<manveru>
not sure how that's related to nix-shell?
<Mic92>
spacekitteh: what is an environment in your case, imports [] loads modules.
<spacekitteh>
as in, instead of keeping a bunch of copies of a shell.nix in various directories which require the same build environment, i can just do like nix-shell "my-env" or whatever
<Mic92>
spacekitteh: what is missing, from what I explained?
proteusguy has quit [(Ping timeout: 245 seconds)]
filterfish has quit [(Read error: Connection reset by peer)]
mizu_no_oto has joined #nixos
<spacekitteh>
i'm not sure
eacamero_ has joined #nixos
<Mic92>
spacekitteh: the basic idea is to create meta packages, which capture all dependencies. These can be then imported.
mizu_no_oto has quit [(Client Quit)]
eacameron has quit [(Ping timeout: 246 seconds)]
mizu_no_oto has joined #nixos
<spacekitteh>
hmm, i see
[0x4A6F] has quit [(Ping timeout: 260 seconds)]
<gchristensen>
ToxicFrog: ping now?
[0x4A6F] has joined #nixos
<spacekitteh>
Mic92: i don't suppose you could write a toy example configuration.nix + environments.nix ? i still haven't really grokked the nix language yet
ambro718 has quit [(Ping timeout: 245 seconds)]
thc202 has quit [(Ping timeout: 250 seconds)]
<ToxicFrog>
gchristensen: ack
<ToxicFrog>
gchristensen: that said, in general, if you have something to say to/ask me, just do it and I'll fish it out of the logs when next I can.
<avn>
spacekitteh: good example of "how to split configs" ;)
<spacekitteh>
avn: pretty
aneeshusa has joined #nixos
warbo has quit [(Ping timeout: 250 seconds)]
<spacekitteh>
avn: is ./localhost.nix special syntax? or do you manually change it on each box?
<avn>
spacekitteh: really I checkout that repo as /etc/nixos/nixos-configs/ and copy configuration.nix.in as /etc/nixos/configuration.nix with s/localhost.nix/realhostname.nix/
* spacekitteh
nods
<avn>
(altough I plan migration to ./hosts/$hostname/default.nix)
<avn>
or I not catched your question
<spacekitteh>
that's really awesome and i definitely plan on using your setup as an example but i don't think it's what i'm trying to achieve right now
<spacekitteh>
i'm trying to set up a bunch of different nix-shell development environments in my configuration.nix
<spacekitteh>
instead of having copies of them scattered everywhere
<avn>
spacekitteh: look on ./envs/*.nix
<avn>
but it use bit deprecated myEnvFun
<spacekitteh>
what should it use instead?
aneeshusa has quit [(Read error: Connection reset by peer)]
<avn>
spacekitteh: no idea, may be someone (or even I am) revamp myEnvFun to something more useful
<spacekitteh>
hmm
<avn>
or just copy definitions, when it finally deprecated and kicked off
aneeshusa has joined #nixos
<avn>
I like idea of build all my environments at once at system build time
chachi has joined #nixos
<spacekitteh>
atm i'm using pkgs.buildFHSUserEnv for an android build environment
<avn>
idk about android really
<spacekitteh>
yeah that doesn't really matter, but it does give me an "env"
<spacekitteh>
which can be directly nix-shell'd into
<chachi>
What is the command to restart my wireless connection? I am using xmonad, no kde4 or anything.
eacamero_ has quit [(Remote host closed the connection)]
<spacekitteh>
avn: so, looking at your envs/haskell.nix, how do you nix-shell into it?
<spacekitteh>
just nix-shell haskellEnv ?
siel has joined #nixos
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
<gchristensen>
or: that commit doesn't do what I think it does?
systemfault has quit [(Quit: Bye!)]
filterfish has joined #nixos
mkoenig has joined #nixos
siel has joined #nixos
<nathan7>
gchristensen: that appears to actually do the thing
aneeshusa has quit [(Quit: WeeChat 1.6)]
<gchristensen>
nathan7: ie: it should be running as 1000?
Rotaerk_ has joined #nixos
<nathan7>
gchristensen: I think so
siel has quit [(Ping timeout: 246 seconds)]
<nathan7>
the issue you're pointing at is May 2015, and was fixed long before 1.6 even was released
Rotaerk has quit [(Ping timeout: 268 seconds)]
<nathan7>
it seems to be failing with EINVAL though
<gchristensen>
yeah, I saw that :/ I was really hoping that patch would help
<nathan7>
which seems like something weirder is up that we should fix
siel has joined #nixos
<gchristensen>
that could be :)
<gchristensen>
the trouble is I don't even know where to start on that issue, it sounds like you have way more clue than I do!
<spacekitteh>
so i get this error: fatal: unable to access 'https://gerrit.googlesource.com/git-repo/': SSL certificate problem: unable to get local issuer certificate
<nathan7>
I happen to be familiar with the internals of that part of the Go stdlib
<spacekitteh>
any idea what causes it?
<nathan7>
spacekitteh: lack of system roots
<nathan7>
spacekitteh: or out-of-date system roots
<spacekitteh>
how do i update them
<spacekitteh>
just nixos-rebuild switch --upgrade?
<nathan7>
I've had `environment.variables.SSL_CERT_FILE = "/etc/ssl/certs/ca-bundle.crt";` set for ages because otherwise nix-shell goes ahead and overrides them
<nathan7>
(when inside nix-shell)
<spacekitteh>
the new cert is only a week old and i haven't rebuilt in over a week
<spacekitteh>
it's not a root cert tho :\
<nathan7>
it wouldn't be
<nathan7>
root certificates are the internet equivalent of nuclear materials
<spacekitteh>
indeed :V
mkoenig has quit [(Read error: Connection reset by peer)]
<nathan7>
(for one, if you're playing with them, an awful lot of "national security" folks start showing an interest in you)
<spacekitteh>
:V
<gchristensen>
they're semi-important ;)
<spacekitteh>
i dream for a day when CAs are no more
<spacekitteh>
XD
<spacekitteh>
but yeah
<spacekitteh>
how can i fix this?
<spacekitteh>
:\
<nathan7>
not sure, seems to work with my not-too-recent roots?
<ToxicFrog>
latest version is 3126 and the derivation in nixpkgs was updated to that three months ago
<ToxicFrog>
And I did a nix-channel --update a few days ago
mkoenig has joined #nixos
siel has joined #nixos
<ToxicFrog>
it is too late to be debugging this but I will definitely be poking at it in the morning
AllanEspinosa has quit [(Ping timeout: 250 seconds)]
<Rotaerk_>
is there some way to mark a derivation as something that may produce a different result if it's run at a later time? I have one that's fetching files from elsewhere, and if they change, running the derivation again will produce different files
<Rotaerk_>
but it seems that when I run the derivation again, it just points me straight to the location in the store that houses what was originally generated
<Rotaerk_>
or maybe a way to explicitly tell it to re-run it
stepcut has quit [(Remote host closed the connection)]
edvorg has joined #nixos
siel has quit [(Ping timeout: 246 seconds)]
<ToxicFrog>
Oh, I see the problem -- it's up to date in master but not in release-16.09
<Rotaerk_>
yay, got my nix expression working that allows cabal2nix to be called for hackage packages from within nix
marusich has quit [(Remote host closed the connection)]
fpletz has quit [(Ping timeout: 246 seconds)]
edvorg has quit [(Ping timeout: 268 seconds)]
siel has joined #nixos
fare__ has quit [(Ping timeout: 245 seconds)]
fpletz has joined #nixos
mkoenig has quit [(Read error: Connection reset by peer)]
siel has quit [(Ping timeout: 246 seconds)]
angerman has quit [(Quit: Bye)]
mkoenig has joined #nixos
alx741 has quit [(Quit: alx741)]
filterfish has joined #nixos
amarsman has quit [(Ping timeout: 268 seconds)]
siel has joined #nixos
zagy has joined #nixos
amarsman has joined #nixos
filterfish has quit [(Ping timeout: 250 seconds)]
siel has quit [(Ping timeout: 260 seconds)]
amarsman has quit [(Quit: amarsman)]
siel has joined #nixos
dubbelosix has joined #nixos
<dubbelosix>
hello...nix newbie here, i'm trying out the process of installing the hello world package using nix...I have my hello.nix file with the necessary functions, is there no way for me to call this directly without using all-packges.nix as the tutorial suggests?
systemfault has quit [(Quit: Bye!)]
siel has quit [(Ping timeout: 258 seconds)]
[0x4A6F] has quit [(Ping timeout: 250 seconds)]
dubbelosix has quit [(Ping timeout: 260 seconds)]
<AtnNn>
dubbelosix: yes, you could use something like `(import <nixpkgs> {}).callPackage'
mkoenig has quit [(Read error: Connection reset by peer)]
proteus-guy has joined #nixos
filterfish has joined #nixos
filterfish has quit [(Remote host closed the connection)]
mkoenig has joined #nixos
filterfish has joined #nixos
siel has joined #nixos
zagy has quit [(Quit: Leaving.)]
zagy has joined #nixos
siel has quit [(Ping timeout: 240 seconds)]
ebzzry has quit [(Ping timeout: 240 seconds)]
siel has joined #nixos
mkoenig has quit [(Read error: Connection reset by peer)]
amarsman has joined #nixos
mkoenig has joined #nixos
siel has quit [(Ping timeout: 258 seconds)]
siel has joined #nixos
Wizek_ has joined #nixos
Wizek has joined #nixos
siel has quit [(Ping timeout: 240 seconds)]
jacob_ has joined #nixos
jacob_ is now known as Guest3002
Wizek__ has joined #nixos
johann__ has joined #nixos
Wizek_ has quit [(Ping timeout: 252 seconds)]
Wizek_ has joined #nixos
siel has joined #nixos
Wizek has quit [(Ping timeout: 250 seconds)]
Wizek__ has quit [(Ping timeout: 250 seconds)]
indi_ has quit [(Remote host closed the connection)]
indi_ has joined #nixos
Itkovian has joined #nixos
deepwater has joined #nixos
moet has joined #nixos
indi_ has quit [(Ping timeout: 265 seconds)]
<moet>
i have services.redshift.enable set to true in my configuration.nix, but i noticed that redshift usually doesn't adjust for different times of day
<moet>
i've also set the lat/lon correctly
<moet>
so i've come here simply to ask if there's a way i can restart the service
<moet>
it's visible under `systemctl status` but i can't seem to call `systemctl restart redshift` or anything similar with success
<moet>
hmm.. i just did pkill redshift and something restarted it..
filterfish has quit [(Read error: Connection reset by peer)]
Wizek has joined #nixos
jp303 has joined #nixos
herzen_ has joined #nixos
mitchty_ has joined #nixos
iyzsong has joined #nixos
offline has joined #nixos
Wizek has quit [(Ping timeout: 264 seconds)]
siel has joined #nixos
jgertm_ has joined #nixos
srhb has joined #nixos
<moet>
ok.. i'll look into it
rly has joined #nixos
garbas has joined #nixos
filterfish has joined #nixos
evangeline has quit [(Ping timeout: 258 seconds)]
filterfish has quit [(Remote host closed the connection)]
siel has quit [(Ping timeout: 245 seconds)]
evangeline has joined #nixos
Wizek has joined #nixos
Wizek has quit [(Ping timeout: 268 seconds)]
indi_ has joined #nixos
jgertm_ has quit [(Ping timeout: 248 seconds)]
filterfish has joined #nixos
filterfish has quit [(Remote host closed the connection)]
michiel_l has joined #nixos
tommyangelo has joined #nixos
indi_ has quit [(Ping timeout: 265 seconds)]
<tommyangelo>
so I made a shell.nix which looks like this: http://pastebin.com/PeeCF0vw but when I run nix-shell in that directory, the docker-compose command is not available. What am I doing wrong?
moet has quit [(Ping timeout: 250 seconds)]
pi3r has joined #nixos
Wizek has joined #nixos
Guest3002 has quit [(Remote host closed the connection)]
Wizek__ has joined #nixos
edvorg has joined #nixos
Wizek_ has quit [(Ping timeout: 252 seconds)]
siel has joined #nixos
filterfish has joined #nixos
zraexy has quit [(Ping timeout: 264 seconds)]
filterfish has quit [(Remote host closed the connection)]
filterfish has quit [(Read error: Connection reset by peer)]
<Mic92>
olejorgenb: you can use a wrapper to extend PATH
<Mic92>
olejorgenb: take a look at makeWrapper
siel has quit [(Ping timeout: 258 seconds)]
johann__ has quit [(Quit: Leaving.)]
<olejorgenb>
Mic92: thanks, but my case is actually a bit worse, since the shell scripts are zsh completion functions. Ie. they are sourced, not executed
<Mic92>
olejorgenb: these binaries are meant for interactive usuage?
Wizek__ has quit [(Ping timeout: 250 seconds)]
<Mic92>
usage
<olejorgenb>
in this particular case I don't think it's actually a problem though. I only depend on coreutils and nix, and it doesn't make sense to use the package without those programs in the path
<olejorgenb>
ic92: yes, they define how "nix" is completed in zsh. (so to clearify: the scripts/shell-code isn't meant to be run, but to be sourced by zsh. The binaries used inside the script is in this case nix, find, etc)
<oleks>
Hi, can someone explain how it is intended that I modify my PATH variable (among other shell profile-related tasks) in NixOS? It seems that my .bash_profile is never loaded on login with slim.
<Mic92>
oleks: does bashrc works for you?
bfrog has quit [(Ping timeout: 260 seconds)]
<oleks>
Mic92: yes
<Mic92>
oleks: slim does probably not run your desktop environment through a login shell.
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/v1xP3
<NixOS_GitHub>
nixpkgs/master 52a220a Ole Jørgen Brønner: nix-zsh-completions: 0.2 -> 0.3
bfrog has joined #nixos
<LnL>
nice! those are the attribute completion changes right?
johann__ has joined #nixos
filterfish has joined #nixos
<olejorgenb>
LnL: yes
filterfish has quit [(Remote host closed the connection)]
* LnL
runs git pull
<olejorgenb>
LnL: great if you test it out
edvorg has quit [(Remote host closed the connection)]
<LnL>
nix-build -A <tab>
<olejorgenb>
Myself my brother have used it for about 6 months, but the possibility of config differences causing trouble is always there I guess
<LnL>
this is awesome!
mudri has joined #nixos
edvorg has joined #nixos
<olejorgenb>
LnL: yep, it makes it much nicer to work on the command line. It would be nice with descriptions (like apt-get have), but we haven't figured out how to make that fast. (ie. avoiding evaluation almost all derivations ^^)
siel has joined #nixos
<LnL>
right, I don't think you can query meta without evaluating the drv
<olejorgenb>
yep, and then the only option left is caching - not sure if it's worth it
pi3r has quit [(Ping timeout: 256 seconds)]
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 2 new commits to master: https://git.io/v1xXE
<NixOS_GitHub>
nixpkgs/master 17322ca Hoang Xuan Phu: certbot: 0.6.0 -> 0.9.3
<NixOS_GitHub>
nixpkgs/master 5a31ad1 Hoang Xuan Phu: simp_le: switch to zenhack/simp_le...
filterfish has quit [(Remote host closed the connection)]
Wizek has quit [(Ping timeout: 250 seconds)]
<rly>
ikwildrpepper: if I do that I get a complaint about allowUnfree missing, but I already have that exact line in my configuration in the right place.
Wizek has joined #nixos
Wizek__ has joined #nixos
<rly>
ikwildrpepper: I suspect this happens, because I use multiple nixpkgs trees.
siel has quit [(Ping timeout: 258 seconds)]
<gchristensen>
niksnut: nathan7 says that Go error is: it attempts to make a kuid/kgid using the userns + uid/gid pairs. if that fails, scm_check_creds returns EINVAL. which is weird, it should be able to construct a kuid/kgid for 0/0 inside the userns, as long as *something* is mapped to root. I think there might be something wrong with how we construct the userns. -- is that possible?
<rly>
ikwildrpepper: so, while I do applaud the error message, I suspect it's not an accurate error message, because it makes assumptions over the specific configuration one is using.
<LnL>
niksnut: that might explain the go failure IIRC
<gchristensen>
niksnut: oh I see... I don't know, then, maybe there should be something mapped to root? I'm out of my real area of knowledge here, I'm just a bit desperate for 16.09 to be back to around 90 failing packages.
hotfuzz has quit [(Quit: WeeChat 1.5)]
filterfish has joined #nixos
<niksnut>
I don't think that's possible (unless we allocate multiple host uids to the sandbox)
arianvp2 has quit [(Quit: WeeChat 1.5)]
<gchristensen>
I see
<niksnut>
the question is why does the go test encounter a 0/0 uid
<LnL>
gchristensen: do you know if there is other stuff that runs into the same issue?
<gchristensen>
no
<gchristensen>
MoreTea: trunk-combined
<LnL>
so it's the only thing?
<MoreTea>
thanks. I should learn more about Hydra
<gchristensen>
LnL: no, I just don't know :)
zagy has joined #nixos
Wizek_ has joined #nixos
<LnL>
alright, I'll look around tonight
<NixOS_GitHub>
[nixpkgs] rardiol opened pull request #21333: nrg2iso: init at 0.4.0 (master...nrg2iso) https://git.io/v1x5u
<gchristensen>
that would be really helpful :)
Wizek__ has joined #nixos
<LnL>
that test looks very specific
<LnL>
so if nothing else runs into issues when building without a root user there's probably not much point in worrying about it
abcrawf has quit [(Remote host closed the connection)]
Wizek has quit [(Ping timeout: 252 seconds)]
c0bw3b has left #nixos []
<gchristensen>
LnL: if that is the answer, then it is the answer... but: it is testing what should be valid behavior
abcrawf has joined #nixos
<LnL>
it does make assumptions
filterfish has quit [(Remote host closed the connection)]
<LnL>
you could say it's similar to /bin/cp existing
filterfish has joined #nixos
<gchristensen>
are you looking at the reproduction he made, or the one in go's code?
<LnL>
bbl
<gchristensen>
ok
<gchristensen>
thank you for helping! :)
<niksnut>
the baffling thing about that go test is how it could *ever* have worked
Wizek_ has quit [(Ping timeout: 260 seconds)]
<niksnut>
it attempts to send a credential that doesn't match the invoking process, which is not allowed
<nathan7>
gchristensen: s/he/they/
<niksnut>
ah, but it checks for EPERM
<nathan7>
yeah
<nathan7>
it specifically wants EPERM
ok2` has quit [(Ping timeout: 252 seconds)]
<nathan7>
this works outside a userns
<gchristensen>
nathan7: oh, I'm sorry, thank you
oida has quit [(Ping timeout: 250 seconds)]
<nathan7>
yup it actually works if root is mapped in the userns
<gchristensen>
nathan7: what do you think about lnl's just-disable-the-test approach?
<nathan7>
I mean, that's obviously possible
<NixOS_GitHub>
[nixpkgs] edolstra pushed 1 new commit to release-16.09: https://git.io/v1xFL
<NixOS_GitHub>
nixpkgs/release-16.09 1dd0fb6 Eelco Dolstra: Fix go test...
<nathan7>
but writing `0 0 1` into the uid_map/gid_map when we set up the userns would also work
sdothum has joined #nixos
<nathan7>
…guess that's settled then
<niksnut>
using a root user causes all sorts of other problems
<rly>
ikwildrpepper: how do I control the version of virtualbox being built when I refer to module options to enable virtualbox? I want it to build the one from master.
JagaJaga has joined #nixos
<nathan7>
mkay
<rly>
Someone already patched virtualbox to work with 4.9 in NixOS, which is very nice, but I need some way to control the version.
Wizek_ has joined #nixos
<nathan7>
niksnut: this isn't about running *as* root though
<nathan7>
niksnut: this is about having the uid/gid mapped
<nathan7>
niksnut: ie root existing inside the userns
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<niksnut>
I'll try that
<niksnut>
not sure whether that works for non-root (diverted store) sandbox builds
Wizek has joined #nixos
<nathan7>
/dev and stuff are owned by nobody inside the userns right now, I figure
<gchristensen>
it does seem to be a fairly reasonable assumption that root will exist
<niksnut>
I can't imagine many cases where that should matter really
<gchristensen>
we apparently have about 26,000 cases where it doesn't :P
<niksnut>
?
<niksnut>
oh
<gchristensen>
the number of builds on hydra that do pass without root
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
Wizek__ has quit [(Ping timeout: 250 seconds)]
Itkovian has joined #nixos
<ToxicFrog>
kmicu: Aah. That would explain it.
<ToxicFrog>
While I'm on the topic, *is* unstable what I want if I want something relatively up to date and rolling-release-ish, but not completely untested? Unstable still runs the tests on Hydra, right?
siel has quit [(Ping timeout: 258 seconds)]
johann__ has joined #nixos
mudri` has quit [(Ping timeout: 246 seconds)]
FRidh has quit [(Remote host closed the connection)]
siel has joined #nixos
<kmicu>
NixOS shines here: try ‘nixos-unstable’ and if you don’t like it switch back to ‘nixos-16.09’. No harm, extremely easy. ( ͡~ ͜ʖ ͡°)
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<kmicu>
That schema issue is still unclear to me. How many times did that change in the past in a breaking way? I hope Nix maintainers do not change it silently w/o an announcement.
mkoenig has joined #nixos
<kmicu>
That “Warning: It is generally safe to switch back and forth between channels. The only exception is that a newer NixOS may also have a newer Nix version, which may involve an upgrade of Nix’s database schema. This cannot be undone easily, so in that case you will not be able to go back to your original channel.” is like half–pregnant woman ― ‘I’m generally pregnant’.
<copumpkin>
niksnut: I raised this at some point in a vaguely related issue, but am curious if you have thoughts on a clang stdenv for linux
* kmicu
wants to avoid situation like ‘they recommended switching between channels on #nixos/in manual and my system is broken now, what a garbage.’.
<gchristensen>
kmicu: one minute, brb :)
Fare has quit [(Quit: Leaving)]
siel has quit [(Ping timeout: 246 seconds)]
cpennington has quit [(Ping timeout: 265 seconds)]
<mpickering>
Can I use nix-shell -A and -p together?
sellout- has quit [(Quit: Leaving.)]
<kmicu>
gchristensen: No rush. If there is a warning message like ‘Dave, I cannot downgrade to previous schema b/c …’ then it’s perfectly ok, but that’s not explained in manuals.
Itkovian has joined #nixos
<gchristensen>
kmicu: I don't know what happens, but for example, if you use unstable and set nix.pkg to pkgs.nixUnstable, it'll use a new schema version from which you can't roll back easily.e
<kmicu>
mpickering: not really, those flags are not designed for that.
<mpickering>
what is the best way to modify the argument to -A to add the things I want to specify with -p?
<kmicu>
gchristensen: but I could set nixUnstable on nixos-16.09 too to mitigate the problem?
siel has joined #nixos
<gchristensen>
kmicu: that may do the trick, right
<gchristensen>
but you'll be unable to go back to stable nix in this situation
<gchristensen>
and: if someone were in this position and reported such an issue, I suspect we'd backport nixUnstable updates to stnable
<mpickering>
My actual problem is I have haskell binaries which are looking for libgmp and can't find it in the environment
<mpickering>
perhaps I am trying to solve this the wrongw ay
<kmicu>
gchristensen: Yep. Now, where are breaking changes in Nix announced? It sounds like important thing. ‘We are breaking schema now! Watch out!’.
dmi3y has quit [(Quit: dmi3y)]
<gchristensen>
no idea :)
mudri` has joined #nixos
<kmicu>
That’s my point :)
<gchristensen>
you definitely have a point, no doubt
<gchristensen>
you usually do
<kmicu>
I will stop recommending mixing stable/unstable channels.
reinzelmann has quit [(Quit: Leaving)]
<gchristensen>
it is okay to use unstable with stable for individual packages, that isn't an issue at all
<kmicu>
Keep stable on system, install unstable things with nix-shell/nix-env.
rardiol has quit [(Ping timeout: 258 seconds)]
<niksnut>
copumpkin: you mean switch to clang by default?
<gchristensen>
another thing that I try and keep in mind is unstable sometimes goes a long time without security patches, due to other breakage
<kmicu>
ToxicFrog: in your systemPackages you could drop ‘(import /root/.nix-defexpr/channels/nixos-unstable/default.nix {}).sublime3’. The correct location for your channel depends on executed nix-channel commands. Above is valid for ‘sudo nix-channel --add https://nixos.org/channels/nixos-unstable nixos-unstable’, and ‘sudo nix-channel --update’.
mitchty_ has quit [(Remote host closed the connection)]
<gchristensen>
aaahhh! we might see a 16.09 channel update today!
<ToxicFrog>
kmicu: awesome, thank you. (Do all channel definitions, including the ones the system was installed with, live in ~root/.nix-defexpr/channels/?)
<jophish>
hmm, I'm able to run nix-channel --update from one user account, but not from another
<gchristensen>
ToxicFrog: more like celebrate, we're coming up on 2 weeks without an update.
<ToxicFrog>
gchristensen: apropos of this, how often does stable get version updates (as opposed to security fixes)? Is a new stable cut twice yearly? The names seem to imply that.
<kmicu>
Only system channels, user channels are under /home/<user>/….
<jophish>
that's weird, I restarted nix-daemon and it's working fine
<ToxicFrog>
kmicu: right, I'm not using nix as not-root except for occasional nix-shell usage
seanz has quit [(Ping timeout: 245 seconds)]
<ToxicFrog>
But I'm also used to stuff in /root being "user specific stuff that belongs to root", like root's .bashrc
<ToxicFrog>
It feels really weird to have system configuration stuff in there instead of in /etc
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<gchristensen>
ToxicFrog: March and September is when new stable versions get released, and users of stable have to take action (ie: change their channel) to get the new stable version. almost immediately after a new stable version is released, the old stable version stops receiving updates
systemfault has joined #nixos
<LnL>
gchristensen: how is the 16.09 eval looking?
<gchristensen>
LnL: 26047 pass / 68 fails / 40 queued, `tested` has no failures yet, so if all goes well we'll have an update
<ToxicFrog>
Ok. And unstable gets frequent updates usually, but may be blocked by failing builds/tests for days or weeks at a time?
<LnL>
gchristensen: so basically go was doing something "weird"
<gchristensen>
yeah
<gchristensen>
oh, what? 16.09 already got an update today *sigh of relief*
siel has quit [(Ping timeout: 258 seconds)]
<kmicu>
Ah, 16.09 update, Guix release, new GCC… seasonal gifts all the way :)
Itkovian has joined #nixos
johann__ has quit [(Quit: Leaving.)]
<itorres>
hi, something funny is happening in my system
<ToxicFrog>
gchristensen: oh, nice, that is handy.
<kmicu>
Hi itorres what’s the problem?
<itorres>
I install nixos.dictdWordnet and use it every other day
<itorres>
I don't know in what moment it has stopped being linked into my system profile
<itorres>
the derivation and the result are in the store, but not linked
<LnL>
how did you "install" it?
<ToxicFrog>
So what would go wrong if I depended on nixpkgs-unstable rather than nixos? CNF stops working?
oida_ has joined #nixos
<itorres>
it's in my /etc/nixos/configuration.nix and I install it with nixos-build switch
georges-duperon has quit [(Ping timeout: 245 seconds)]
<LnL>
ToxicFrog: your system will build, but it might not work
<kmicu>
Maybe boot is full or you are booting into a generation w/o dictdWordnet.
<ToxicFrog>
(is nixpkgs-unstable intended for people using Nix without using NixOS?)
<gchristensen>
right, ToxicFrog
oida has quit [(Ping timeout: 264 seconds)]
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<gchristensen>
ToxicFrog: the tests do other things like make sure the bootloader installs correctly
<LnL>
ToxicFrog: yes, or for packages (you can add packages from an other version of nixpkgs to your systemPackages for example)
<kmicu>
ToxicFrog: nixpkgs-unstable is not tested for NixOS and can brick your booting procedure. Do not recommend using it cuz it’s gives no benefits over nixos-unstable.
<ToxicFrog>
Ok. So as a practical matter: if I'm using Nix on not-NixOS, I want nixos-unstable. On NixOS, nixos-<version> is well tested and updates frequently, but updates mainly focus on security/stability updates, not version bumps; every 6 months a new release is cut and switching to that channel gets me a whole pile of updates.
<itorres>
mmhh, ok, running "for d in /nix/var/nix/profiles/system-*; do test -d $d/sw/share/dictd && echo $d ; done" I see that I "lost it" a couple of generations ago
<itorres>
the funny thing is that it is in the configuration.nix file
<kmicu>
Add a new package to systemPackages and check if it’s available after nixos-rebuild switch.
Itkovian has joined #nixos
<ToxicFrog>
Meanwhile, nixos-unstable is based on nixpkgs master, and as a result tends to have more up to date software versions, but is less well tested and sometimes wedges completely and doesn't update at all for days/weeks while nixos-<version> is still getting security patches.
<kmicu>
ToxicFrog: for a standalone Nix you want ‘nixpkgs-unstable’ and that channel should be already set by default during installation.
<LnL>
ToxicFrog: exactly, but you can mix and match by importing form another channel if you want newer stuff
* ToxicFrog
nods
<LnL>
that's the nice thing about nix
<ToxicFrog>
So probably what I want here is to depend on nixos-<version>, switch channels promptly when a new release is cut, and if I need a more up to date version of a specific package, I import that from the unstable channel
<LnL>
it doesn't matter if one of your packages depends on a newer glibc or whatever, you can just install it
<copumpkin>
niksnut: yeah, or at least make such a package distro viable for linux
<copumpkin>
would probably eliminate a lot of the friction with darwin
<LnL>
that's an interesting idea
slack1256 has joined #nixos
<ToxicFrog>
Would it be worth having a "nixos-latest" channel that always points to the latest nixos-<version>, so people don't have to manually switch channels on every release?
<copumpkin>
of course, tons of packages today do stdenv.isDarwin when they mean stdenv.isLLVM
<gchristensen>
ToxicFrog: that'd be pretty neat
<copumpkin>
but that's stuff that we could fix
<ToxicFrog>
gchristensen: should I open an issue for it?
<gchristensen>
sure
* LnL
should automate his pypi packages generation properly
<ToxicFrog>
Is nixpkgs the right repo for that?
<itorres>
ok, installed dmidecode, it gets installed under /bin, but there's no trace of the links to share :/
pi3r has joined #nixos
<kmicu>
Sole ‘nixos-latest’ alias could be dangerous, w/o a proper update procedure. Each release has a lot of breaking changes and it can be surprising if you execute ‘nixos-rebuild switch --upgrade’ and it prints two pages of errors :)
<gchristensen>
oh true
rly has quit [(Ping timeout: 250 seconds)]
<kmicu>
It’s a surprise twice a year. Not a bad trade–off.
seanz1 has quit [(Quit: Leaving.)]
<LnL>
yeah I was thinking the same thing
ThatDocsLady has quit [(Quit: Arma-geddin-outta-here!)]
<LnL>
but a warning from nixos-rebuild would be nice
ryantrinkle has joined #nixos
<kmicu>
itorres: What’s under ‘ls /nix/store/*dmidecode*/’?
<ToxicFrog>
Yeah, I'd rather have that surprise than have the "whoops nothing has been getting updated for months because the channels switched and I didn't remember to update my configs" surprise
goibhniu1 has joined #nixos
<ToxicFrog>
noisy surprises > quiet surprises
goibhniu has quit [(Ping timeout: 252 seconds)]
<avn>
ToxicFrog: explicit better than implicit ;)
seanz has joined #nixos
<gchristensen>
if we look past the proposed solution to the real question at hand ("how do I make sure I'm on the latest release?") there are lots of ways to solve the problem, and it is a valid question and problem.
<NixOS_GitHub>
nixpkgs/master dd968d0 Thomas Tuegel: okular: propagate build inputs...
zagy has joined #nixos
phreedom has quit [(Ping timeout: 240 seconds)]
zagy has quit [(Client Quit)]
<gchristensen>
niksnut: 1. the patch for 1.7 should probably be applied to 1.6 as well, 2. did the patches get applied to unstable? related: I was reading in our cherry-picking style workflow, it is common to frequently mege stable back up in to unstable.
mkoenig has quit [(Read error: Connection reset by peer)]
mkoenig_ has joined #nixos
kahiru has joined #nixos
<kahiru>
hey, I have a second hard drive with encrypted partition, which has lvm inside and want to unlock it with a keyfile located on my encrypted rootfs. I tried using fileSystems.<name?>.encrypted, but when I set the keyfile there it never gets unlocked. However if i comment out the keyfile line it prompts me for a password for that partition and everything works. How can I make this work with a keyfile?
slack1256 has quit [(Remote host closed the connection)]
<LnL>
gchristensen: is there some sort of documentation regarding the flow?
<gchristensen>
LnL: cherry-picking workflows?
<gchristensen>
or our workflow?
fiddlerwoaroof has joined #nixos
<gchristensen>
LnL: `man gitworkflows` for the first :)
<LnL>
no our workflow
<Profpatsch>
I forgot how to access the native ghc version from the ghcjs derivation.
<LnL>
like master <-> staging
<LnL>
and the nixos release branches
<Profpatsch>
There was an attribute for that
<Profpatsch>
ghcjs.nativeGHC or something like that.
<gchristensen>
LnL: oh, not a lot... I could try and find some, and maybe write some.
<srhb>
Does anyone else experience openvpn client dropping after suspend, and maybe know of a fix? :)
<gilligan_>
can't get vpn to work in the first place hehe
pdobrogost has joined #nixos
<LnL>
gchristensen: that looks familiar, didn't know that was in the manual
\\\\\\\\\\\ has joined #nixos
warbo has quit [(Quit: Leaving.)]
mudri` has quit [(Ping timeout: 264 seconds)]
<jophish>
I've got a package I'm installing from a binary distribution. When I try to start an executable in this package it complains that it can't find libpng12.so.0. Clearly I've got to add libpng12 to the RPATH of this executable
<jophish>
I've done this before for some other binary packages, but I'm wondering if there's an automated way to do this for all the binaries in this package
<jophish>
one annoying constraint is that this package (Quartus) is very large (20GB installed) and it takes a very long time to build this derivation
<jophish>
also, installing and running this executable N times for N missing library paths is going to take a very long time :/
xwvvvvwx has joined #nixos
xwvvvvwx has quit [(Client Quit)]
MoreTea has quit [(Ping timeout: 240 seconds)]
zagy has joined #nixos
cpennington has quit [(Ping timeout: 256 seconds)]
<LnL>
jophish: you can use ldd to get a list of the shared libraries it's looking for
moet has joined #nixos
alx741 has quit [(Quit: alx741)]
alx741 has joined #nixos
siel has joined #nixos
Shou has quit [(Read error: Connection reset by peer)]
<gchristensen>
LnL: are there outstanding questions from that, that you'd be interested in knowing?
oida_ has quit [(Ping timeout: 258 seconds)]
pi3r has quit [(Ping timeout: 258 seconds)]
cpennington has joined #nixos
siel has quit [(Ping timeout: 258 seconds)]
Shou has joined #nixos
<LnL>
I don't think so, but there are some things to note about master <-> staging
zagy has quit [(Quit: Leaving.)]
k0001 has joined #nixos
<LnL>
I don't know who usually merges staging in, but I've noticed that it regularly causes rebuilds
zagy has joined #nixos
zagy has quit [(Client Quit)]
<LnL>
people should not assume that there where no rebuilds on master since the last merge
nh2_ has joined #nixos
<k0001>
Somebody here has experience using `node2nix` to package a project that needs to use `gulp-cli` during the building process? It looks like `node2nix` is not creating a suitable environment for `gulp-cli` to run, and I'm not sure how to fix it :)
<LnL>
gchristensen: and something else that's not very clear to me is what sort of things are ok to backport
dmi3y has joined #nixos
<gchristensen>
ok, cool
goibhniu has joined #nixos
<LnL>
eg. should certain darwin specific fixes be backported to release channels?
<gchristensen>
LnL: maybe open an issue and tag me in it? that way you can add to it, and then I can also work on that off-line
<gchristensen>
almost certainly not :P
zagy has joined #nixos
<gchristensen>
LnL: back in a while, good questions
<k0001>
In particular, when trying to run `gulp` within the environment created by `node2nix`, it complains about some `babel`-related stuff being missing: https://dpaste.de/uNPU
<jophish>
LnL: oh, it sems libpng is sufficient. Now I just need a way to find all the elfs to patch
<jophish>
It can't hurt adding libpng12 to the RPATH of everything
siel has joined #nixos
<LnL>
jophish: patchelf with --set-rpath or --replace-needed
<LnL>
but ideally you would fix it in the build system
<jophish>
LnL: closed source I'm afraid
<LnL>
ah it's unfree, in that case you have to use patchelf and/or wrapProgram to fix stuff
<jophish>
yeah, I've done this before for another closed source program. The pain in this case is that this package has probably hundreds of executables and takes a long time for the derivation to build
<jophish>
ldd tells me that lots of things are going to ubuntu's /usr/lib/
siel has quit [(Ping timeout: 246 seconds)]
<Dezgeg>
try ldd from nix's glibc package
<jophish>
Dezgeg: I'm using lddtree from pax-utils. I'll try nix's ldd
<LnL>
is there something special about our version?/
siel has joined #nixos
<Dezgeg>
it won't try looking in /usrbin
<jophish>
Dezgeg: nix's ldd gived the same output (aside from the offsets changing)
<Dezgeg>
doesn't sound right if it's giving things pointing to /usr/lib
<LnL>
Dezgeg: it's an unfree package
<Dezgeg>
well, that could happen if it's shipped with a rpath containing /usr/lib
ixxie has joined #nixos
aneeshusa has joined #nixos
<gilligan_>
is there anything analogous to pkgs.writeFile which would allow me to capture a a whole directory as-is as a package? Easy enough to just write out by hand of course. Was just wondering if there was some shortcut ;)
<jophish>
gilligan_: there's something like srcOnly, just a second I'll find it
<gilligan_>
jophish, oh sweet
<jophish>
yeah, pkgs.srcOnly might be what you want
<jophish>
you can even apply patches which is super useful
<gilligan_>
jophish, great. thank you! will check it out
<gilligan_>
jophish, thanks!
siel has quit [(Ping timeout: 240 seconds)]
<gilligan_>
jophish, it is used *once* in nixpkgs ;-)
<jophish>
hah, really?
<LnL>
I've never heard of it
<jophish>
we use it a bunch at work
<gilligan_>
in twister/default.nix
<jophish>
oh, actually only once here
<jophish>
L/
<gilligan_>
be that as it may- it is *exactly* what I read right now
redmq has joined #nixos
MatejC has joined #nixos
mguentner2 is now known as mguentner
jgertm_ has quit [(Ping timeout: 260 seconds)]
iMatejC has quit [(Ping timeout: 250 seconds)]
<LnL>
I feel like that should be moved to the file with all the other trivial builders
aneeshusa has quit [(Ping timeout: 250 seconds)]
redmq has quit [(Ping timeout: 258 seconds)]
<LnL>
also doesn't use stdenvNoCC or preferLocalBuild
Fare has joined #nixos
siel has joined #nixos
derjohn_mobi has quit [(Ping timeout: 252 seconds)]
bfrog has quit [(Quit: WeeChat 1.6)]
jgertm_ has joined #nixos
FareTower has joined #nixos
fare__ has joined #nixos
Fare has quit [(Ping timeout: 240 seconds)]
cpennington has quit [(Ping timeout: 264 seconds)]
siel has quit [(Ping timeout: 260 seconds)]
FareTower has quit [(Ping timeout: 258 seconds)]
rly has joined #nixos
siel has joined #nixos
<LnL>
wow, staging is in a pretty bad state at the moment
jgertm_ has quit [(Ping timeout: 252 seconds)]
<gchristensen>
LnL: probably from the breakage
<gchristensen>
and hasn't had a rerun since
<gchristensen>
LnL: not sure why a darwin fix needs backporting to 16.09?
<LnL>
that's what I thought, but I was not sure
aneeshusa has joined #nixos
siel has quit [(Ping timeout: 240 seconds)]
cpennington has joined #nixos
siel has joined #nixos
<LnL>
but I would not mind to manage a darwin channel that is a bit more stable then nixpkgs-unstable
<NixOS_GitHub>
nixpkgs/master ea46420 Eelco Dolstra: Use overlayfs instead of unionfs-fuse in the VM tests...
oida_ has joined #nixos
TheAppleMan has joined #nixos
mudri`` has joined #nixos
TheAppleMan has quit [(Changing host)]
TheAppleMan has joined #nixos
ryantrinkle has left #nixos []
aneeshusa has joined #nixos
mudri` has quit [(Ping timeout: 268 seconds)]
siel has joined #nixos
amarsman has joined #nixos
fare__ has quit [(Ping timeout: 258 seconds)]
siel has quit [(Ping timeout: 260 seconds)]
nh2_ has quit [(Ping timeout: 260 seconds)]
moet has quit [(Quit: leaving)]
Sonarpulse has quit [(Ping timeout: 258 seconds)]
AllanEspinosa has joined #nixos
<gchristensen>
gilligan_: ping
<gchristensen>
gilligan_: thinking about your build server, I think it might be beneficial for you to track nixpkgs by pinning it in your code, instead of whatever the build server happens to be on
siel has joined #nixos
<gilligan_>
gchristensen, that is what i am doing
<gilligan_>
gchristensen, this is really only for one special scenario
<gilligan_>
gchristensen, i only use <nixpkgs> to obtain pkgs.fetchFromGithub which will then receive a specific version
<gchristensen>
oh hehe ok
<gilligan_>
gchristensen, thanks for your getting back to me with that though. appreciated
<gchristensen>
you're welcome!
<LnL>
oh rip
<gchristensen>
hum?
<LnL>
you have to rebuild the entire ghc chain when you change the haskell builder?/
<gchristensen>
you poor thing :|
<LnL>
I guess that's for tomorrow
* LnL
compiles...
ixxie has quit [(Remote host closed the connection)]
<NixOS_GitHub>
[nixpkgs] pSub pushed 1 new commit to master: https://git.io/v1pK1
<LnL>
I don't want to know how many times that machine has compiled a ghc compiler
<avn>
LnL: faster than chromium ;) be sure ;)
<LnL>
yeah that one is just insane
muyfine has joined #nixos
* manveru
really wishes fetch* functions where all builtin :)
siel has quit [(Ping timeout: 258 seconds)]
<gchristensen>
yikes, why?
msd has joined #nixos
<LnL>
so many zsh updates today :D
<LnL>
gchristensen: have you seen the new completion?
<manveru>
because a lot of times when i cd into a project, it downloads nixpkgs-unstable just for fetchFromGithub
<manveru>
and then downloads the specific nixpkgs version anyway
<manveru>
hmm
<manveru>
fetchurl is builtin, no?
<LnL>
there is builtins.fetchTarball
<manveru>
nope :(
ryanartecona has quit [(Quit: ryanartecona)]
<manveru>
ah, maybe that was the one i thought of
<LnL>
but we don't have a real pure one AFAIK
<manveru>
and it doesn't even need a checksum
<manveru>
weird
<manveru>
also fetchTarball is not written as `builtins.fetchTarball` in the manual
oida_ has quit [(Ping timeout: 246 seconds)]
<manveru>
must be a typo
<manveru>
but i think i'll use that in future... just wish there was a pure version
\\\\\\\\\\\ has left #nixos []
<avn>
LnL: ~6 hours on my not very slow machine
<avn>
but spinning disks, ssd should be faster
<domenkozar>
gchristensen: sorry? :)
<gchristensen>
domenkozar: to view CVE history
msd has quit [(Remote host closed the connection)]
msd has joined #nixos
<msd>
LnL: we should chat about nss
<domenkozar>
gchristensen: sorry, very busy these days :S
<msd>
LnL: there's an isse with porting it to work on Mac. A certain compile flag's being drawn in specific to clang, when gcc's being used to build the system
<msd>
LnL: That's why it doesn't compile for Darwin
amarsman has quit [(Quit: amarsman)]
<gchristensen>
domenkozar: that is okay :)
<gchristensen>
I thihk you'll like it, though, and will only take a few minutes.
<msd>
LnL: However, the cflag's being pulled from the environment, and the Makefile flag for suggesting an alternative compiler's not working
<LnL>
msd: yeah, that sometimes happens
<msd>
LnL: it seems like we shouldn't enforce users to use clang vs gcc if unnecessary. What's the general action in this case?
sigmundv has quit [(Ping timeout: 240 seconds)]
<domenkozar>
gchristensen: what's the command again?
<msd>
LnL: in specific, -Qunused-arguments is being pulled in
<LnL>
it's the project that adds those, right?
<gchristensen>
domenkozar: git fetch origin; git remote add grahamc git@github.com:grahamc/nixpkgs.git; git fetch grahamc refs/notes/security:refs/notes/security; git checkout origin/release-16.09; git log --notes=security and then scroll down to see commit 478fa3583f8c74a86269cf6b53460433562eb5df
<dtzWill>
msd: is the Nix derivation using "isDarwin" when it should be testing something like "isClang" ?
<dtzWill>
i've also seen fixes for darwin that were clang specific that I /wanted/ --build failures that after I open up the file I realize someone arleady fixed it but just applied it to darwin instead xD :)
<domenkozar>
gchristensen: so that's editable?
<dtzWill>
so thought i'd be related O:)
<gchristensen>
domenkozar: yeah, to look behind the curtain `git checkout notes/security`
<gchristensen>
LnL: ^
<msd>
dtzWill: LnL think I know how I'm going to edl with it
<msd>
*edal
<msd>
*deal
<dtzWill>
haha okay hooray
<gchristensen>
domenkozar, LnL: (and then `cat 478fa3583f8c74a86269cf6b53460433562eb5df` which is that commit I told you to look for)
<LnL>
gchristensen: you don't push those to nixos/nixpkgs?
<gchristensen>
probably 6 years ago I sat down and read the man page for `git` and picked it up.
<gchristensen>
well, I haven't
<gchristensen>
it is still experimental
<msd>
LnL: if we're fixing an issue, should we chunk multiple package repairs (specifically meant for building on Darwin) into multiple PRs if it can be done?
<msd>
LnL: or should it all be pushed as one larger PR for easier organization?
<LnL>
it's fine to group things if they depend on each other
ryanartecona has joined #nixos
<msd>
LnL: if one package depends on another independent PR (one independent package, one dependent), should one decouple those packages?
<LnL>
if you keep the commit separate it's still easy to cherry-pick one change out
aneeshusa has quit [(Ping timeout: 265 seconds)]
<msd>
LnL: makes sense. Thanks!
Shoue has joined #nixos
Shou has quit [(Ping timeout: 258 seconds)]
<LnL>
msd: about those arguments, have you tried to build it using clang without those?
<msd>
LnL: the inability to sandbox was confusing the Makefile. I think I have it under control for now.
<msd>
LnL: the system sets using GCC to 1 is Clang is found, because if Clang is found, the Makefile assumes you're implicitly on a Darwin build
<msd>
LnL: and then it attempts to drag in flags meant for Clang using GCC, and compile with the GCC front-end that the Darwin systems comes with
<jasom>
Am I doing something wrong, or is the newest version of the firefox flash plugin really 4 years out of date?
<msd>
LnL: now it's just failing on a linker error
<msd>
LnL: Another issue with the package is that GCC isn't explicitly declared as a build input (assuming that's because on a Linux system it's drawn in implicitly) so the Makefile gets confused
<msd>
LnL: this is a pain in the ass
civodul has joined #nixos
<LnL>
yes the linux stdenv uses gcc while the darwin stdenv has clang
<msd>
LnL: does the stdenv.clang pull in the gcc symlink?
<msd>
LnL: no, because that would make an impure system, right?
<jophish>
I have a package which puts its binaries in $out/foo/bin. What's the correct way to handle this so that tools such as nix-shell work correctly wrt PATH
jsgrant_ has joined #nixos
<manveru>
jophish: either symlink to $out/bin or tell the program to use / as prefix
muyfine has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<jophish>
manveru: symlinking seems like the nicest approach, thanks
alx741 has joined #nixos
georges-duperon has joined #nixos
<msd>
LnL: is inherit (darwin) cctools; what brings in the clang/gcc symlink?
<msd>
(In a more elegant style?)
<LnL>
no that's stuff like ld and install_name_tool
<LnL>
domenkozar: I have to rebuild everything to test my dyld workaround
<gilligan_>
LnL, if the code is in better shape than nox-review i am interested ...
<domenkozar>
LnL: fun :)
<gchristensen>
LnL: I haven't seen it...
siel has joined #nixos
<gchristensen>
really that would only inform ordering for me, it is very annoying to do a mass-rebuild first and then have to wait forever to test the rest of my patches :$P
<LnL>
domenkozar: I really hope that I didn't make a typo or something
<LnL>
gchristensen: it's fine, I'm just talking about it because you were wondering what would rebuild
cpennington has quit [(Remote host closed the connection)]
<gchristensen>
ah yeah
<gchristensen>
"basically everything"
<gchristensen>
gstreamer plugins are a pain in my butt.
gilligan_ has quit [(Ping timeout: 260 seconds)]
<gchristensen>
lol ... rebuild glibc, then stdenv, then zlib ... hooray
mudri|srcf has joined #nixos
<gchristensen>
domenkozar: for these zlib issues, should I stick them in staging first? they're all rated at "low" and "normal" severity
kampfschlaefer has quit [(Ping timeout: 258 seconds)]
<gchristensen>
domenkozar: normally I'd just go straight to master / 16.09, but I feel a bit unsure about such a big rebuild :)
siel has quit [(Ping timeout: 260 seconds)]
tvon has quit [(Quit: System is sleeping...)]
ryanartecona has quit [(Quit: ryanartecona)]
<domenkozar>
gchristensen: best to go to staging I'd say
sellout-1 has quit [(Quit: Leaving.)]
siel has joined #nixos
<gchristensen>
domenkozar: can-do
<gchristensen>
domenkozar: I'm going to put it right in to release, though. no need to wait there, I think.
dredozubov has quit [(Read error: Connection reset by peer)]
siel has quit [(Ping timeout: 260 seconds)]
<NixOS_GitHub>
[nixpkgs] abbradar pushed 6 new commits to master: https://git.io/v1p5H
<NixOS_GitHub>
nixpkgs/master ecf4e22 Nikolay Amiantov: pythonPackages.virtkey: init at 0.63
<NixOS_GitHub>
nixpkgs/master 8000204 Nikolay Amiantov: pythonPackages.python-uinput: init at 0.11.2
<NixOS_GitHub>
nixpkgs/master c22b243 Nikolay Amiantov: gnome2.gnome_python: use standard build system
msd has joined #nixos
dredozubov has joined #nixos
siel has joined #nixos
sitedyno has quit [(Ping timeout: 246 seconds)]
<LnL>
rofl, that python-html5lib thread
NeQuissimus has quit [(Ping timeout: 246 seconds)]
<gchristensen>
gsnedders means well :(
siel has quit [(Ping timeout: 240 seconds)]
m3tti has joined #nixos
Wizek has quit [(Ping timeout: 240 seconds)]
<m3tti>
hi there
<m3tti>
i'm using nix 16 and i'm trying to get xmonad with xmobar working
<m3tti>
but my system tells me there is no xmobar
<m3tti>
what can i do
<LnL>
gchristensen: and here I was thinking that pip going from 0.15.x to 15.x was stupid
NeQuissimus has joined #nixos
goibhniu has joined #nixos
<aristid>
m3tti: i'm pretty sure xmobar exists, but the package name is probably different than you exist
<kmicu>
m3tti: could you share error message?
goibhniu1 has quit [(Ping timeout: 264 seconds)]
<gchristensen>
oh hi NeQuissimus :)
<m3tti>
xmobar not found
<m3tti>
is the error
<kmicu>
m3tti: what command gives you that?
<aristid>
m3tti: the attribute name to use is pkgs.haskellPackages.xmobar (possibly nixos or nixpkgs instead of pkgs)
indi_ has joined #nixos
<aristid>
quite interestingly nix-env -qaP|grep xmobar also does not list it for me
<kmicu>
To search Haskell packages you need to use ‘nix-env -qaP 'xmobar.*' -f '<nixpkgs>' --description -A haskellPackages’
<aristid>
but this attribute name i gave works
<aristid>
kmicu: ah, haskell packages are not included in the default list anymore?
<LnL>
the haskell are not searched by nix-env
<kmicu>
That’s current (and annoying) limitation of Nix.
<aristid>
i think they were in there like a year ago?
danharaj has quit [(Quit: Leaving.)]
<m3tti>
how could i fix that
<LnL>
I don't think so, there are way to many packages to list them by default
mw has quit [(Quit: ~{^.^})]
msd has quit [(Remote host closed the connection)]
<aristid>
m3tti: i already told you...
<m3tti>
i've found something mentioning installing haskellPlatform
<kmicu>
m3tti: are you on NixOS? Installing xmonad/xmobar is very easy on NixOS.
sitedyno has joined #nixos
wak-work has joined #nixos
<aristid>
m3tti: try it like this: nix-env -iA pkgs.haskellPackages.xmobar
<kmicu>
m3tti: there is no haskellPlatform in Nix(pkgs).
<aristid>
kmicu: not anymore :)
<aristid>
a VERY old blog post might reference it
mw has joined #nixos
<m3tti>
error: attribute ‘pkgs’ in selection path ‘pkgs.haskellPackages.xmobar’ not found
<aristid>
m3tti: try nixos or nixpkgs instead of pkgs
<kmicu>
(Old post tend to confuse users.)
* hodapp
realizes, slowly, he can run NixOS on servers
<m3tti>
it was nixos
<aristid>
kmicu: we just need so many blog posts that there's always a new post for google to show :D
<m3tti>
and now how could i put it into my config.nix
<kmicu>
m3tti: if you are on NixOS use ‘nix-env -iA nixos.haskellPackages.xmobar’.
<m3tti>
youst in the package section
<m3tti>
or should i put it into the extraPackages section in xmonad
<aristid>
hmm, extraPackages in xmonad might work, not sure
<hodapp>
maybe, maybe, this (plus the ability to run things in a container) will simplify some of the crap I hate about setting up certain kinds of dev envirenments, like old PHP crap
<aristid>
environment.systemPackages will probably work too
<kmicu>
m3tti: In configuration.nix in systemPackages put ‘pkgs.haskellPackages.xmobar’.
siel has joined #nixos
<kmicu>
aristid: will not work.
<m3tti>
ok thanx
<m3tti>
i love nix allready
<m3tti>
:-D
<m3tti>
thanx for the great help
<aristid>
hodapp: i think it works fine on servers, but keep in mind of course that nixos does not yet have a workforce making sure all security patches to all applications are always applied in time
m3tti has quit [(Remote host closed the connection)]
<kmicu>
extraPackages in xmonad module is for adding haskell libraries visible for xmonad.hs.
<hodapp>
aristid: whether my existing servers do is another matter
<aristid>
kmicu: i've used nix-env for some time, but i still find it confusing.
<m3tti>
i'm also trying to get into haskell
<spacekitteh>
LnL: is staging for doing mass updates to push to master all at once so hydra only builds it once?
<LnL>
spacekitteh: yup
<kmicu>
aristid: What is confusing? Could you give an example?
<spacekitteh>
kk
<m3tti>
i'm reading "learn you a haskell for great and good"
<LnL>
spacekitteh: but it usually takes a while before it will end up in master
<m3tti>
love that language already
filterfish has quit [(Ping timeout: 245 seconds)]
<aristid>
kmicu: like, LnL's example with the -qaA. i did not know that was how you would do that. i'd just git grep my ~/nixpkgs for that, to be honest
<spacekitteh>
damn, true. it's stopping me from building android :(
<hodapp>
anyone here running NixOS on EC2 or DigitalOcean or anything like that though?
<LnL>
aristid: -qa lists all packages, -A selects an attribute instead of the top level attribute
<aristid>
hodapp: not me, but the nixops application for deploying nixos to servers has native ec2 support :)
<aristid>
LnL: i think the main thing that trips me up with nix-env is how it doesn't use NIX_PATH's <nixpkgs> by default, but instead prefers channels, unless i'm saying it wrong out of confusion :)
<kmicu>
nix-mode is just highlighting :) It will not help you much.
<LnL>
aristid: yeah, that's why I generally add -f '<nixpkgs>'
indi_ has quit [(Remote host closed the connection)]
<aristid>
a while ago niksnut was alluding to getting rid of channels
<kmicu>
B/c nix-env uses ~/.nix-defexpr as default value for -f flag. That is also explained in manual :)
<aristid>
maybe that would be good :D
siel has quit [(Ping timeout: 260 seconds)]
<kmicu>
New Nix could remove ‘nix-channel’, not channels in general.
<aristid>
kmicu: i think it was using http:// urls in the NIX_PATH
indi_ has joined #nixos
<aristid>
so i guess that URL still points to a channel
amarsman has quit [(Ping timeout: 258 seconds)]
<kmicu>
(Nix tooling is a mess, but that mess is explained very well in manuals.)
copumpkin has quit [(Quit: My MacBook Pro has gone to sleep. ZZZzzz…)]
<m3tti>
one more question. Why is nodejs and ghci segfaulting?
<aristid>
kmicu: i'm not saying i can't do what i need to do when i spend enough time in the manuals, but then i end up forgetting some of the messier parts again and again :)
<aristid>
m3tti: more context please :)
<kmicu>
m3tti: what commands did you execute?
<m3tti>
just node
<m3tti>
to get to the node prompt
<m3tti>
i've installed nodejs package
<m3tti>
and with ghc it's the same
<kmicu>
‘nix-shell -p ghc --command ghci’ works here w/o problems.
<kmicu>
The same with ‘nix-shell -p nodejs --command node’.
<LnL>
what's "nobody" supposed to be in that context then?
<nathan7>
LnL: 65534/65534
<LnL>
interresting
indi_ has quit [(Remote host closed the connection)]
RickHull has joined #nixos
<RickHull>
benley: hey :)
mkoenig has quit [(Read error: Connection reset by peer)]
mkoenig has joined #nixos
<RickHull>
I'm looking at NixOps, and I'm trying to understand how it can help me stand up a cluster. say I need to edit config files for e.g. zookeeper to tell it IP addresses of other nodes in the cluster. Does NixOps help with this?
Wizek has joined #nixos
<RickHull>
what if I want the machine to put the ZK log on a dedicated disk? how can I express this?
<RickHull>
Is this better suited for a layer above NixOps?
<simpson>
RickHull: In most cases, you don't need to use IP addresses if hostnames are allowed in that configuration. All nixops machines know each other by name.
<RickHull>
simpson: let's say I'm standing up a new ZK cluster, while the old one is still serving. I may have autogenerated hostnames that still need to be plugged in
<RickHull>
i.e. static hostnames won't work, if the new cluster members start talking to the old cluster
<simpson>
RickHull: I have no good advice for migrating a ZK cluster without downtime into nixops.
<LnL>
spacekitteh: where the changes green with the new cmake?
erasmas has quit [(Ping timeout: 256 seconds)]
<spacekitteh>
nope
<RickHull>
I'm just using zookeeper as an example. this problem exists with most clustered software in some form or another
<spacekitteh>
g++ threw another internal error in cmake
<nathan7>
RickHull: nixops sets up /etc/hosts on every cluster member automatically
civodul has quit [(Quit: ERC (IRC client for Emacs 25.1.1))]
<RickHull>
i'm wondering about the general problem of needing custom configuration on cluster members, per whatever the app or service is using for configuration. e.g. How do I declare that my ZK cluster has 5 nodes, and make it so those 5 nodes can talk to each other? And what if I want ZK configured to use a secondary disk? How do I coordinate that with provisioning and express that in NixOps for the ZK config?
<RickHull>
I'm picking on ZK because I've tackled this problem in Chef and it's not pretty there
<nathan7>
you… declare 5 nodes in the config
<nathan7>
and you declare their config?
<nathan7>
I've never used Chef, but it seems implausible to me that this wouldn't be easy…?
<RickHull>
Hm, ok. Just trying to wrap my head around the Nix way of doing things
<nathan7>
like you can declare extra disks and mount 'em places