import docopt"* you mean
<symphorien>
whatever you put in the build.nix
<nikivi>
also this is python 2 script but I think I get the ia
<nikivi>
idea*
cnu- is now known as cnubidu
<__monty__>
nikivi: I'd recommend reading the manual, both nix and nixpkgs before reading the theses (at least I think there's 2). You'll get why you can't just nix-env -i pythonlib if you do.
<nikivi>
Is there a hacky way I can use to run this script now and make it work?
Arcaelyx has quit [Ping timeout: 245 seconds]
<__monty__>
I'd also have recommended against removing your system's package manager tbh, I'm currently on mac too and for some things I just have to fall back to brew.
<symphorien>
nix-shell -p python lib1 lib2 lib3
<nikivi>
__monty__ Doesn't using other global package managers defeat the purpose of Nix?
<Huw>
I needed to use debug1devices to zfs list; then zpool import rpool; then I can boot the system
simukis has quit [Ping timeout: 248 seconds]
<nikivi>
symphorien error: undefined variable 'docopt' at (string):1:103
<symphorien>
python27Packages.docopt
<__monty__>
nikivi: Nope, that's the beauty of nix. Trust me you'll choose to avoid the others in no time but for some things it's just more convenient.
<nikivi>
I tried run: nix-shell -p python docopt
<Huw>
But that's not a great workflow for the family
<{^_^}>
[nixpkgs] @symphorien opened pull request #41133 → os-prober: fix dependencies for version >= 1.75 → https://git.io/vhmek
<__monty__>
nixy: You want pythonPackages.docopt
<__monty__>
nixy: Woops, meant nikivi.
<__monty__>
nikivi: Easiest solution would be adding these two lines to the top of your script:
<__monty__>
Then make the script executable, put it on your PATH and just call it.
<nikivi>
and remove `#!/usr/bin/env python2`?
<__monty__>
It's not necessarily what I'd recommend but it's the solution that'll take the least effort.
<symphorien>
yes
sanscoeur has joined #nixos
<nikivi>
okay wow, that worked, thanks __monty__
<nikivi>
will read the manual now, it's soo long :)
PLPD-Bot has quit [Remote host closed the connection]
Plato[m] has quit [Remote host closed the connection]
<nikivi>
I do want to understand how to deal with language specific package managers in a painless way though
<nikivi>
things like `npm install -g ..`
<__monty__>
nikivi: Once you've read both manuals you won't want to ; )
<nikivi>
So that I don't have to write nix derivations for all of that and repeat the efforts of people
smallville7123 has quit [Quit: rosa]
<__monty__>
nikivi: Your script sounds like an ideal candidate for your first nix package btw.
<nikivi>
So if I get this write, even small scripts that you write should be packaged under nix?
<nikivi>
right*
<__monty__>
You're not repeating efforts, you'll have to put in effort here and there though, that's definitely still necessary if you want to get the most out of nix.
<nikivi>
So if I write a small python script. Or only write nix deriviation for it if it tries to call some external libs?
PLPD-Bot has joined #nixos
<__monty__>
nikivi: You don't *have* to, you'll want to. Once you've written a couple you'll realize it's no big deal and it'll come in handy while developing.
<nikivi>
Also is Nix going to support zsh shell as alternative to bash? Without me having to prefix nix shell like so `nix-shell --run zsh` every time
<freusque>
/nix/store/b0zlxla7dmy1iwc3g459rjznx59797xy-binutils-2.28.1/bin/ld: /nix/store/1cpfm79phw1n8jrx3aayxgppfyp4766c-ghc-8.2.2/lib/ghc-8.2.2/rts/libHSrts.a(StgStartup.o): relocation R_X86_64_32S against symbol `stg_enter_info' can not be used when making a shared object; recompile with -fPIC
<tfc[m]>
symphorien: fresheyeball thx this works! i am currently building a lot of nix expressions to define the setup flow from a lot of scattered dependencies to some running services. and one thing i always stumble over is not finding the right documentation for such simple builtin/library things.... i always have the nixpkgs and nix documentation open and grep for "functionName = " in the nixpkgs repo, but often to no avail. any
<tfc[m]>
advise where to find more written help as a newbie who often looks for primitive helpers?
<freusque>
tfc[m]: I think the nix pills are a good starting point
<tfc[m]>
sorry i meant to answer to freusque and symphorien not fresheyeball
<tfc[m]>
freusque: hm i read those already. when i first read them they confronted me with a lot of things that i was not able to remember immediately when i needed them next. although they helped a lot already.
<tfc[m]>
freusque: using that, too. that's sometimes more sometimes less helpful than grepping through the nixpkgs repo. but it also does not find the definitions/docs of builtin functions, does it?
<freusque>
no it doesn't
<freusque>
unless the comments from the source are helpful
<tfc[m]>
hm too bad looks like something to just get through in the beginning.
<freusque>
if they exist, they usually are though
<tfc[m]>
to bad it's not a statically typed language. that would be so helpful. :D
ixxie has joined #nixos
<ixxie>
heya
jmiven has quit [Quit: co'o]
jmiven has joined #nixos
<gonz_>
Too bad that it's its own language at all, really
<gonz_>
If we're talking about nix
ericsagnes has quit [Read error: Connection reset by peer]
<tfc[m]>
gonz_: true. although it makes sense that it's purely functional.
<gonz_>
There were already alternatives for that as well
<tfc[m]>
however, i have a new problem right now. i wanted to play with builtin functions so i installed `nix-repl` with `nix-env -i nix-repl`. when i start `nix-repl` now i get an error `nix database directory /nix/var/nix/db is not writable: permission denied`
<gonz_>
`nix` as a language has no actual reason for existing
<tfc[m]>
what could i have made wrong possibly?
<ixxie>
well there is Guix if you want a Nix with a general purpose language (scheme)
<sphalerite>
tfc[m]: nix-repl is the old thing, the new fancy thing is `nix repl`
<sphalerite>
tfc[m]: which comes with nix
<ixxie>
but I think there is much to be said for the value of Nix being a DSL
<tfc[m]>
oh i see! so i better forget that old one.
<tfc[m]>
thank you! :)
<sphalerite>
tfc[m]: if you want to use nix-repl rather than nix repl for some reason, you need to set NIX_REMOTE=daemon in the environemtn for it
<tfc[m]>
i have no reason to use nix-repl. i did just not know that `nix repl` exists and that works for me now. thank you
<gonz_>
ixxie: It could've been a DSL in Haskell. There was already plenty of precedent set, it's obviously the better choice. It might work fine the way it is, but that doesn't mean it was the correct chocie.
<gonz_>
s/chocie/choice
<ixxie>
gonz_: that would allow a lot more freedom, which has advantage but also great disadvantages
<tfc[m]>
btw. does "gonz_" abbreviate gabriel gonzalez? :D
<gonz_>
tfc[m]: No
<tfc[m]>
ok. it seemed obvious sorry
<ixxie>
with that much freedom, Nixpkgs would be a lot more heterogenous
<MichaelRaskin>
A DSL in Haskell is not an obviously better choice, because bootstrapping
<MichaelRaskin>
Apparently even the Perl bits were too much pain
<ixxie>
bootstrapping?
<tfc[m]>
hm hard to say. i would certainly like it, but i know of sooo many people who would damn much profit from nixOS's selling points and values - but they feel so alienated by the expression language. haskell would also not fix it.
<ixxie>
tfc[m]: also, better documentation (need to get back to working on the Wiki!)
<MichaelRaskin>
ixxie: building everything just right for a platform where nobody has built Nix before is sometimes annoying even as it is.
aristid has joined #nixos
<tfc[m]>
ixxie: ok that's right. more documentation always helps more. sometimes there _is_ already enough docs already but as a newbie i feel overwhelmed where to look
tzemanovic has joined #nixos
<tfc[m]>
btw. when i have a mkOption that is of `types.path` and i want to check if the user provided this at all - how to do that? currently i am at `0 == builtins.stringLength (builtins.toString cfg.foobarPath)`
<tfc[m]>
in order to use it in `mkIf`
<ixxie>
tfc[m]: I am trying to think about how to structure the Wiki in order to improve the newbie experience
<ixxie>
tfc[m]: but we lost momentum in the winter
tzemanovic has quit [Ping timeout: 252 seconds]
<ixxie>
now I am trying to think about how to proceed
<tfc[m]>
ixxie: i'd love to participate but currently i must concentrate on learning this myself
<ixxie>
of course ^^
<ixxie>
I think I feel the same and I have been around for two years
<ixxie>
its probably a common blocker for documentation efforts
<ixxie>
everybody prioritises coding xD can't blame them! I feel the same
<tfc[m]>
i think i measure myself being ready to help when i wrote my first >10lines nix expression without searching 20 minutes of documentation for every single bit of it. :D
<__monty__>
MichaelRaskin: Doesn't GHC with LLVM backend make it equivalent to C++ for bootstrapping (at least in theory)?
<tfc[m]>
then i would try to take complex existing nix expressions, and tear them apart to comprehensively explain it to readers by example. this is something i learn best from myself.
iqubic` has joined #nixos
<ixxie>
MichaelRaskin: how does having our own language help with bootstrapping?
ericsagnes has joined #nixos
<MichaelRaskin>
ixxie: once you have C++ — and you likely do have it — building Nix is actually the easy part
<ixxie>
tfc[m]: we wanna have tutorials in the wiki but didn't get around to it
Huw has joined #nixos
sanscoeur has joined #nixos
coot has joined #nixos
iqubic has quit [Ping timeout: 260 seconds]
smallville7123 has quit [Quit: rosa]
sanscoeur has quit [Ping timeout: 264 seconds]
joehh has quit [Ping timeout: 252 seconds]
andromeda-galaxy has quit [Ping timeout: 240 seconds]
Ariakenom has quit [Read error: Connection reset by peer]
andromeda-galaxy has joined #nixos
tzemanovic has joined #nixos
rauno has joined #nixos
<ixxie>
anybody have any clue as to why the package count as indicated by Repology would have doubled in the last month?
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Arcaelyx has joined #nixos
troydm has joined #nixos
troydm has quit [Ping timeout: 264 seconds]
sanscoeur has joined #nixos
<tfc[m]>
what is the correct way to merge 2 configs in a module that can be independent? like `config = (mkIf a { ...}) // (mkIf b {...});`
<Orbstheorem>
Hi, I was wondering if there's any reason why there's no source archive like debian which host an a copy of the sources of every packages
<tfc[m]>
but that does not work for me
<symphorien>
actually, hydra does have a copy of the source of all packages it has ever built
<Orbstheorem>
tfc[m]: you could put both in a set with actual names and select the right one wherever you use it
<tfc[m]>
Orbstheorem: but this is not an XOR choice but an inclusive or
<tfc[m]>
i have 2 very related services and user should be able to decide to run only A, only B, or both.
<symphorien>
Orbstheorem: for example https://nixos.org/nixos/packages.html#pdf+shuff the source of this version of psf-shuffler is nowhere to be found on the internet, but it is on hydra.
<tfc[m]>
so my `config = ...` expression must be able to also enable both at the same time
<LnL>
there's also refs/pull/${num}/head if there's some context where you don't know the source branch
patrl has joined #nixos
<LnL>
I should really play with declarative jobsets sometime...
johnw has quit [Ping timeout: 260 seconds]
camsbury has quit [Quit: WeeChat 2.0]
taktoa has quit [Ping timeout: 276 seconds]
<mjrosenb>
ok, so I'm running into the issue where nothing is working quite as expected on OSX. I'm not 100% clear on the cause, but it may be related to the unstable channel only being turned on for root account. Is there any harm in also adding that to my account?
<ryantrinkle>
LnL: hmm, i'm having a bit of an issue with it
<sonne>
i'd like to manage vimrc with nixos configuration, for what i can see online i should use home-manager. however it seems that home-manager doesn't only configure stuff but also installs user-contained versions of the packages. isn't there a middle ground?
<adisbladis[m]>
sonne: You can manage just files with home-manager
<MichaelRaskin>
Well, how do you want to configure vimrc contentx?
Lisanna has joined #nixos
<sonne>
adisbladis[m]: how so? i tried "program.vim { enable = true; };" and it installed the package, same for program.git {}
<MichaelRaskin>
You can trivially create a package that just copies a file into store, then add it to systemPackages (make sure pathsToLink include its location), then symlink or source it
<sonne>
MichaelRaskin: by putting my vimrc contents in some config variable i'd suppose?
<sonne>
MichaelRaskin: that sounds hackish
<sonne>
(also i just installed nixos 10 minutes ago for the first time)
sanscoeur has joined #nixos
woodson has joined #nixos
<aristid>
omg, somebody managed to package gnucash 3.1
<aristid>
amazing
* aristid
jumped from 2.4 to 3.1 directly for that reason
<aristid>
(2.6 was packaged technically, but very broken)
sanscoeur has quit [Ping timeout: 264 seconds]
<adisbladis[m]>
sonne: home-manager has a nice abstraction for that: `home.file.".vimrc".source = ./path/to/vimrc;`
FRidh has quit [Read error: Connection reset by peer]
blankhart has joined #nixos
<Orbstheorem>
Hello, what's the best way to debug this?:
<Orbstheorem>
May 27 19:59:31 Triglav kernel: xmonad-x86_64-l[2265]: segfault at 0 ip 00007fb241defcc4 sp 00007ffe07215298 error 4 in libX11.so.6.3.0[7fb241dbd000+139000]
q3k has joined #nixos
woodson has quit [Ping timeout: 240 seconds]
jtojnar has joined #nixos
jtojnar has quit [Remote host closed the connection]
<Orbstheorem>
Make a derivarion that produces libX11 with debugging symbols and recompile all my system?
iqubic has joined #nixos
alexteves_ has joined #nixos
<clever>
Orbstheorem: you can arrange it to only affect xmonad
woodson has joined #nixos
<Orbstheorem>
Also, Is there a decent way to rebuild my configuration with “options”? ie. I'd like to create multiple profiles and do something like nixos-rebuild switch -E 'profile=powersave'. Instead of overriding the environment variable
<clever>
not really
<Orbstheorem>
:(
* Orbstheorem
thinks there should be an option for that
blankhart has quit [Ping timeout: 276 seconds]
<ldlework>
Hmm I feel like the ZNC nixpkg options are not ideal
<ldlework>
they only allow you to express a single irc user
<ldlework>
Like you can only express one user, and the networks they have - but not multiple users.
alexteves has quit [Ping timeout: 240 seconds]
<ldlework>
If your nickname is different on oftc, oh well.
<__monty__>
That's an annoying limitation...
<clever>
Orbstheorem: ok, so the haskell xmonad package depends on the haskell X11 package, which depends on xorg.libX11
<{^_^}>
[nixpkgs] @bobvanderlinden opened pull request #41147 → Use Nix-builtin fetchurl where possible → https://git.io/vhmZ7
<clever>
haskellPackages.X11.override { libX11 = enableDebugging xorg.libX11; } would create a custom X11 library using a debug build of libX11
<Myrl-saki>
cannot find any of ["libHSprimitive-0.6.2.0-4578caNkWQ54Gt1mxLF2Yh.a","libHSprimitive-0.6.2.0-4578caNkWQ54Gt1mxLF2Yh.p_a","libHSprimitive-0.6.2.0-4578caNkWQ54Gt1mxLF2Yh-ghc8.0.2.so","libHSprimitive-0.6.2.0-4578caNkWQ54Gt1mxLF2Yh-ghc8.0.2.dylib","HSprimitive-0.6.2.0-4578
<Myrl-saki>
caNkWQ54Gt1mxLF2Yh-ghc8.0.2.dll"] on library path
<symphorien>
is there short way to write if x != null then x else default ?
<iqubic>
Is it just me or did videos stop working in the latest version of FireFox?
<ryantrinkle>
clever: what output should i expect?
<ryantrinkle>
i was seeing some errors in there before, when i had some permissions wrong
<steveeJ>
can someone recommend a screencast utility that's packaged in nixpgks?
<clever>
ryantrinkle: you can also manually run `hydra-notify build-finished 63657` or swap in build-started (as `sudo -u hydra -i`) to manually claim something has happened and update github
<clever>
ryantrinkle: grab an actual build# from the hydra UI
<manveru>
steveeJ: i used recordmydesktop
<ryantrinkle>
clever: do i run that as a particular user?
<ryantrinkle>
oh sorry, i see that nwo
contrapumpkin has quit [Quit: My MacBook Pro has gone to sleep. ZZZzzz…]
<michas_>
Hi, what is the difference between `nix-build` and `nix build`. The output of `nix` refers to the man page, but it looks like there is none. :/
patrl has joined #nixos
<steveeJ>
manveru: I'll give that a try, thanks!
<ryantrinkle>
clever: that ostensibly finished successfully (0 exit code)
<MichaelRaskin>
I remember once using xvidcap
<ryantrinkle>
maybe my regex is off?
<clever>
ryantrinkle: that runs thru the same regex
<ryantrinkle>
right; it didn't appear to do anything
<MichaelRaskin>
steveeJ: that was long ago, though
<ryantrinkle>
and then exited successfully
<clever>
ryantrinkle: ah, can you paste your regex and your job name?
<manveru>
given something like `{x = drv; y = drv;}`, is there an easy way to make a derivation that combines the outputs into symlinks in a directory with the key as directory name?
iqubic has quit [Remote host closed the connection]
<ryantrinkle>
do i not even have this plugin loaded at all?
<clever>
ryantrinkle: its always loaded
<ryantrinkle>
ah
sanscoeur has joined #nixos
<clever>
ryantrinkle: copy .hydra-notify-wrapped to a tmpdir and also copy the path like `/nix/store/8dxcx2jv168c3f6a31fag1a3qwkps170-hydra-2017-11-21/libexec/hydra/lib` to a tmpdir (read the hydra-notify script), and copy hydra-notify as well
<clever>
ryantrinkle: then modify hydra-notify to use the copy of -wrapped, and the copy of lib
<clever>
then start adding prints to it
iqubic has joined #nixos
vaninwagen has joined #nixos
<clever>
libexec/hydra/lib/Hydra/Plugin/GithubStatus.pm is where the PR status is handled
jensens has quit [Ping timeout: 245 seconds]
blob has quit [Quit: Leaving]
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
Orbstheorem: that uses the haskell overrides to replace xmonad and X11, then gets a .ghcWithPackages from the modified set
<ixxie>
manveru: by raw count yes, but not if you look at the freshness too
<ixxie>
which is important!
<manveru>
:)
<ixxie>
and repology considers it important enough to be the primary statistics
<manveru>
true, still feels like cheating
<manveru>
but i guess packages are packages, not applications
<ixxie>
well how many package managers support both applications and language specific packages? its quite rare and very cool ^^
<manveru>
;)
<MichaelRaskin>
Approximately all of them?
InstantGratifica has joined #nixos
<ixxie>
anyways congratulations to all the contributers
<manveru>
i just improved my crystal2nix wrapper btw
<MichaelRaskin>
Which repositories is a better question
<ixxie>
right
<ixxie>
s/support/have a rich count of/
<bhipple[m]>
I'd like to make a stdenv/glibc overlay that ads `sssd` as a dependency, so that user lookups with libnss_sss.so work as expected. I've tried the naive thing of trying to just override propagatedBuildInputs but am having some issues; is there a standard way to do this?
<ixxie>
but yeah of course it doesn't say how good the coverage is in what area
asuryawanshi has quit [Remote host closed the connection]
<ixxie>
I hope at some point this summer to get around to doing some nixpkgs data science crawl, hooking into repology and doing some topic modeling to figure out what categories of packages are well covered / badly covered in nixpkgs
<ryantrinkle>
clever: ah! i changed it so that hydra was running on refs/pull/$num/merge
<iqubic>
How does one use NixOS to change the default browser? I specifically want the default to be qutebrowser.
<manveru>
ixxie: do they have data dumps?
Mic92 has quit [Quit: WeeChat 2.1]
<manveru>
iqubic: set the BROWSER env variable
<ryantrinkle>
but apparently statuses are supposed to be reported against the head of the PR
<clever>
ryantrinkle: yeah
<ixxie>
manveru: better - an API
<clever>
ryantrinkle: the status is reported on a specific revision
<ryantrinkle>
clever: yeah, that makes sense, but why do i care what the status of the head is?
<clever>
ryantrinkle: and that special fake-merge commit doesnt show up in the PR page
<manveru>
ixxie: i'd rather have a pg db dump like rubygems does :)
<Orbstheorem>
clever: Thanks, that seems to have worked!
<ryantrinkle>
i care what will happen if someone mashes the "merge" button
<ryantrinkle>
isn't that what pull/$num/merge represents?
<ixxie>
manveru: well you can make a simple script to get a dump; its not that complex really
<clever>
ryantrinkle: yeah, but hydra doesnt know to post the status back to the head commit, and github cant show the status of the merge commit
<ryantrinkle>
clever: i see
<ryantrinkle>
bizarre
<clever>
ryantrinkle: hydra would have to be improved to know about head vs merge internally, and test merge but post to head
<manveru>
damn, does ofborg still not support fetchGit?
<ixxie>
anyway I am hoping to be able to use some NLP/ML magic to classify our packages; would that be of use / interest to anybody?
<__monty__>
manveru: Missed opportunity to call it onix ; )
<Orbstheorem>
So, why am I overriding haskellpackages and then getting ghcWithPackages instead of overriding ghcwithpackages directly? Is it because gwp is a function that resolves a package set?
<LnL>
manveru: why would it support that?
* manveru
silently weeps
<clever>
Orbstheorem: i think think ghcWithPackages accepts .override at all
<clever>
Orbstheorem: so you have to override the whole packageset it came from
<Orbstheorem>
Alright ^^
<Orbstheorem>
Thanks!
<clever>
yep
<manveru>
__monty__: what?
<__monty__>
manveru: Crystal and nix.
<manveru>
__monty__: well, it's not set in stone yet ;)
<__monty__>
You got me interested in crystal just because of the pun potential now.
ilyaigpetrov has quit [Quit: Connection closed for inactivity]
<clever>
Orbstheorem: also, now that we have switched to using a proper haskell override, you can just override X11, and it will affect xmonad automatically
sbdchd has joined #nixos
<iqubic>
Is there a way to find the location of an executable I have installed?
<clever>
iqubic: run type or which on it, then realpath on that path
<iqubic>
I installed it via system.environmentPackages.
<iqubic>
Why do I have to run realpath?
<clever>
iqubic: type/which give a path like /run/current-system/sw/bin/foo
<clever>
which is a symlink to the real binary
<clever>
if you need realpath or not, depends on what you want to do
<iqubic>
I see.
sbdchd has quit [Ping timeout: 268 seconds]
simukis has quit [Read error: Connection reset by peer]
<ryantrinkle>
clever: hmm, now i changed it to pull/${num}/head, but the job won't reevaluate
<clever>
ryantrinkle: youll need to make a new commit on the PR that actually causes a nix level rebuild
<nkaretnikov>
ryantrinkle, elvishjerricco: see above
mahalel_ has joined #nixos
<nkaretnikov>
that's when i tried running try-reflex with your binary cache
<ryantrinkle>
nkaretnikov: looking
<nkaretnikov>
could it be because i tried without the cache beforehand and aborted that?
<ryantrinkle>
clever: ah, what about if someone's just changing random stuff?
<elvishjerricco>
nkaretnikov: Ah, that's an annoying one. Your store is corrupted because you built some of the Haskell derivations and downloaded others
Mic92 has joined #nixos
<ryantrinkle>
nkaretnikov: yes, it could be
<elvishjerricco>
Haskell derivations are unfortunately nondeterministic
woodson has quit [Ping timeout: 252 seconds]
ariutta has joined #nixos
<{^_^}>
[nixpkgs] @7c6f434c pushed to master « libftdi1: fix build when docs are enabled »: https://git.io/vhmW8
qewzee has joined #nixos
<nkaretnikov>
are you sure the cache works? if i run collect garbage, it'll take days to catch up
<elvishjerricco>
So if you download a Haskell derivation, the ones it depends on must also have been downloaded from the same place
<elvishjerricco>
nkaretnikov: Yes, the cache is fine
<nkaretnikov>
okay, let me collect some garbage then
Adluc has joined #nixos
<ryantrinkle>
clever: by which i mean, say someone changes README.md only; will their PR not get marked as OK?
<clever>
ryantrinkle: depends on if you have a src = ./. that reads that file
<ryantrinkle>
clever: ah, should i just add that?
<clever>
also, the reval caching is within a jobset
<ryantrinkle>
that seems silly, but hey, if it makes this work, it's fine by me :P
<clever>
so the PR will always get a first eval when opened
<ryantrinkle>
ah, ok
<{^_^}>
[nixpkgs] @7c6f434c pushed to master « pythonPackages.pathlib2: fix build by removing obsolete test removal »: https://git.io/vhmWg
alexteves has joined #nixos
Lisanna has quit [Ping timeout: 252 seconds]
<nh2[m]>
aszlig: on your Hetzners what are the network interfaces called? I'm having a problem right now with on a new server eth0 being named eno1 instead. But since `usePredictableInterfaceNames` is true by default, I'm wondering why on all my other Hetzners the interface is called eth0
<nkaretnikov>
ryantrinkle, elvishjerricco: do you know that the tutorial link on the reflex homepage points to the homepage itself? is it on purpose?
rihards has quit [Quit: rihards]
<ryantrinkle>
nkaretnikov: i believe those links are broken at the moment
<ryantrinkle>
gonna hopefully get that fixed today :)
<elvishjerricco>
ryantrinkle: You see my PR about the homepage in reflex-platforM?
<ryantrinkle>
elvishjerricco: yeah, the one that brings it all in-tree?
<ryantrinkle>
it's an interesting idea :)
<elvishjerricco>
ryantrinkle: Yea
<ryantrinkle>
i haven't thought through the consequences yet, haha
<ryantrinkle>
just trying to get hydra to behave
<ryantrinkle>
though i've just learned that github's status checks stuff is fundamentally broken
<elvishjerricco>
ryantrinkle: ?
sanscoeur has joined #nixos
<ryantrinkle>
it wants you to check the unmerged head of the PR, not the thing you'll get when you merge it
<aszlig>
nh2[m]: there is a udev rule for setting that interface persistent
<ryantrinkle>
so, e.g., pushing to master doesn't invalidate the status checks
<ryantrinkle>
even of things that are meant to be merged to master
patrl has quit [Ping timeout: 256 seconds]
woodson has joined #nixos
<elvishjerricco>
ryantrinkle: Yea. But pretty much all CIs check the actual branch, not the merged version, right? Hydra does I believe.
<aszlig>
nh2[m]: see _get_udev_rule_for() in HetznerState
T_S_ has joined #nixos
<ryantrinkle>
elvishjerricco: i mean, regardless of whether that's what they do, it seems blatantly incorrect
<ryantrinkle>
who cares whether the unmerged branch works? the thing we're trying to guarantee is that when someone mashes the "merge" button, the resulting thing works
<ryantrinkle>
also, i did read that travis tests the merged branch
<ryantrinkle>
but then reports it to the head on github
<ryantrinkle>
which seems like about the best you can do with the current API
<nh2[m]>
aszlig: you are right, I can see it. And it's in /etc/udev/rules.d/99-local.rules on my server. Yet it's called eno1. Let' me have some more look or try trigger it manually ...
<clever>
ryantrinkle: the metadata on an eval has to be improved to track both head and merge, and then the github status plugin has to use that metadata, then it might be as smart as travis
<ryantrinkle>
clever: yeah, that makes sense
<ryantrinkle>
clever: i think i'm not understanding how to add ./. to my results
<nh2[m]>
aszlig: yeah ok the mac address is wrong. I think something went wrong with this server, it also had a wrong IP in its physical spec, the one of a different server of mine. Maybe I swapped some IPs accidentally on creation, I'll recreate it
<clever>
ryantrinkle: it would also be nice if hydra could do both head and merge builds, to confirm both pass
<ryantrinkle>
clever: agreed
<nh2[m]>
aszlig: thanks for the quick pointer!
<clever>
ryantrinkle: the ./. is just a question of if changing readme.md causes a rebuild, and id usualy try to avoid that
Ariakenom has quit [Quit: Leaving]
<clever>
ryantrinkle: the contents of readme.md dont affect if things pass or fail
sanscoeur has quit [Ping timeout: 252 seconds]
<ryantrinkle>
clever: hmm, ok; maybe i'm ok without that, for future builds
<aszlig>
nh2[m]: although it would be nice if there would be a way to regenerate the physical spec, for example when you got a hardware replacement without disks and thus the MAC has changed
<{^_^}>
[nixpkgs] @baracoder opened pull request #41153 → spring: Fixed compiling and version bump → https://git.io/vhmWN
pyrtsa has joined #nixos
<nh2[m]>
aszlig: yeah right now I fixed the IP with `sqlitebrowser`, but forgot to update the mac
<nh2[m]>
aszlig: I think the most important thing will be to write it in the manual so it's clear what is where
<Orbstheorem>
clever: the thing is that I'm trying to debug segfault I don't know how to trigger. It has happened multiple times over the last couple of weeks. I was planning to run on debuggable libs and when it crashes recover the coredump and then study it. But in the meanwhile, continue using my system
michas_ has quit [Read error: Connection reset by peer]
<clever>
Orbstheorem: in your override-xmonad\ \(1\).nix file, just strip out all of the overrides, and run `LD_LIBRARY_PATH=${pkgs.enableDebugging pkgs.xorg.libX11}/lib ${xmonad/bin/xmonad &`
<clever>
Orbstheorem: then the main xmonad will always use that debug build
<clever>
Orbstheorem: then you need one other small change
<judson>
Tangential but: how would you set up dnscrypt-proxy to work within a corporate VPN? The challenge is that there are names that only exist in the VPN DNS servers.
<judson>
Bonus points if internal names don't leak to the dnscrypt servers.
<clever>
Orbstheorem: running ulimit before you start xmonad will handle what this is descripting
<MichaelRaskin>
Hm. I cannot make Nix trust an extra public key from command line… How do I do this (as root)? nix-serve seems to sign the cache so I hope I did succeed in generating keys
<Orbstheorem>
clever: I don't think that worked. I switched, restart display-manager and gdb attach to the pid of the running instance and no lib has debugging symbols
<Orbstheorem>
Anyway, thank you very much
<Orbstheorem>
I have an exam tomorrow morning and it's almost 1am, I should be sleeping :P
<kandinski>
my nixos-rebuild fails and it looks this is the culprit: http://paste.debian.net/1026839/ any suggestion? I'm on 18.03
sbdchd has quit [Ping timeout: 256 seconds]
xy2_ has quit [Ping timeout: 240 seconds]
srdqty has quit [Quit: WeeChat 1.9.1]
sigmundv has joined #nixos
<kandinski>
kandinski: is there anything I can do to go around this issue?
alexteves__ has quit [Remote host closed the connection]
<MichaelRaskin>
ARGH
alexteves_ has joined #nixos
<MichaelRaskin>
Turns out it has cached in ~/.cache/nix that the binary cache didn't provide a signature, so the configuration changes were not seen even if I deleted the cache in /nix/var/nix/
alexteves_ has quit [Remote host closed the connection]