<deepfire>
LnL, absolutely, but there's a lot going on inside these phases
<LnL>
the haskell mkDerivation just has a bunch of extra phases that happen before configure/install/...
<deepfire>
sure
copumpkin has quit [(Quit: My MacBook Pro has gone to sleep. ZZZzzz…)]
<deepfire>
what I would like is some kind of way to step them -- although at this point I'm beginning to realise it's not really amenable to automation
nixnode has quit [(Ping timeout: 260 seconds)]
<LnL>
best thing I can think of is set -x; eval "$setupCompilerEnvironmentPhase"
danharaj has quit [(Ping timeout: 240 seconds)]
<LnL>
that should show you the commands that are executed an give you some idea of what went wrong
<deepfire>
yes, that might be the way to go, indeed
<deepfire>
I was just hoping for something.. more helpful : -)
<deepfire>
there's this ill-at-ease kind of marriage between imperative shell and purely functional nix in these builder definitions
<deepfire>
no real modularity or introspectability or structure in how the thing is constructed
<NixOS_GitHub>
[nixpkgs] AtnNn opened pull request #22009: vmTools: add Ubuntu 16.10 and Debian 9pre (master...vmtools) https://git.io/vMHPo
<deepfire>
it's feels like some kind of ad-hoc construct, growing more appendages as time goes..
<justan0theruser>
I'm trying to run this install script: https://paste.debian.net/plain/909683 but I'm getting error: cannot auto-call a function that has an argument without a default value (stdenv)
<justan0theruser>
when I run nix-env -i -f install-cudnn.nix
justan0theruser has quit [(Ping timeout: 248 seconds)]
<ToxicFrog>
"string doesn't represent an absolute path"
endformationage has joined #nixos
<gchristensen>
ooooeuoeu
<gchristensen>
oops
<deepfire>
clever, I have tried but hadn't got much more information from: export NIX_DEBUG=true; nix-build -A haskellPackages.gi-pango --check '<nixpkgs>' >pango.log 2>&1
<gchristensen>
I reduced hydra-queue-runner's run time by 300ms per job :o
<clever>
deepfire: its not an env variable, its an attribute you have to set inside the nix expression
<gchristensen>
no that can't be right... but I reduced it a lot ;)
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
<deepfire>
clever, from what 'ag' tells me, it is something that is at very least /used/ as a shell variable
eacamero_ has joined #nixos
<deepfire>
clever: e.g. it only textually happens inside .sh files in nixpkgs
<clever>
deepfire: nix will nuke the environment variables at the start of every build, and initialize it with all attributes inside the derivation its building
<deepfire>
*appears
<clever>
deepfire: so env vars outside of nix-build cant leak in, only attributes set on the derivation, which affect its hash
<deepfire>
aha, I just thought it's a magic thing, passed inside
<clever>
this is also how src = fetchurl ... winds up in $src during the build
eacameron has quit [(Ping timeout: 240 seconds)]
<deepfire>
so, specifying NIX_DEBUG inside the attrset argument to mkDerivation is a no-go
mudri has joined #nixos
<clever>
and haskell may be messing with the attrset some
<clever>
the nix functions for haskell
eacamero_ has quit [(Ping timeout: 240 seconds)]
<deepfire>
yes, got it..
vaibhavsagar has joined #nixos
eacameron has joined #nixos
eacamero_ has joined #nixos
c74d has joined #nixos
eacameron has quit [(Ping timeout: 252 seconds)]
eacamero_ has quit [(Ping timeout: 264 seconds)]
<Fare>
oh noes, kernel compilation! :-(
<Fare>
I don't understand why what I changed in my configuration.nix saved a type error, but oh well.
<Fare>
maybe the kernel will be done compiling tomorrow
<clever>
ive found that kernel compiles are fairly short now, compared to things like chromium
<pikajude>
short compared to chromium, lol
eacameron has joined #nixos
<Fare>
how do I increase parallelism while compiling the kernel, already?
* Fare
sets kernel to 4.9 instead of testing, and there, no more kernel compilation.
eacamero_ has joined #nixos
* Fare
reboots... see you in a better NixOS!
Fare has quit [(Remote host closed the connection)]
eacameron has quit [(Ping timeout: 255 seconds)]
mudri has quit [(Quit: WeeChat 1.6)]
AllanEspinosa has joined #nixos
* ToxicFrog
flails at nixops
<ToxicFrog>
It badly needs either:
<ToxicFrog>
- a way to recover the generated deployment keys so you can add them to authorized_keys manually when they get removed by 'nixos-rebuild switch'
<ToxicFrog>
- or a way to tell it NOT TO GENERATE KEYS IN THE FIRST PLACE and just trust that 'ssh root@host' will do the right thing
<ToxicFrog>
Documentation would also be nice.
<ToxicFrog>
'send-keys' looks like it tries to be what I want, but if it thinks you have keys for those machines it tries to use them, which is a problem because if you're running send-keys obviously those keys won't work!
<clever>
ToxicFrog: related, there is ssh-copy-id
<clever>
which assumes you can either enter a pw, or your ssh agent has a key not in ~/.ssh/id_rsa that can get in
<ToxicFrog>
clever: yeah, but that requires me to feed it the key to copy
<ToxicFrog>
and nixops stuffs the key into an sqlite3 database file somewhere
<clever>
ouch
<ToxicFrog>
And then if the key exists in the db tries to use it exclusively even if the matching authorized_key entry on the remote got deleted
<clever>
need to store it to a file and -i then
<ToxicFrog>
Right, but that requires me to extract it from the sqlite3 db first
<clever>
yeah
<ToxicFrog>
To be clear, this isn't some kind of nixops key-import mechanism; the first time you run it it generates a nixops-specific keypair for each host and uses that exclusively for all future operations.
<ToxicFrog>
On the controller it goes in the sqlite3 state file; on the remotes it goes into /etc/ssh/authorized_keys.d/root
<ToxicFrog>
...but the latter will get destroyed by a 'nixos-rebuild switch' on the remote, leaving the only copy of the public key in the sqlite3 file.
<ToxicFrog>
Which isn't a problem if you're using nixops to manage livestock, but if you're using it to manage pets -- like, say, making sure all your family computers have working, consistent backups -- it's really fragile.
<clever>
/etc/ssh/authorized_keys.d/root is managed by setup-etc.pl, and should never be touched directly
<clever>
you should either be using ~/.ssh/authorized_keys, or the configuration.nix
<clever>
so id say nixops has to have the path fixed
<ToxicFrog>
clever: the keys get inserted into the built configuration by 'nixops deploy'
<ToxicFrog>
So they aren't in configuration.nix, but they are part of the nix configuration
<clever>
also, you shouldnt be using nixos-rebuild on a nixops target
<clever>
ahh
<clever>
nixops servers shouldnt even have a configuration.nix, as far as im concerned
<ToxicFrog>
But that's *only* the configuration on the nixops controller; nixops makes no attempt to make sure the contents of /etc/nixos bear any resemblance to the configuration built on the controller.
<clever>
the config must be built by the original master that made it
Fare has joined #nixos
<ToxicFrog>
And even if you manually sync it, it won't contain the configuration injected by 'nixops deploy'.
<clever>
i would always delete configuration.nix from nixops machines
<ToxicFrog>
What I'm getting here is that nixops is fundamentally not the right tool for what I want to do and I should be using ansible on top of nix or something.
<ToxicFrog>
Deleting configuration.nix from the managed machines isn't an option because some of them are end-user machines; they need to be able to make local edits and 'nixos-rebuild switch' without destroying the world.
mizu_no_oto has joined #nixos
<Fare>
what are you trying to do and what's wrong with nixops?
<Fare>
or disnix
<ToxicFrog>
I think the root problem here is that I want nixops to only manage part of each system
<Fare>
why not all?
<ToxicFrog>
e.g. making sure everything has backup software installed, while still letting end users tinker with their local configs
<ToxicFrog>
Fare: because these are pets, not livestock. Family laptops and such.
<Fare>
if users can have their default.nix ?
zagy1 has joined #nixos
<Fare>
and/or their overrides to import?
<ToxicFrog>
This was pretty easy to do with ansible, but I figured since moving to nix I should try nixops.
<Fare>
do they build software, or just install different packages/
<Fare>
?
<Fare>
or just share the list of packages
zagy has quit [(Ping timeout: 245 seconds)]
<ToxicFrog>
(this also means that nixops is incompatible with nixos's auto-upgrade option, because every morning it'll run 'nixos-rebuild switch' and toast the nixops deployment keys; presumably the assumption is that you leave that off and instead have a timer on the controller to build and push updated configs)
<simpson>
ToxicFrog: This has led me to consider a hypothetical bastion server which manages nixops deployments.
<simpson>
Analogous to the role of a Puppetmaster.
<ToxicFrog>
(nixops also assumes that you actually *want* to build everything on the controller and then nix-copy-closure, which is the exact opposite of what I want)
<ToxicFrog>
simpson: I'm not really familiar with puppet
<simpson>
ToxicFrog: You push a git repo, CI runs, a "Puppetmaster" checks out the repo. Async with that, the Puppetmaster delivers Puppet configurations to machines, each of which check in with the Puppetmaster nightly (or whenever.)
<simpson>
So, when you push a configuration change, your fleet slowly updates over the next day/interval.
<ToxicFrog>
That's kinda-sorta similar to what I want
<ToxicFrog>
Thinking on it more, I think the behaviour I'm after is something like
<ToxicFrog>
On deploy, check all machines to see if they have local changes
<ToxicFrog>
If they do, commit and pull them
<ToxicFrog>
Then push local config to all machines and nixos-rebuild switch them
<ToxicFrog>
s/local/master/
AmiableCitizen has joined #nixos
<endformationage>
When using patchelf to change the linker's path, does it matter whether I use $NIX_CC/nix-support/dynamic-linker or ${stenv.cc.libc.outPath}/... ?
<endformationage>
Is one preferred?
AmiableCitizen has quit [(Ping timeout: 240 seconds)]
<clever>
endformationage: most places ive seen use $NIX_CC/nix-support/dynamic-linker
<endformationage>
Alright.
deepfire has quit [(Ping timeout: 276 seconds)]
deepfire has joined #nixos
<eacamero_>
Is there a way to tell nix to build a package locally instead of sending it pre-built from a cache?
<eacamero_>
In my case the cache is large and the build step is easy.
athan has quit [(Remote host closed the connection)]
<clever>
eacamero_: --option binary-caches "" i think
eacamero_ is now known as eacameron
<eacameron>
clever: What's that an option to? I thought maybe there was a way in the derivation to tell it to prefer local builds or something.
<clever>
oh that, thats also possible
<clever>
grep on nixpkgs should confirm
<clever>
preferLocalBuild = true; i think
<eacameron>
clever: Sure enough!
<eacameron>
Woot!
<eacameron>
Thanks.
mbrgm has quit [(Ping timeout: 264 seconds)]
mbrgm has joined #nixos
paroneayea has quit [(Changing host)]
paroneayea has joined #nixos
<eacameron>
Sigh...I'm getting this all wrong. I need a way to depend on things without making them part of the closure. I.e. static link. Except in my case it's literally just copying files.
<clever>
eacameron: if you copy the files, and they dont contain any internal references, there should be no runtime dependency
<clever>
eacameron: you can use grep to check for its old path within the file and see if thats causing the issue
<eacameron>
clever: Right...but nixops still thinks it needs to copy the whole closure
<clever>
eacameron: run nix-store --query --tree on the final result
<eacameron>
clever: I'd just use a symlink...but for some dumb reason this PHP app can't handle it. I think it reflects on its own path and then freaks.
<clever>
and confirm if its a dependency or not, and via what
<clever>
ahh, if your copying it from a->b, then it cant reuse the a between copy-closure runs, and every time b changes, yeah that can be a pain
<clever>
id just fix the php code
<eacameron>
clever: Yah that makes sense, and that command is awesome. So yes it's actually depending on another folder which is largely identical. I'd be fine with it copying the result, but it's copying both the result and the dependency, which is just where it copies stuff from.
<eacameron>
clever: Essentially I'm using `./src`, copying it, modifying it. But it's copying both the original source and the modified version. I don't care about the original.
<clever>
eacameron: can you pastebin the output of --tree and point out which path you dont want?
<clever>
eacameron: aha, line 4 tells me everything, the src refers to itself
<clever>
so if you copy the source, it will still internaly be refering to itself
<clever>
eacameron: run this, "grep -r l557a8749l88icgwhh8g5m247zg3n3i4 /nix/store/a3cfnz9zm2dm79kvmfqaa2bmqivmlvmn-craft-app"
<eacameron>
clever: Oh geez you're right! And that's a bug!
<eacameron>
clever: You're insane!
<eacameron>
in a good way. ;)
acarrico has quit [(Ping timeout: 260 seconds)]
dramforever has joined #nixos
RchrdB has quit [(Quit: Leaving)]
hexagoxel has joined #nixos
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
<clever>
eacameron: nix will basicaly run the same grep command every time a build finishes, to figure out what the runtime deps are
<eacameron>
clever: Seriously?
<eacameron>
Wow.
<clever>
yep
<eacameron>
So it basically *infers* the closure...
<clever>
it even runs it on binary files, thats how it finds the elf files you depend on
<eacameron>
I thought it just assumed that if it was used during the build in needed it.
<clever>
and if you have debug symbols enabled, pointing to the source, you now depend on the source at runtime
<eacameron>
clever: That's sick!
<eacameron>
Nix never ceases to amaze me.
ericsagnes has quit [(Ping timeout: 256 seconds)]
acarrico has joined #nixos
<c74d>
NixOS doesn't put program crash memory-dumps somewhere by default, does it?
<clever>
coredumps are off by default
<c74d>
a QEMU VM of mine just crashed trying to free() an invalid pointer and now I hear lots of disk I/O
<c74d>
okay, thanks
deepfire has quit [(Ping timeout: 256 seconds)]
deepfire has joined #nixos
<eacameron>
clever: Could I hardlink instead of symlink in the derivation?
<eacameron>
clever: That would solve it. ;)
<clever>
maybe, but copy-closure wont see the hardlink
<eacameron>
clever: Oh...so it'll still copy it.
ertes has joined #nixos
<clever>
yeah
deepfire has quit [(Ping timeout: 245 seconds)]
phreedom has quit [(Remote host closed the connection)]
dhess has joined #nixos
<dhess>
Hi, I've got a headless Intel NUC running NixOS. I want to use it to encode live streams using ffmpeg with hardware vaapi support. Anyone around who has experience with vaapi on a headless box?
<dhess>
Unfortunately /run/opengl-driver doesn't exist, which is breaking everything. I've got hardware.opengl.driSupport = true; and hardware.opengl.extraPackages = [ pkgs.vaapiIntel ]; in configuration.nix
<justanotheruser>
(I guess my message didn't transmit last time)
<clever>
justan0theruser: are you just testing it, or do you want to install it?
<clever>
justan0theruser: and further down, you want tar -xzvf $src, otherwise the build will fail when the sandbox is enabled
<justanotheruser>
I want to intsall it, the right(tm) way
<clever>
then you can just nix-env -iA nixos.cudnn
<clever>
justanotheruser: something else that will both help testing, and save you trouble down the road, is to install it as a packageOverride in ~/.nixpkgs/config.nix
<clever>
and months later, when you want to rebuilt it for whatever reason, you dont have to remember the exact command we used previously
<moet>
how can i tell if my processor is 32 or 64 bit? it's a Intel Atem N450.. i have very little ram (1gb) and i'm about to reinstall nixos.. i'm trying to decide if it's better to use the 32bit version to save on ram..
<clever>
moet: look for lm (long mode) in /proc/cpuinfo
<clever>
gchristensen: also, i ran this against a racklodge server lastnight, it booted up just fine, and i was able to image the entire sda, and i plan to install nixos next
<justanotheruser>
so using that paste exactly, I get this when building with evironment.systempackages containing cudnn https://paste.debian.net/plain/909784
mizu_no_oto has joined #nixos
<clever>
justanotheruser: something elsewhere in your system is referencing cudann7
<clever>
justanotheruser: review the contents of your configuration.nix and any files it references
<clever>
gchristensen: ah, i sometimes do that kind of thing with mysql, havent really looked into hydra's pgsql db yet
<justanotheruser>
clever: only references hardware-configuration.nix, and cudnn isn't anywhere except evironment.systemPackages
<clever>
justanotheruser: can you gist the full configuration.nix and nixos-rebuild output?
<gchristensen>
clever: there are loads of useless indexes on hydra's db
<clever>
gchristensen: part of it depends on the read to insert ratio
<clever>
thats the runtime graph, not the compile-time graph
<clever>
justanotheruser: if you want the compile-time graph, you need "nix-store -q --tree /nix/store/gvfhv333z17zs8b1spw27wpdyhw3rbpm-nixos-system-dawn-16.09.1512.6b28bd0.drv"
<clever>
(that path came from your nixos-rebuild pastebin)
<justanotheruser>
yep its there!
<moet>
clever: thanks for your help! going for the plunge
moet has quit [(Quit: leaving)]
<justanotheruser>
I'll accept 7.0, thank you for your service
xwvvvvwx has quit [(Ping timeout: 252 seconds)]
mguentner2 has joined #nixos
dram_phone has joined #nixos
mguentner has quit [(Ping timeout: 264 seconds)]
mexisme2 has joined #nixos
mexisme has quit [(Ping timeout: 240 seconds)]
Ralith_ has quit [(Ping timeout: 248 seconds)]
athan has joined #nixos
Ralith_ has joined #nixos
Rotaerk has joined #nixos
esmiurium has quit [(Read error: Connection reset by peer)]
Sonarpulse has quit [(Ping timeout: 255 seconds)]
AllanEspinosa has quit [(Ping timeout: 258 seconds)]
dram_phone has quit [(Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org)]
systemfault has quit [(Quit: Bye!)]
[0x4A6F] has quit [(Ping timeout: 240 seconds)]
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
mizu_no_oto has joined #nixos
mexisme2 has quit [(Read error: Connection reset by peer)]
mexisme has joined #nixos
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
georges-duperon has quit [(Ping timeout: 252 seconds)]
uralbash has joined #nixos
ebzzry has joined #nixos
justan0theruser has joined #nixos
alx741 has quit [(Quit: alx741)]
justanotheruser has quit [(Ping timeout: 258 seconds)]
cgdub has joined #nixos
cgdub has quit [(Remote host closed the connection)]
cgdub has joined #nixos
cgdub has quit [(Read error: Connection reset by peer)]
phanimahesh has joined #nixos
<hodapp>
ugh, I am about to give up even getting Turtl to *run*, much less to have any sort of sane Nix package for it
phanimah1sh has quit [(Ping timeout: 240 seconds)]
<hodapp>
right now it's just failing silently when I try to do 'make run', or 'make package' and then 'nw' to run it with nwjs
<hodapp>
it simply sits there and doesn't open any sort of window, then if I hit Ctrl-C I get "[2591:0121/004854:WARNING:x11_util.cc(1490)] X error received: serial 138, error_code 3 (BadWindow (invalid Window parameter)), request_code 4, minor_code 0 (X_DestroyWindow)"
phanimahesh has quit [(Ping timeout: 248 seconds)]
phanimah1sh has joined #nixos
<hodapp>
but I doubt that "application run from nw.js doesn't produce any sort of window on NixOS" rings a bell with anybody
endformationage has quit [(Ping timeout: 240 seconds)]
eacameron has quit [(Remote host closed the connection)]
<GiGa|Laptop>
I might update the definition for it though, as that's a really old version
altphi has joined #nixos
<GiGa|Laptop>
and it's easy enough to update the definition too
<GiGa|Laptop>
ta
<globin>
GiGa|Laptop: no worries, sure, would be happy to merge :)
<GiGa|Laptop>
Cool, might get that done later if I have chance
<GiGa|Laptop>
Depends on the family
GiGa|Laptop has quit [(Quit: Leaving)]
deepfire has joined #nixos
<Unode>
hi all, what is the recommended way to have thunderbird + enigmail working in nixOS? I have it installed but enigmail always fails decryption because (I assume) it fails to find pinentry or to open its GUI
<Unode>
so far I've been opening the keychain manually by decrypting a file on a terminal but this gets annoying pretty quickly.
<jophish_>
I need to produce a list of dependencies a package depends on, as well as the licensing information (from the meta field). Getting the list of packages isn't a problem (nix-store --query --references), it's getting the meta information from that which I'm unsure of how to do
GiGa|Laptop has joined #nixos
fadenb has quit [(Quit: WeeChat 1.6)]
fadenb has joined #nixos
bennofs has joined #nixos
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
sternenseemann has joined #nixos
LUuM has joined #nixos
Itkovian has joined #nixos
reinzelmann has quit [(Ping timeout: 255 seconds)]
<GiGa|Laptop>
globin, when it comes to requesting a pull, I've forked nixpkgs on Github, are you able to pull from my fork?
MP2E has quit [(Quit: sleep time)]
<globin>
GiGa|Laptop: creating a PR on github most likely is the easiest
simendsjo has joined #nixos
<globin>
GiGa|Laptop: just ping me there :)
ebzzry has quit [(Ping timeout: 240 seconds)]
e3d3 has left #nixos ["Leaving"]
<GiGa|Laptop>
globin, Must confess I wouldn't have a clue how to do that
<globin>
GiGa|Laptop: then push and PR that branch
Itkovian_ has joined #nixos
<GiGa|Laptop>
Ah poo, that's included both the files which have changed (as I'm doing this in the Github UI). I'll have to check it out with git properly
<GiGa|Laptop>
Got to come back to this later I'm afraid, heading out
GiGa|Laptop has quit [(Quit: Leaving)]
Itkovian has quit [(Ping timeout: 255 seconds)]
ertes has quit [(Ping timeout: 255 seconds)]
ertes has joined #nixos
eacameron has joined #nixos
Itkovian_ has quit [(Ping timeout: 240 seconds)]
<NixOS_GitHub>
[nixpkgs] timbertson opened pull request #22015: buildPythonPackage: accept developmentPrefix and send `pip` output to stderr (master...python-stderr) https://git.io/vMQvy
ebzzry has joined #nixos
<Unode>
Ralith: finally found the font culprit. Seem like the broken font is "Roboto" but I still don't know where it's coming from.
eacameron has quit [(Ping timeout: 255 seconds)]
zml has quit [(Ping timeout: 240 seconds)]
shanemikel has quit [(Quit: My MacBook Air has gone to sleep. ZZZzzz…)]
Itkovian has joined #nixos
<NixOS_GitHub>
[nixpkgs] bjornfor pushed 1 new commit to release-16.09: https://git.io/vMQvp
<NixOS_GitHub>
[nixpkgs] globin pushed 1 new commit to gcc-6: https://git.io/vMQJ8
<NixOS_GitHub>
nixpkgs/gcc-6 3bc2154 Robin Gloster: spidermonkey_1_8_5: add patch to fix build with gcc6
<Unode>
erm... does anyone know which softwares bundle DejaVu Sans?
<Unode>
I'm trying to get rid of that font (the problem wasn't fixed, just hidden by installing Roboto) to see if it solves the problem but can't figure out where the font is coming from.
Itkovian has quit [(Quit: My MacBook has gone to sleep. ZZZzzz…)]
p0le has joined #nixos
dermetfan has joined #nixos
<sternenseemann>
Unode: fonts.fonts = with pkgs; [
<sternenseemann>
erm sry
ebzzry has quit [(Ping timeout: 255 seconds)]
xwvvvvwx has quit [(Quit: WeeChat 1.6)]
<sternenseemann>
Unode: you need something like this in your configuration.nix fonts.fonts = with pkgs; [ corefonts dejavu_fonts … ];
<sternenseemann>
Unode: then dejavu is available system-wide
<Unode>
sternenseemann: I have that, but currently I removed all the fonts included there and I still get dejavu on the system
<ToxicFrog>
Oh, and very belated, but the answer to "why not disnix" for me is "I kind of lost interest when the link to the manual 404ed"
<sternenseemann>
Unode: i guess it might be part of corefonts?
<Unode>
sternenseemann: I basically have a buggy font and I'm trying to figure out which, and where it's coming from
<sternenseemann>
or some browser bundles it?
<sternenseemann>
Unode: fc-cache -v shows all fonts fontconfig knows of
<sternenseemann>
Unode: err, all folders
<sternenseemann>
Unode: fc-list is better, shows all fonts in fontconfig cache
<gchristensen>
fpletz: don't forget about that signature I sent you
Itkovian has quit [(Client Quit)]
civodul has joined #nixos
<jophish_>
hmm, doesn't seem to be working
LUuM has quit [(Ping timeout: 260 seconds)]
<jophish_>
I've added a darwin machine in remote-systems.conf (x86_64-darwin) and am trying to build a package for darwin. My machine complains that it's unable to build it and doens't seem to be looking at anything in remote-systems.conf
<jophish_>
I have sourced remote-build-env
<jophish_>
$NIX_BUILD_HOOK is set properly
<clever>
jophish_: NIX_REMOTE_SYSTEMS as well?
<jophish_>
clever: yeah, that's correct too
<jophish_>
I've restarted nix-daemon too
<jophish_>
my guess is that nix-daemon is interfering with things
<LnL>
jophish_: are the env variables set for the daemon?
<jophish_>
NIX_REMOTE is set to daemon
<jophish_>
LnL: oh, probably not
<jophish_>
I was hoping that I could set them from the user shell
<clever>
yeah, its the daemon that will need NIX_BUILD_HOOK/NIX_REMOTE_SYSTEMS set, not the slave
<LnL>
those local variables only work for a single user install
<jophish_>
gotcha
<LnL>
if you're on nixos I would recommend using nix.distributedBuilds and nix.buildMachines
<jophish_>
LnL: sadly I'm on Ubuntu
<jophish_>
getting this working on my NixOS machine was a breeze :)
Itkovian has joined #nixos
<LnL>
how's the daemon running?
<jophish_>
LnL: it's a systemd service
<LnL>
you can look at the generated definition of your nixos machine then
<jophish_>
domenkozar: the plan is to get another linux machine up at the moment, the darwin business was just to test if remote-systems was being read at all!
alx741 has joined #nixos
<domenkozar>
jophish_: ah :)
<LnL>
oh yes, you also have to do that :p
mizu_no_oto has quit [(Client Quit)]
muhammadn has joined #nixos
<LnL>
I have ''pre-build-hook = '' in the nix.conf of my builder
<jophish_>
my long build is nearly completed! I better get this working soon!
<muhammadn>
I am trying to build a package with my webapp from github but i need to use stateful directory like /var/www/webapp
<muhammadn>
how do i achieve this in my .nix file?
<jophish_>
gchristensen: hehe. All of the queries Hercules runs are generated using opaleye
<jophish_>
(a nifty haskell package for doing type safe postgresql queries)
<gchristensen>
well that is fine, almost all the queries hydra run are generated by a nifty perl package for doing postgresql queries, the point is make sure they're not bad :P
<LnL>
gchristensen: I made it easier to rebuild my container without needing a special nixpkgs checkout :)
<gchristensen>
nice!
<jophish_>
I thought that this was less of a concern with postgres as it's got a better optimizer than most databases
<gchristensen>
no
<gchristensen>
it isn't magic, you still need to treat it well, and hydra contains millions and millions of rows
<gchristensen>
heck, _my_ hydra, only a week old, has over a million rows
<LnL>
gchristensen: I was thinking that it's probably possible to add an option to nixos that does something similar to apply custom patches on top of the nixpkgs input
<gchristensen>
jophish_: I don't mean to be a jerk :/
<jophish_>
gchristensen: I'll keep that in mind, thank you!
<gchristensen>
hydra has lots of weird behaviors that aren't very good, and I frequently see simple updates take 25s to execute
iyzsong has quit [(Quit: bye)]
<LnL>
yeah, I've also noticed that
<gchristensen>
jophish_: ah ... so, opaleye, with sensible inputs should produce sensible outputs and get good performance. however, it is not impossible to give opaleye nonsensical input
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
<philipp[m]>
Anybody else here have the gajim otr plugin breaking on nixos recently?
<gchristensen>
jophish_: (I'm assuming, since it can be type-safe and not good input) like a query with a subselect which has a subselect which has an eal
<gchristensen>
jophish_: on a technical level, I think that comment is saying that if you have an index on (finished, buildstatus) and your query is `WHERE buildstatus = 0 AND finished = 1` (ie: a different order than the index) MySQL will not use your index, whereas postgresql might
<gchristensen>
that is to say, it is sensible in it reorders your query in order to fit optimal indexes
<muhammadn>
i have permission denied if i 'mkdir -p /var/www/webapp' and cp -Rp * /var/www/webapp
<jophish_>
gchristensen: as far as I know that query isn't possible in opaleye, but don't quote me on that :)
<gchristensen>
jophish_: the specifics don't matter, it is going to be possible to have poorly performing queries
ebzzry has quit [(Ping timeout: 240 seconds)]
<jophish_>
I don't doubt it. I'll keep that in mind when we come to be using some more complex ones :)
<gchristensen>
and there are behaviors that are undesirable, like doing selects in a loop, which can also have a nasty impact
xadi has quit [(Quit: Leaving.)]
<gchristensen>
one interesting behavior I'm looking at is it does an exclusive lock on a table with high write activity where there are potentially better options to avoid that lock ... sometimes that lock can take 20+s to be allocated
<NixOS_GitHub>
[nixpkgs] schneefux opened pull request #22018: vim plugins: add gruvbox (master...pkg.vim-gruvbox) https://git.io/vMQtJ
<fpletz>
gchristensen: I noticed yesterday that our signing-party package is missing some tools I need, just finishing up that work :)
<fpletz>
mbrgm: yup, I would really love to see that, but currently don't have the time to implement this myself… but happy to talk about it and share ideas
<Fare>
simpson: other solution, burn it with fire.
<Ralith>
Unode: does it affect things other than github?
<shlevy>
domenkozar: Another request to disable travis builds, please :(
<Unode>
at this point, if I could just figure out what font is actually being used for the ugly glyphs I'd be happy :/ fc-match tells me DejaVu Sans but I can't figure out how to disable it (bundled).
<MichaelRaskin>
Fare: I guess
<Unode>
Ralith: yes many other websites
<MichaelRaskin>
you could try to run mandb in buildEnv
<simpson>
Fare: Sorry, lemme back up. I want to run this program. If I wanted to remove all bad software from my computer, I would simply throw it out the window.
<shlevy>
domenkozar: Still getting meaningless emails
<Unode>
Ralith: to the point that I had decided to ignore and just live with it but the less style a page has, the more prone to using this font it is.
<Unode>
it's like the default failover font is just bad
<Ralith>
Unode: have you tried nuking all your font settings other than `fonts`? I'm suspicious of the enable* and fontconfig stuff
<simpson>
Fare: If you're interested in helping me with my XY problem, I have a website and I would like to spider a couple levels into it in order to harvest realistic URLs for a load test.
<Unode>
Ralith: I can try again
<Ralith>
I have no idea why you'd set `antialias = false` for example
<MichaelRaskin>
simpson: I think I put urls-from-page script into NixPkgs somewhere
<Unode>
that was me trying to get rid of some bluring
<Unode>
Ralith: rebooting after nuking everything fonts related
<Ralith>
if that config didn't sort it out in general I'm stumped
<Ralith>
are you on unstable? maybe latest chromium is just weird
<Unode>
yep it didn't
<Unode>
Ralith: no I'm on stable or should be.
jsgrant- has joined #nixos
<Unode>
chromium is in the config file I linked
<Ralith>
maybe chrome's developer tools can give you some insight into the font selection/rendering
<Unode>
so now firefox shows github without problems (chromium still bad) but for instance: https://www.fontspring.com/license/demo has some bad fonts in both browsers.
mizu_no_oto has joined #nixos
<MichaelRaskin>
TFW NixPkgs manual needs OpenJDK to build
eacameron has quit [(Remote host closed the connection)]
<gchristensen>
ha!
<Unode>
Ralith: tried that... so here's what I can tell
<MichaelRaskin>
I feel getting behind, I thought last time it needed GHC
<Unode>
If the page references a font that I don't have on my system I get the broken font.
<shlevy>
gchristensen: I'm sure I could, but it will take a while :/
deepfire has joined #nixos
* shlevy
hates "clever" ORMs
<Unode>
is there any way I cant perform changes to the store such as moving a file temporarily out of the way to test something?
* Fare
hates even more *stupid* ORMs.
<MichaelRaskin>
The problem is that these are the same ORMs
<gchristensen>
shlevy: me too, I really really wish I could just put that SQL in.
<shlevy>
Eh, it's not my favorite, but I can understand having an ORM for the very basics of databse interaction
<shlevy>
Beyond that though just give me the SQL
eacameron has joined #nixos
<Unode>
shlevy: you need an SQL abstraction on top of the ORM
<Unode>
that will surely fix it
<shlevy>
:D
<Unode>
and now that I put that silly idea out, I'm going to hit google and look for someone that probably already did it
mizu_no_oto has quit [(Quit: Computer has gone to sleep.)]
ntc has joined #nixos
paperd has quit [(Quit: WeeChat 1.5)]
eacameron has quit [(Read error: Connection timed out)]
<gchristensen>
shlevy: so I think this'll dramatically cut load times on this page: https://prs.nix.gsc.io/project/nixos from over 10s to below 6s. not fabulous, but a nice improvement none the less.
eacameron has joined #nixos
systemfault has quit [(Quit: Bye!)]
<NixOS_GitHub>
[nixpkgs] fpletz pushed 3 new commits to master: https://git.io/vMQ8T
<NixOS_GitHub>
nixpkgs/master ab90eac Franz Pletz: networking: fix typo in resolvconf option edns0
<NixOS_GitHub>
nixpkgs/master ce3b98d Franz Pletz: linux: 3.18.45 -> 3.18.47
<NixOS_GitHub>
nixpkgs/master 61caacb Franz Pletz: linux: 4.1.36 -> 4.1.38
graphite has joined #nixos
sellout- has joined #nixos
xwvvvvwx has joined #nixos
<xwvvvvwx>
Hey :)
<xwvvvvwx>
When I try to rebuild my system I get an error that a URL cannot be found
<Unode>
Ralith: ok so I managed to remove all fonts to the point of having chromium segfaulting for lack of a fixed/sans font :) But up until the end the bugged font was still there
<Unode>
which makes me thing that somehow it's defaulting to one of the X bitmap fonts
<Unode>
think*
<fpletz>
xwvvvvwx: you have to wait for the next channel release or use a nixpkgs git clone :)
<xwvvvvwx>
ah
<Unode>
is there any way I can completely disable any bitmap font from X?
<xwvvvvwx>
I just need to modify the NIX_PATH env var for this local checkout right?
<clever>
xwvvvvwx: you can also use -I to override things, -I nixpkgs=~/apps/nixpkgs/
<fpletz>
xwvvvvwx: yeah, or use -I nixpkgs=/path/to/your/nixpkgs
graphite has left #nixos ["Leaving"]
<xwvvvvwx>
so: sudo nixos-rebuild switch -I nixpkgs=/path/to/nixpkgs
<xwvvvvwx>
but this will potentially much more unstable?
ChongLi_ is now known as ChongLi
glenn has quit [(Ping timeout: 255 seconds)]
<fpletz>
xwvvvvwx: yup, channels are released if the tests pass, so might be a bit unstable
<fpletz>
xwvvvvwx: are you using 16.09 or master?
<xwvvvvwx>
right now I'm on the unstable nixos channel
<xwvvvvwx>
That's master but only commits where hydra passes right?
<fpletz>
you might even need to compile some stuff if hydra hasn't finished building all packages for a specific commit
<fpletz>
the master branch can have (which becomes the unstable channel if the tests pass) can be broken :)
<xwvvvvwx>
can be quite a long time between passing hydra builds?
<MichaelRaskin>
clever: that I would never know: I don't use NixOS enough to have a large chance of getting caught up in such stuff
<MichaelRaskin>
With literally NixOS
<gchristensen>
yep
<xwvvvvwx>
love it :)
<MichaelRaskin>
gchristensen: you keep mentioning Packet.net-donated resources for Hydra, does its output also got to cache.nixos.org?
<xwvvvvwx>
was there ever any thought of integrating application test suites into the hydra builds?
<gchristensen>
if you want to see terror, go find out how debian does this testing :|
<MichaelRaskin>
That actually does happen sometimes
<clever>
MichaelRaskin: ive read the entire info doc on grub before, so i'm familiar with how grub works
alx741 has quit [(Ping timeout: 258 seconds)]
<gchristensen>
MichaelRaskin: no, but I believe niksnut wants to merge these projects
<xwvvvvwx>
gchristensen: you have a link?
<gchristensen>
xwvvvvwx: nope, because they don't do it.
<gchristensen>
(AFAIK...!)
<xwvvvvwx>
:D :D claws out :D
<xwvvvvwx>
yeah I can imagine
<MichaelRaskin>
clever: I have messed up GRUB config in multiple times, I know how GRUB works, and _if_ the failure is missing /boot/ … But we have seen funnier failures
<clever>
MichaelRaskin: ive even gotten a little bored, and i read the assembly behind the stub in the MBR
<Yaniel>
like rEFInd spontaneously refusing to run on a beautiful saturday morning?
<clever>
MichaelRaskin: and then used it to decode an MBR on an actual disk, and figure out where stage 1.5 was
<MichaelRaskin>
Yaniel: I classify rEFInd and everything Apple-related as «oh the pain»
<Yaniel>
this didn't even happen on a mac
<Yaniel>
anyway, it was a fun day
<Yaniel>
*Fun
<MichaelRaskin>
Urist McRefindUser confirms?
res0nat0r has quit [(Quit: Leaving)]
<MichaelRaskin>
clever: the good thing about EFI is that now stage2 is located using an actual filesystem
<MichaelRaskin>
And stage1 is not needed
eacameron has quit [(Remote host closed the connection)]
<clever>
MichaelRaskin: yeah
<clever>
MichaelRaskin: GPT legacy solves that, by forcing you to put stage 1.5 into a dedicated "bios boot partition"
<clever>
that partition has no filesystem, grub just writes the raw stage 1.5 executable directly to it
<MichaelRaskin>
The most Fun story about GRUB that I remember is when it turned out that some large number of NixOS generations in config make GRUB fail badly. Not as in «ignore all after N», bad as in «ignore config»
<Yaniel>
MichaelRaskin: yeah my desktop mobo uses UEFI
<MichaelRaskin>
clever: I know that well
eacameron has joined #nixos
<Yaniel>
somewhat surprisingly systemd-boot hasn't caused any problems
<MichaelRaskin>
And that's also better than default
<MichaelRaskin>
They haven't changed much since the gummiboot days
<shlevy>
Have they changed anything?
<MichaelRaskin>
They cannot destroy all common sense _instantly_ you know
<shlevy>
Even back when it was gummiboot it was run by the same guys :P
<clever>
MichaelRaskin: the guid type code for the "bios boot partition", when converted to ascii, is "Hah!IdontNeedEFI"
<MichaelRaskin>
I know
<clever>
i recently had to fix the MBR on one of my drives
<clever>
and used hexdump to confirm things
<shlevy>
It always had the "bootloader timing interface" thing that was (and still is) supposedly generic but as far as I know only systemd actually supports it
<clever>
turns out, the partition label was bios boot partition, but it had the wrong typecode
<clever>
and a random magnetic in the box was holding the MBR
<clever>
until the drives got shuffled about
<clever>
the SSD's failed to boot on their own
<MichaelRaskin>
Great
<clever>
MichaelRaskin: and the entire reason i had to shuffle the drives, memtest86 said the ram in my desktop was bad
<clever>
MichaelRaskin: even if i swapped it for ram from another pc
<MichaelRaskin>
I think I have done all working combos: MBR + GRUB, GPT + BIOS-boot + GRUB, GPT + gummiboot, GPT + UEFI + GRUB
<clever>
except, memtest still said the ram in 3 other systems was also bad
<MichaelRaskin>
A damaged memtest copy?
<clever>
the gcc hardening in nixpkgs
<MichaelRaskin>
Ahaha
Fare has quit [(Ping timeout: 258 seconds)]
<clever>
memtest86 on nixos-unstable gives false errors
<MichaelRaskin>
Great
xwvvvvwx has quit [(Quit: WeeChat 1.6)]
<clever>
as for weird things ive done with grub
<clever>
i have booted my laptop with ipxe + iscsi + grub legacy + gpt
Fare has joined #nixos
<clever>
ipxe uses iscsi to open the root hdd over the LAN, and it then hijacked the legacy bios API for the primary master hdd, and ran the MBR
<clever>
so grub legacy boots via the legacy bios api, thinking its a local drive the entire time
xwvvvvwx has joined #nixos
<clever>
and that gets it far enough to load the kernel+initrd
<clever>
and then a copy of iscsistart in the initrd is enough to make linux happy
<MichaelRaskin>
Nice
<clever>
i have since switched that laptop to an internal copy of nixos, it now has ZFS on LVM on LUKS, with a plaintext /boot, all ontop of GPT
<MichaelRaskin>
I should learn to use LUKS, then make an initramfs with a ton of tools in the plain /boot/, and have unlock-capable boot on a USB key
sdothum has quit [(Quit: ZNC - 1.6.0 - http://znc.in)]
eacameron has quit [(Remote host closed the connection)]
Fare has joined #nixos
eacameron has joined #nixos
sdothum has joined #nixos
Jackneill_ has quit [(Read error: Connection reset by peer)]
mojjo has joined #nixos
eacameron has quit [(Ping timeout: 240 seconds)]
<mojjo>
hi! has anybody experience with using haskell packages which depend on external libs. I'm having troubles using 'gloss' as a dependency in a stack project. stack --nix install gloss gives me 'Missing dependency on a foreign library.'
shanemikel has quit [(Quit: My MacBook Air has gone to sleep. ZZZzzz…)]
<Unode>
hi all, a question about tmpfs filesystems. I have: http://dpaste.com/247NXRW but for some reason the size option is being ignored.
glenn has joined #nixos
<Unode>
Is the syntax incorrect?
<Unode>
by ignored I mean that the actual mount produces a 2G filesystem (default given 4G RAM) and not 3G
michaelpj has joined #nixos
glenn has quit [(Remote host closed the connection)]
glenn has joined #nixos
glenn has quit [(Remote host closed the connection)]
<bennofs>
I get Jan 21 23:16:04 c-cube hydra-server[14959]: Compress::Raw::Bzip2 version 2.07 required--this is only version 2.064 at /nix/store/v4b73k27ijng18d38d79hghxmjisif1k-hydra-perl-deps/lib/perl5/site_perl/5.22.2/IO/Compress/Adapter/Bzip2.pm line
sdothum has quit [(Quit: ZNC - 1.6.0 - http://znc.in)]
jarlg has quit [(Quit: WeeChat 1.5)]
sdothum has joined #nixos
<bennofs>
hmm looks like CompressRawBzip2 was updated in nixpkgs master
tvon has quit [(Quit: Peace out, y'all!)]
glenn has quit [(Remote host closed the connection)]
<NixOS_GitHub>
nixpkgs/master 56c6a43 Franz Pletz: zfs: add hint to try unstable version, fix typo
<Unode>
bennofs: probably due to problems with Hydra
xadi has joined #nixos
<LnL>
what problems?
pxc has joined #nixos
<mojjo>
is there a haskell hacker using stack on nixos?...
<bennofs>
mojjo: I'm not using stack yet, but planning to do that soon. What's your question?
xadi has quit [(Quit: Leaving.)]
<mojjo>
I'm trying to use the haskell package gloss in a stack project. It depends on external libraries. Stack has a explicit nix flag, which is meant to handle those cases: That's why I could not compile the project in the beginning, but after putting the needed nixos(not haskell) dependencies in the stack.yaml the compilation worked. but when trying to execute the build, an error appears which indicates that there is sti
<mojjo>
ll a lib missing.
<bennofs>
mojjo: can you show that errro?
<mojjo>
my-exe: user error (unknown GLUT entry glutInit)
xadi has joined #nixos
peti has quit [(Ping timeout: 252 seconds)]
<mojjo>
and the only dependency I added to the yaml to make it compile was 'mesa'
<bennofs>
mojjo: can you try running the program from outside the nix shell?
jsgrant- has quit [(Read error: Connection reset by peer)]
jsgrant- has joined #nixos
xadi has quit [(Ping timeout: 276 seconds)]
glenn has quit [(Remote host closed the connection)]
mojjo has quit [(Ping timeout: 252 seconds)]
glenn has joined #nixos
faffolter has quit [(Ping timeout: 245 seconds)]
target_i has quit [(Quit: leaving)]
peti has joined #nixos
mojjo has joined #nixos
<mojjo>
bennofs: sry, connection as broke. but read your msg in the log... wait
<mojjo>
bennofs: I tried this: ./.stack-work/install/x86_64-linux-nix/lts-7.0/8.0.1/bin/my-exe
<mojjo>
(outside a nix shell) but the error is the same
michaelpj has quit [(Ping timeout: 255 seconds)]
<mojjo>
The only way I got the package working was by: $ nix-shell -p "haskellPackages.ghcWithPackages (pkgs: [gloss])", then it is known to ghc and it works fine.
<mojjo>
rather pkgs.gloss
jgertm has joined #nixos
civodul has quit [(Ping timeout: 255 seconds)]
mexisme2 has joined #nixos
civodul has joined #nixos
wildjim has joined #nixos
mexisme has quit [(Ping timeout: 240 seconds)]
wildjim has quit [(Read error: Connection reset by peer)]
mexisme2 has quit [(Ping timeout: 240 seconds)]
Kingsquee has joined #nixos
mexisme has joined #nixos
<glenn>
Hi all, is this the right place for noobie (user as opposed to developer) questions? (relates to nix-env on darwin)
<mojjo>
glenn: go ahead..
<MichaelRaskin>
Are there any Nix users on Darwin who don't become developers really quickly, if only to remove an obsolete Linux-only flag?
<ToxicFrog>
glenn: I ask a lot of newbie questions here and people are pretty helpful, go for it
<MichaelRaskin>
(which means there are no newbie user questions, just newbie developer questions!)
<glenn>
coolio, thanks - so, having installed nix package manager, i'm playing with different profiles.. the prob i've to is that once i switch profile all my nix* commands vanish effectively as they are in the default profile.
<glenn>
i can kinda work around this by always having /nix/...profiles/default in the path behind the currently selected profile, but what is the correct way of handing this?
<mojjo>
glenn: what exactly do you mean by nix commands?
<glenn>
sorry, for example nix-env
<LnL>
adding entries to your path is fine
<LnL>
I don't even use the profiles.d/nix.sh helper
<glenn>
ok, cool, so thats normaal
<glenn>
ok
<glenn>
thanks LnL
cpennington has quit [(Ping timeout: 240 seconds)]
<glenn>
and mojjo
<LnL>
yes it's just a package manager, it doesn't manager your system