<Lisanna>
I've done (drv1 // drv2) and see that the result is drv2. Which attributes are being replaced which allow Nix to identify which derivation this evaluates to, such that if I didn't replace them it would continue to evaluate to drv1?
<samueldr>
that's not on the actual one I use
<clever>
samueldr: that only works until you move the drive to a different sata port
<fresheyeball>
samueldr: that's where it hangs, on mounting
<clever>
fresheyeball: the grub layer is working perfectly, and its failing to find the rootfs by uuid
vidbina has joined #nixos
<samueldr>
(I love how the displays crops to hecking)
<fresheyeball>
clever: that sounds right
<samueldr>
yep, so grub can boot, but it's *possible* that the next installs are on the wrong drive though
<clever>
fresheyeball: can i see more to the right on that last line of text?
<fresheyeball>
clever: I would need to reproduce that state
<clever>
fresheyeball: does it just happen when you boot the default option?
<fresheyeball>
clever: yes, and I tried an old generation too
<clever>
fresheyeball: can you see the same uuid its looking for, if you run `blkid /dev/sd*` from the livecd?
<fresheyeball>
clever: yes, and I mounted it
<clever>
fresheyeball: simplest fix i can see then, is to mount the root and boot fs under /mnt as normal, nixos-install --chroot, and then `nixos-rebuild boot` to repair /boot
ma271 has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 1 new commit to staging: https://git.io/vAYi7
<NixOS_GitHub>
nixpkgs/staging eb862c4 Franz Pletz: systemd: 234 -> 237...
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz deleted systemd-237 at 02dcf84: https://git.io/vAYi5
NixOS_GitHub has left #nixos [#nixos]
<fresheyeball>
clever: can you be more explicit about those commands?
<clever>
fresheyeball: mount the rootfs to /mnt, and the boot fs to /mnt/boot/, then just run `nixos-install --chroot` and `nixos-rebuild boot`
<fresheyeball>
clever: what is the root fs vs the boot fs?
<clever>
fresheyeball: the filesystem you used for / and /boot
<fresheyeball>
so I did this
<samueldr>
(it is possible to have only one partition with grub, so it's possible you don't have a /boot partition)
<samueldr>
(with grub on legacy boot)
<clever>
i prefer zfs, and i dont trust grub with zfs, so /boot has to be an older FS
<fresheyeball>
clever I did this: mount /dev/disk/by-uuid/adsafsadfkl /mnt
<fresheyeball>
and ls /mnt shows what I expect for /
<fresheyeball>
I have no clue as to boot
<clever>
fresheyeball: and did you have an fs for /boot/?
<fresheyeball>
I do have /mnt/boot currently
<`_>
hey that's not a valid uuid
<clever>
or was it just the boot directory on the root fs?
<fresheyeball>
I don't know
<samueldr>
`_: some would argue that `_ isn't a valid nickname :)
`_ is now known as validnickname
<clever>
fresheyeball: what does /dev/disk/by-uuid/adsafsadfkl currently point to?
<validnickname>
checkmate
<fresheyeball>
clever: it's the root when I boot normally
<clever>
fresheyeball: what is the device path?
<chreekat>
is there a way to clean up after nix-build -K? other than 'sudo rm'
<clever>
chreekat: i dont think there is any way to automatically clean it up
hakujin has quit [Quit: WeeChat 2.0.1]
ma271 has quit [Ping timeout: 256 seconds]
validnickname is now known as `_
vidbina has quit [Ping timeout: 240 seconds]
zearen has joined #nixos
<clever>
fresheyeball: for example, /dev/sda1 ?
<Ralith>
clever: is there a way to programmatically get the build path from the shell?
halfbit has quit [Ping timeout: 260 seconds]
<clever>
Ralith: from inside the builder, or after nix-build has failed?
silver_ has joined #nixos
spietz has joined #nixos
<Ralith>
clever: after it failed; I'm trying to integrate nix with an external CI pipeline, and I want to extract some logfiles after the build fails
<clever>
Ralith: ah, i can think of something better, one minute
ndrei has quit [Quit: WeeChat 2.0]
<Ralith>
(and also clean up the failed dir after extracting the files)
<dhess>
just had to set GOPATH, never used go before
dan_b has quit [Ping timeout: 240 seconds]
kelleyNif has joined #nixos
<Ralith>
clever: what if the failing derivation is a dependency of the derivation I'm actually passing to `nix-build`?
<clever>
Ralith: then the result symlink wont be made, and youll need to parse some of the nix output, to follow the dependency chain of failures, and find the one that passed and failed
<clever>
Ralith: hydra doesnt have this issue, because hydra manually builds every step itself, rather then letting nix walk the tree for you
<Ralith>
hm, I guess I can just build each by hand and bail out early, that'll work fine
<Ralith>
clever: also, any way to get nix-build to give me symlinks for each output?
<Ralith>
for a single multiple-output derivation
<clever>
Ralith: only for the outputs of the final derivation you pointed to with -A
<Ralith>
right, that's what I want now
<clever>
[clever@amd-nixos:/tmp]$ nix-build '<nixpkgs>' -A glibc.out -A glibc.bin
<clever>
[clever@amd-nixos:/tmp]$ ls -l result result-bin
<clever>
Ralith: try these 2 commands
<Ralith>
ahah, thanks
<clever>
Ralith: and also nix-store --query --outputs $(nix-instantiate '<nixpkgs>' -A glibc)
<clever>
or eval glibc.outputs in nix, via either nix-repl or nix-instantiate
<clever>
also have a look at the flags in nix-store -q
<Ralith>
whee, I can even combine -E and -A
<clever>
[clever@amd-nixos:/tmp]$ nix-store --query --binding postConfigure $(nix-instantiate '<nixpkgs>' -A glibc)
<clever>
Ralith: it can do fun things like this, which will get the value of the postConfigure env variable within the glibc derivation
<clever>
ah, that can also query the outputs on a .drv file
<clever>
so you can find them after you instantiate
<Ralith>
@_@
<Ralith>
sometimes nix is weird and obtuse and sometimes it is like here have twenty different interesting and useful solutions to your problem
k0001 has joined #nixos
<clever>
heh
<Havvy>
Which directory would I put a tool for generating HTML from markdown? tools/typesetting?
smichel17 has quit [Quit: smichel17]
<clever>
Havvy: maybe near pandoc, if thats not inside the hackage-packages.nix blob
<Havvy>
Hmm, github says that all-packages.nix has exactly 1000 contributors. I wonder if that's true or that's just the limit it counts.
<Havvy>
samueldr: I find it easier on my browser than anything else.
kelleyNif has quit [Quit: Leaving.]
<Havvy>
:/ Can't fetch the Rust source from Github because there's no cargo.lock
kelleyNif has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #34916: xz: set CONFIG_SHELL to /bin/sh, fix retained reference to bootstrap (staging...fix/xz-invalid-references) https://git.io/vAYSk
NixOS_GitHub has left #nixos [#nixos]
Xalt has left #nixos ["ERC (IRC client for Emacs 25.3.1)"]
thc202 has quit [Ping timeout: 248 seconds]
kelleyNif has quit [Quit: Leaving.]
alex`` has quit [Ping timeout: 255 seconds]
kelleyNif has joined #nixos
orivej has joined #nixos
Ralith_ has quit [Ping timeout: 248 seconds]
d4g_ has joined #nixos
<nahamu>
gchristensen: you around?
<nahamu>
I really think that this line 'config_evaled = import "${pkgs.path}/nixos/lib/eval-config.nix" config;' is somehow sidestepping the overlay.
<nahamu>
the code as currently pushed works, but if I uncomment the line that tries to pull in an overlay package it bails.
<petersjt014[m]>
is there any builtin function to compare config files?
<nahamu>
I can repush in the broken state if that would be more helpful.
mbrgm has joined #nixos
<adisbladis>
rycee: Hmm. Regarding https://github.com/NixOS/nixpkgs/pull/33836 I'm thinking that debian and gentoo probably has the same test failures as us in that PR except they dont run the test suite and hence never catch it. I'm not really familiar with debian packaging but it seems to me based on the contents of https://sources.debian.org/src/gpsbabel/1.5.4-2/debian/ that they dont run tests.
stephenjudkins has quit [Remote host closed the connection]
<nahamu>
brilliant!
<gchristensen>
great!
<nahamu>
thank you so much.
<nahamu>
I knew I was getting very close.
<gchristensen>
you're welcome :) so glad to help
<gchristensen>
btw you probably want to move { nixpkgs = { overlays = [ (import ./overlay.nix) ]; }; } in to a file, and then reference the file instead
<gchristensen>
otherwise the errors can be sort of nonsensical
<nahamu>
will do.
<nahamu>
how would you expect such a file to be named?
<gchristensen>
I don't have a good suggestion :(
<gchristensen>
maybe overlay-module.nix or something
<petersjt014[m]>
where in the fs would I find old generations/config files?
<gchristensen>
/nix/var/nix/profiles/system-*
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<gchristensen>
ok nahamu I'm heading out for the night. good luck, have fun :)
notdaniel has joined #nixos
<nahamu>
thanks gchristensen! have a great evening.
<gchristensen>
you too
xcmw has joined #nixos
<petersjt014[m]>
Hmm. This is getting tricky
<petersjt014[m]>
Is there any got way to just cat out an old `configuration.nix`?
notdaniel has quit [Client Quit]
<petersjt014[m]>
*(any way)
<petersjt014[m]>
I broke something between configs and I'm trying to do a diff somehow to see what
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Ubercow has quit [Quit: Ubercow]
Ubercow has joined #nixos
alfie is now known as {{refimprove}}
{{refimprove}} is now known as alfie
xcmw has joined #nixos
kelleyNif has quit [Quit: Leaving.]
<fpletz>
petersjt014[m]: unfortunately not because it is compiled, but there is the system.copySystemConfiguration option for the future
kelleyNif has joined #nixos
<fpletz>
it's no silver bullet though as it conly copies the actual configuration.nix file and nothing more (i.e. imports)
<adisbladis>
It wont help you now but protip for the future: Keep your configs in git (or your revision control system of choice)
<petersjt014[m]>
sounds good
pkill9 has quit [Ping timeout: 256 seconds]
<petersjt014[m]>
though is there an easy way to show a config file after the imports have been imported
<nahamu>
on my system, one of those matches my current /etc/nixos/configuration.nix
Supersonic112 has joined #nixos
Supersonic112 is now known as Supersonic
<petersjt014[m]>
`find: warning: you have specified the -maxdepth option after a non-option argument -type, but options are not positional (-maxdepth affects tests specified before it as well as those specified after it). Please specify options before other arguments.`
<petersjt014[m]>
but after browsing there otherwise I did find it
<nahamu>
swap the ordering.
<petersjt014[m]>
I think
<nahamu>
"find /nix/store/ -maxdepth 1 -type f -name \*configuration.nix\*"
<nahamu>
"find /etc/nixos/configuration.nix /nix/store/ -maxdepth 1 -type f -name \*configuration.nix\* -exec md5sum {} + | sort" if you want to be fancy
<nahamu>
anyway, good luck!
<petersjt014[m]>
It did find it, but apparently followed a symlink to /nix/store/configuration.nix
<petersjt014[m]>
hmm. well it's a start
kelleyNif has quit [Quit: Leaving.]
<petersjt014[m]>
I might just switch between gens and cat out each to a file
<petersjt014[m]>
and then set up a git thing later
<genesis>
prefix = /usr/local in a lot of makefile, i used to override to / but should i ,
<genesis>
?
Ross has joined #nixos
<genesis>
and prefix=$out is not so convenient when they prepend DESTDIR to $prefix
<Lisanna>
I'd like to add some derivations as "tests" to some function I've written in Nix. Meaning, when this function is used, I would like my test to be forced to be built, even though it's not listed as a dependency for any of the real derivations I'm actually trying to build (and probably shouldn't be). Is there a way to force evaluation and building of a derivation when some Nix code gets evaluated?
<adisbladis>
Lisanna: You could just add a default.nix with your tests as attributes and build it?
<adisbladis>
Or am I misunderstanding?
<Lisanna>
err
<Lisanna>
I'm trying to gate my function's usage with some tests passing
<Lisanna>
I have some pure Nix tests written in the form of asserts
<adisbladis>
Ahh check. Now I get it.
<Lisanna>
so if someone tries to use this function and it doesn't pass the tests, the asserts will fire
<adisbladis>
Hmm..
<Lisanna>
one of the tests is a runCommand though, and so I would like to force that runCommand to be built if this function gets used somewhere
<Lisanna>
so that an error can be printed if it exits with non-zero
<Lisanna>
I keep thinking back to the trace functions, wondering if I could use those somehow
<Lisanna>
since they can force evaluation
<Lisanna>
not sure if they can force building though
<infinisil>
Lisanna: builtins.seq maybe
<infinisil>
takes 2 arguments, evaluates the first, returns the second one
<Lisanna>
infinisil nope, just tested. forcing evaluation doesn't force building
<infinisil>
Lisanna: Oh, how about importing the result of it
<infinisil>
then it has to build it to evaluate the nix code
kelleyNif has quit [Quit: Leaving.]
<Lisanna>
isn't that only in 2.0?
<infinisil>
nah, that's import-from-derivation
<Lisanna>
x_x
<infinisil>
But you might want to put some meaningful nix in $out
kelleyNif has joined #nixos
<infinisil>
or maybe just builtins.readFile actually
<Lisanna>
wait, you're telling me if I do: import "${drv}/default.nix" then it will evaluate drv, build drv, and then evaluate the import?
<infinisil>
Lisanna: indeed!
<Lisanna>
witchcraft
<Lisanna>
how did I not know about this
<infinisil>
that's what IFD is all about
<Lisanna>
I'd never heard of IFD before now
<Lisanna>
ugh\
<Lisanna>
...now I need to go back and check over my entire stack to see if I'm doing waay more work than I need to be somewhere
<Lisanna>
thank you for telling me about this ^^
<Lisanna>
for some reason I thought this wasn't coming until 2.0... I guess I was confusing this with fetching nix code from the internet and then evaluating it
<infinisil>
Lisanna: But, IFD is slow
<infinisil>
Lisanna: and a no-go for nixpkgs
<Lisanna>
I'm sure, it has to do a whole build cycle as part of the evaluation
<infinisil>
exactly
<Lisanna>
even if the build is cached it'll add a lot of overhead
<Lisanna>
will it do the IFD every time it's encountered in a Nix expression, or does it memoize the result?
<infinisil>
Lisanna: and what *is* coming with 2.0 is builtins.fetchgit, which can fetch stuff fast during evaluation (the current pkgs.fetchgit does it during build time), which means the result can be used to evaluate nix code without requiring the slow IFD
<infinisil>
as far as i understand it
<infinisil>
Lisanna: well if the result is built already it won't have to build it again if all the inputs are the same (meaning the result path will be the same)
<infinisil>
so it does do caching
<Lisanna>
yeah, not talking about the build itself, but the overhead from doing the IFD itself
<Lisanna>
so if you use the import in 3 places, do only pay the build overhead penalty once, or three times?
<Lisanna>
(Assuming in both cases the /nix/store path for it already exists)
<infinisil>
Not sure about the IFD overhead itself
<Lisanna>
I'm assuming there's some overhead of checking and retrieving the result from /nix/store, and computing the hash so it knows where to look in /nix/store
<infinisil>
Lisanna: only once, the nix daemon won't start more than 1 build on the same derivation
<Lisanna>
I guess my general question is does Nix do memoization
<contrapumpkin>
well, a thunk is still only evaluated once
<contrapumpkin>
so you can build stuff like memocombinators
<infinisil>
not sure if this new builtins.memoise is in master
<Lisanna>
oh, okay, guessing that's coming in 2.0
Ubercow has quit [Quit: Ubercow]
<Lisanna>
anyways, cool, this could be used to write a general-purpose "tryRunCommand" that will return true or false if the given commands succeed or fail
Ubercow has joined #nixos
<Lisanna>
...and with a bit more work, a generic "tryBuild" function
nuncanada has quit [Ping timeout: 256 seconds]
<Li[m]>
so if I understand overlays correctly, that could be a way to build a system with different init, or libc. correct?
<infinisil>
Li[m]: I think so yeah
slyfox has quit [Ping timeout: 256 seconds]
slyfox has joined #nixos
<contrapumpkin>
libc would be tricky
<contrapumpkin>
init would work differently
<contrapumpkin>
this is more for replacing or tweaking dependencices, or augmenting the package set
<Li[m]>
but nixpkgs defines the whole OS
<contrapumpkin>
it defines the entire package set, and then nixos is built on top of that
<contrapumpkin>
the part that says thou shalt use systemd is just a nixos module
<Li[m]>
so we could override `services`
<contrapumpkin>
admittedly a module that's super widespread
<contrapumpkin>
pretty sure we have packages for other init systems
<contrapumpkin>
just not modules that use them
<Li[m]>
but lazily called, so a minimal system would be easier to build, right?
<Li[m]>
i cant wait to build router images for those constrained cheap mips
<Li[m]>
nix is so naturally fit for that
Ubercow has quit [Quit: Ubercow]
<contrapumpkin>
porting to different systems can sometimes be tricky
<contrapumpkin>
we're still ironing out the kinks in our ARM packages
<Li[m]>
I bet
<contrapumpkin>
but yeah, it's a good fit for a lot of stuff :)
<Li[m]>
arm64 is a really interesting platform too, but the HW is all locked out
<Li[m]>
maybe the new ios bootloader code leak can help
<Li[m]>
nixos on an iphone?
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
xcmw has joined #nixos
<tnks>
I'm pretty sure it's possible to retrieve a binary from a Nix cache with just the /nix/store path.
<tnks>
I think I did it before, but forget with which commad.
<Ralith>
tnks: `nix-store --realise`?
<tnks>
Ralith: just /nix/store/hashhashhash-name-whatever?
<Ralith>
check the manpage
Ubercow has joined #nixos
schoppenhauer has quit [Ping timeout: 256 seconds]
schoppenhauer has joined #nixos
hamishmack has quit [Quit: hamishmack]
codygman has joined #nixos
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Lisanna>
what are all the unique ways in which a derivation can fail to build? The two that I can think of right now are: the process exits with non-zero exit, and failure to produce one of the output paths
<Lisanna>
are there any others?
halfbit has joined #nixos
<contrapumpkin>
fixed-output derivation hash doesn't match
<contrapumpkin>
no available build users :)
<Ralith>
various types of resource exhaustion
<Lisanna>
OK... slightly different question: if the builder has started executing, what are all the unique ways it can fail to build?
<simpson>
What happens if e.g. the build phase doesn't return, but runs forever?
<Lisanna>
...I think nix has a build timeout option?
<Lisanna>
that's a good one though
<simpson>
What's your goal?
<Lisanna>
I'm writing a function called tryRunCommand, which returns true if a given runCommand succeeds, and false if it fails
<Lisanna>
so, trying to cover as many bases as feasible so that you get "false" instead of your whole Nix expression evaluation getting interrupted
<simpson>
Interesting.
<Lisanna>
not sure if I could handle the build timeout one without knowing what timeout value the nix config is set to though
<Lisanna>
also infinisil : I tested the IFD thing and if it's already been built once (the /nix/store path exists) it seems to evaluate without any noticeable overhead, so wondering why IFD isn't allowed in nixpkgs if you only need ot pay the penalty once?
<infinisil>
Lisanna: there are some comments on github explaining it, hold on I'll try to find them
<infinisil>
Lisanna: Well I can't find them, but the arguments are: distributed builds not possible, cross-compilation not possible, and some other things
<infinisil>
or maybe cross compilation is possible
<infinisil>
there are some problems with build being run by the evaluator and not the queue runner
<petersjt014[m]>
I have this nice little command that I picked up somewhere, and can't seem to make it work as a script--not sure how to the escapes work
<adisbladis>
Shells do not do any kind of interpolation in single quoted strings
<petersjt014[m]>
single quote is liter and un-literalif?
<petersjt014[m]>
I guess that makes sense
<petersjt014[m]>
that's what quotes are i guess
<petersjt014[m]>
still, the possible nesting doesn't sit well with me
Drakonis has quit [Read error: Connection reset by peer]
<petersjt014[m]>
same reason lisp scares me i guess
pie_ has joined #nixos
Arcaelyx has joined #nixos
reinzelmann has joined #nixos
Arcaelyx_ has quit [Ping timeout: 260 seconds]
Guanin has quit [Ping timeout: 268 seconds]
endformationage has quit [Quit: WeeChat 1.9.1]
pie_ has quit [Ping timeout: 240 seconds]
Lisanna_ has quit [Quit: Lisanna_]
Lisanna_ has joined #nixos
dj_goku has quit [Ping timeout: 248 seconds]
lsyoyom has quit [Ping timeout: 268 seconds]
chrisbarrett has quit [Ping timeout: 264 seconds]
dj_goku has joined #nixos
dj_goku has quit [Changing host]
dj_goku has joined #nixos
Guanin has joined #nixos
Lisanna_ has quit [Ping timeout: 256 seconds]
lsyoyom has joined #nixos
Lisanna has quit [Read error: Connection reset by peer]
<infinisil>
Probably the cleanest way to do that command is to use '{ file }: with import <nixpkgs>{}; callPackage file {}' and add --argstr file "$1"
rauno has joined #nixos
<infinisil>
petersjt014[m]: and make $1 absolute via realpath or so
<infinisil>
Which should work with any file names, however odd they might be
pie_ has joined #nixos
Mateon3 has joined #nixos
Mateon1 has quit [Ping timeout: 268 seconds]
Mateon3 is now known as Mateon1
strflw has joined #nixos
Ubercow has quit [Quit: Ubercow]
jmeredith has quit [Quit: Connection closed for inactivity]
taktoa has joined #nixos
Arcaelyx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cocreature>
does nix 2.0 have a replacement for "nix-env -i"?
orivej has quit [Ping timeout: 240 seconds]
seanparsons has quit [Read error: Connection reset by peer]
MercurialAlchemi has joined #nixos
<infinisil>
cocreature: I'd be surprised if it doesn't
<cocreature>
infinisil: well nix-env is still a thing, I was wondering about the unified nix command (sorry should have been more precise)
<infinisil>
cocreature: Hmm, I thought so too, but it apparently doesn't have a replacement (yet?)
<cocreature>
infinisil: at least that means that I’m not just too stupid to find it, thanks :)
<infinisil>
cocreature: Pretty much all of the previous commands will still be in nix 2.0 anyways though, stuff like nix-build, etc.
<cocreature>
sure but I’d like to use the new shiny things where possible :)
seanparsons has joined #nixos
<infinisil>
yea same
Tobba_ has joined #nixos
Tobba has quit [Ping timeout: 256 seconds]
JosW has joined #nixos
ssmike has joined #nixos
hyper_ch2 has joined #nixos
Lisanna has joined #nixos
<Lisanna>
i just thought of a way to implement isos in nix
<Lisanna>
step 1: use md5 hashes
<Lisanna>
step 2: implement a nix drv which calculates md5 collisions
<Lisanna>
step 3: modify the two drvs you want to be isomorphic according to what the collision finder says
<Lisanna>
(in all seriousness, I wish nix had a feature like this. to choose between two drvs depending on if one is already available)
<Lisanna>
but I guess there's no w
<Lisanna>
way to test if a nix drv is already built or not
kiloreux has joined #nixos
kiloreux has quit [Remote host closed the connection]
kiloreux has joined #nixos
alex`` has joined #nixos
<infinisil>
Lisanna: this is on a higher level than nix
<infinisil>
Since it's stateful
<infinisil>
So if you want to do something like this, use the commands nix-store, etc. to do the logic
<Lisanna>
wait, couldn't I use fixed output derivations for this?
<infinisil>
for what?
<Lisanna>
if two fixed output drvs have the same output, but a different means of getting there, if one of them was already built, it can just use that cached product, right?
<Lisanna>
since they should have the same /nix/store outpaths
<Lisanna>
the sha of a fixed output drv only depends on the product, not the inputs, right?
<infinisil>
Lisanna: that's not how cryptographic hashes work though, it's practically impossible to do that
boxofrox has quit [Ping timeout: 256 seconds]
Tucky has joined #nixos
<infinisil>
yeah sure it would work, if you managed to get the same hash for 2 different things
<infinisil>
but that defeats the whole point of fixed output derivations
<Lisanna>
"touch $out" vs. echo -n "" > $out
rogue_koder has quit [Quit: Konversation terminated!]
<infinisil>
Lisanna: yeah that works, but not sure how this could be useful
<Lisanna>
I think I have quite a cool use case for this
<Lisanna>
imagine you have a test suite
asuryawanshi has joined #nixos
<Lisanna>
there are two ways to run a test from that suite against your reference model: by itself, or with other tests
<Lisanna>
and you may want to do one or the other depending on your motivations
boxofrox has joined #nixos
<Lisanna>
running one test by itself is faster overall for the one, but slower if you did that for each individual one
<sphalerite_>
michalrus: I think you also need gc-keep-derivations
<Lisanna>
brb
<infinisil>
Lisanna: interesting!
<sphalerite_>
michalrus: for example, for hello, hello instantiates to ...-hello.drv which depends on ...-hello.tar.bz2.drv, so ...-hello and ...-hello.tar.bz2 will be kept. Except if you run a GC which deletes the .drvs which are what are keeping the sources alive
Myrl-saki has joined #nixos
<infinisil>
Lisanna: but.. you need a hash to even be able to build a fixed-output derivation, which means you'll have to run every test once anyways, and then you have the result and it won't be run again. Updating a single test will update the hash, so you have to do that over again
<infinisil>
(if a test change doesn't change the hash, it will use the already built result)
<infinisil>
And even if you could have the same hash for 2 ways of doing the tests, you still have to build a single derivation, there's nothing in nix that chooses either derivation
<infinisil>
(you need to use nix-store level commands for that, because it's stateful)
orivej has joined #nixos
<Lisanna>
infinisil if you specify what the expected result of a test is in the definition of the test, you could compute the sha
<Lisanna>
but, that's not ideal
<Lisanna>
ideally you run the test against the reference model and record the results, then compare those results against running it against the real thing
<sphalerite_>
Li[m]: re "i cant wait to build router images for those constrained cheap mips" I got some nixpkgs-built stuff running on my old router yesterday!!
Ross has joined #nixos
<infinisil>
sphalerite_: Li[m] = Lisanna btw
<Lisanna>
what
<Lisanna>
...am I?
zzamboni has joined #nixos
kiloreux has quit [Remote host closed the connection]
<sphalerite_>
cocreature: AFAIK a replacement for nix-env -i in 2.0 is indeed planned but still not implemented.
<adisbladis>
Lisanna: Now you are.
<Lisanna>
...o...ok..
kiloreux has joined #nixos
zzamboni has quit [Client Quit]
zzamboni has joined #nixos
<sphalerite_>
Lisanna: yeah your hash collision thing + using md5 as the output hash algorithm could indeed work.
<sphalerite_>
In fact I seem to recall seeing some sort of abuse like that in nixpkgs :D
<sphalerite_>
I keep getting out of memory while tyring to use nix copy to get some stuff from one machine to another. I'm guessing this is actually "out of process address space" and not "out of available memory" since I've got 32GB of swap... But anyway, I do really need to get these paths across. Does anyone know an alternative solution?
adfaure has quit [Read error: Connection reset by peer]
<hyper_ch2>
adisbladis: that I don't know :)
adfaure has joined #nixos
zzamboni has joined #nixos
<hyper_ch2>
sphalerite_: do you have parted also installed? :)
<sphalerite_>
hyper_ch2: yeah just tried that. It said "The resize command was removed in parted 3.x"
<sphalerite_>
ended up copying the files over, deleting, recreating, and copying them back
roberth has joined #nixos
<hyper_ch2>
sphalerite_: :)
<adfaure>
Hello Nix, I have a question. I a m using nix on a grid, and I install nix on all nodes of my job. But for some reason, after the installation nix-shell is broken. I get the error: Error in tempdir() using /run/user/114/nix-shell.XXXXXX: Could not create directory /run/user/114/nix-shell.HSTMkg: Permission denied at /nix/store/gy4yv67gv3j6in0lalw37j353zdmfcwm-nix-1.11.16/lib/perl5/site_perl/5.24.3/x86_64-linux-thread-multi/Nix/Utils.
<hyper_ch2>
you have so much ram, you can just copy stuff to ram, format stick and put it back :)
<adfaure>
Ohh no my bad, I did not read. It is just a permission issue. Sorry for the noise
<sphalerite_>
hyper_ch2: no, I copied it to /tmp which is on a HDD. Although the contents of that filesystem woudl have fit into RAM too
wip_ has quit [Ping timeout: 256 seconds]
<sphalerite_>
(it's my EFI /boot)
leat has joined #nixos
zzamboni has quit [Remote host closed the connection]
spietz has quit [Ping timeout: 268 seconds]
orivej has quit [Read error: Connection reset by peer]
zzamboni has joined #nixos
<hyper_ch2>
efi is just some evil witchcraft :)
<oahong>
Hi, Anyone have broken .nix-profile in $HOME? it's a fresh 17.09 installation.
<sphalerite_>
oahong: broken how?
Lisanna has quit [Quit: Yaaic - Yet another Android IRC client - http://www.yaaic.org]
<oahong>
broken symlink, point to non-exists dir.
<sphalerite_>
oahong: as in a broken link? It's probably because you haven't got anything in your user profile yet. If you install something using nix-env it will create the first generation of your profile and "unbreak" the link
<sphalerite_>
right
<sphalerite_>
but even if you don't do that, it shouldn't be a problem
spietz has joined #nixos
asuryawanshi has joined #nixos
<oahong>
ok, thanks
asuryawanshi has quit [Remote host closed the connection]
zzamboni has quit [Remote host closed the connection]
asuryawanshi has joined #nixos
zzamboni has joined #nixos
goibhniu has joined #nixos
zzamboni has quit [Remote host closed the connection]
zzamboni has joined #nixos
wip_ has joined #nixos
<hyper_ch2>
if flexget could be made to work again, I could move another server to nixos :)
<adisbladis>
hyper_ch2: I often do that on my laptop :) Copy files to ram that is
<adisbladis>
Git clone to tmpfs <3
arjen-jonathan has joined #nixos
<sphalerite_>
adisbladis: doesn't caching usually make that useless, at least for performance?
goibhniu has quit [Ping timeout: 255 seconds]
<sphalerite_>
adisbladis: and I don't see why else you'd do it?
<adisbladis>
sphalerite_: Yes its not su much for performance I do it
<sphalerite_>
what then?
<hyper_ch2>
adisbladis: how much ram does your notebook have?
zzamboni has quit [Ping timeout: 248 seconds]
<hyper_ch2>
btw, is there any nice littel powerful notebook with 2 nvme slots?
<adisbladis>
sphalerite_: It's an excellent way of not having to remember to clean up temp stuff. Like checking out a source tree to make a single patch
<adisbladis>
hyper_ch2: 32G
<hyper_ch2>
adisbladis: your notebook has more ram than mine :)
<sphalerite_>
adisbladis: I just use /tmp for that, and have boot.cleanTmpDir set
thc202 has joined #nixos
<hyper_ch2>
I still wonder, all distros I've used so far to clean /tmp upon reboot (or power off... not quite sure when exactely)... only nixos doesn't clean it by default
<hyper_ch2>
what was the decision for that behaviour?
<adisbladis>
sphalerite_: My /tmp is on tmpfs so :>
<adisbladis>
hyper_ch2: Good question.
<hyper_ch2>
it just something that strikes me as odd... there's probably a good reason for that default behaviour - I just don't see it
<sphalerite_>
Guanin: java stuff in nixpkgs is nasty and hard AFAIK. Maybe not the best target for a first PR :/
<hyper_ch2>
sphalerite_: what's your phone number? :)
sigmundv has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 3 new commits to master: https://git.io/vAOmC
<NixOS_GitHub>
nixpkgs/master 47adc4e Peter Simons: LTS Haskell 10.5
<NixOS_GitHub>
nixpkgs/master be67670 Peter Simons: hackage-packages.nix: automatic Haskell package set update...
<NixOS_GitHub>
nixpkgs/master 4b39930 Peter Simons: Update hspec family packages to 2.4.8 when building with GHC 8.4.x.
NixOS_GitHub has left #nixos [#nixos]
<Guanin>
sphalerite_, I know that java is (by itself) pretty ugly to handle, but I want to use that software on my server :)
<sphalerite_>
hyper_ch2: secret :p
<Havvy>
sphalerite: Is it 128-230-8028?
<sphalerite_>
Guanin: probably easier to just use nix-shell to get the build tools and build it without nix unfortunately. Although maybe we have some java wizards here who can help?
<hyper_ch2>
it's in unstable-small and possbily in different channels as well
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] samdroid-apps opened pull request #34919: WIP/requesting-comment: add plotinus and module (master...wip-plotinus) https://git.io/vAOYr
NixOS_GitHub has left #nixos [#nixos]
<sphalerite_>
hyper_ch2: since it was added last APril it should be in all the maintained channels
<hyper_ch2>
I'd actually might test rss2email
wip_ has joined #nixos
rauno has quit [Ping timeout: 268 seconds]
<hyper_ch2>
instead of running a seperate rss reader, I can just create a new mail account and use sieve to filter things into subfolders/categories
<hyper_ch2>
that's something for the weekend to do :)
<sphalerite_>
let me know how that goes!
ottidmes has joined #nixos
<hyper_ch2>
will do so
MP2E has quit [Remote host closed the connection]
spear2 has quit [Ping timeout: 268 seconds]
rauno has joined #nixos
oida has quit [Ping timeout: 256 seconds]
<hyper_ch2>
there's one caveat though... I like for xkcd and oots and some other feeds the full url to be loaded, not just the rss entry
<sphalerite>
Ooooh oots, haven't read that in ages either... That'll probably be a lot to catch up on, especially since I don't remember where I left off
<hyper_ch2>
sphalerite: I'm surprised you know oots
<hyper_ch2>
sphalerite: if in doubt, just start over with #1
<hyper_ch2>
sphalerite: *spoiler*
Neo-- has joined #nixos
<hyper_ch2>
nah, won't spoil
i-am-the-slime has quit [Ping timeout: 268 seconds]
<sphalerite>
Yeah... That will take a while xD
<sphalerite>
Probably best to do it when I'm done with uni
<hyper_ch2>
when you're done with uni you won't have time
<hyper_ch2>
because you'll have to work then
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] peti pushed 1 new commit to master: https://git.io/vAO30
<NixOS_GitHub>
nixpkgs/master afb83e0 Peter Simons: multi-ghc-travis: update to current Git master
<WilliamHamilton>
hi, anyone has an idea about why the command `pdflatex somefile.tex` complains about the lack of ifxetex.sty, even if I think it's installed? I could not find anything in the nixpkgs issue tracker
<WilliamHamilton>
here is the relevant part of my nixos configuration:
<adfaure>
Hello, is someone familliar with the buildRustPackage please. The registry doest not find one of my dep, I have to use the unstable channel. It is not something that I want to do. Do you know if I have another solution please? I hope my question is clear. Thx in advance.
<michalrus>
sphalerite: ooooh, thank you! I’ll check that out immediately.
<cmacrae>
Hey guys - is 'nix-prefetch-git' available in any package in any channels? I found it in the git tree under 'fetchgit' but I can't see that package in any channels
Neo-- has quit [Ping timeout: 260 seconds]
zzamboni has quit [Quit: Leaving.]
<cmacrae>
I want to be able to arbitrarily run that script to get the hash of remote repos
zzamboni has joined #nixos
cmacrae has quit [Read error: Connection reset by peer]
cmacrae has joined #nixos
<michalrus>
sphalerite_: oh. :( Look:
<michalrus>
gc-keep-derivations
<michalrus>
If true (default), the garbage collector will keep the
<michalrus>
It’s true by default…
ThatDocsLady has joined #nixos
<hyper_ch2>
cmacrae: just add nix-prefetch-git to your list of installe systempackages
<hyper_ch2>
or nix-env -iA nixos.nix-prefetch-git
romildo has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nix/master 081f14a Shea Levy: Allow using RegisterPrimop to define constants....
<NixOS_GitHub>
[nix] edolstra pushed 2 new commits to master: https://git.io/vAO8s
<NixOS_GitHub>
[nixpkgs] andir pushed 4 new commits to release-17.09: https://git.io/vAOB2
<NixOS_GitHub>
nixpkgs/release-17.09 8d9dda8 Vladimír Čunát: qpdf: nitpicks after update...
<NixOS_GitHub>
nixpkgs/release-17.09 6060847 Andreas Rammhold: qpdf: 6.0.0 -> 7.0.0 (fixes several CVEs)...
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/release-17.09 bc62cd9 Andreas Rammhold: qpdf: 7.0.0 -> 7.1.1...
zzamboni has quit [Client Quit]
zzamboni has joined #nixos
cbarrett has quit []
dgpratt has quit []
dgpratt has joined #nixos
cbarrett has joined #nixos
zzamboni has quit [Client Quit]
zzamboni has joined #nixos
NixOS_GitHub has joined #nixos
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] andir closed pull request #34676: diffpdf: move to using qt5 (master...feature/diffpdf-qt5) https://git.io/vAvfD
kelleyNif has joined #nixos
<__monty__>
Have a remote buildserver setup and it works great but transfering 100s of MBs of missing derivations to the server is quite annoying, especially since there's also a binary cache running on this server. How do I avoid this step, just having the remote buildserver get everything from the cache just like how my local machine gets all these things from the cache?
<dtz>
sphalerite_: $#!%% okay. thank you! Dumb q maybe, but do you have sandbox enabled?
velovix has quit [Remote host closed the connection]
velovix has joined #nixos
<sphalerite_>
I do not! Trying again with it
<sphalerite_>
also could you have a look at the PR above? Since you're the only one who reacted to the issue it fixes :p
tmaekawa has joined #nixos
<sphalerite_>
I keep forgetting that the sandbox isn't enabled by default :/
<sphalerite_>
I did get error: derivation '/nix/store/1w1sqwcawm97cwpj5nfqkn9d9q5awsh7-xz-5.2.3.drv' may not be deterministic: output '/nix/store/khh0hwhrzpmjb5fijxrcm0s1iapkl9jg-xz-5.2.3-bin' differs
<sphalerite_>
but it built alright
<dtz>
\o/ that's progress. If you look at the log, during configurePhase, what's on the line starting with "checking for a shell that conforms to POSIX..."
<dtz>
?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] fpletz pushed 2 new commits to master: https://git.io/vAO29
<Myrl-saki>
symphorien: It uses cmake or something. I have no idea how to do this.
zzamboni has quit [Quit: Leaving.]
<symphorien>
I mean : nix-locate libclang.so
<Myrl-saki>
IS there a nix-shell-like-thing that does ${} subsitition?
<Myrl-saki>
OH
<dtz>
pallav: do you mean nix copy only downloads packages one-by-one when using your cache, or that populating the cache is terribly slow because it does it one at a time?
<pallav>
dtz: Populating the cache for the mirror is slow. I am assuming its because its downloading one at a time (from cache.nixos.org)
<Myrl-saki>
But ${} is being used with the shell. How do I make nix intercept that?
<symphorien>
Myrl-saki: install nixpks.nix-index to unlock it
<dtz>
IIRC I had similar concerns when populating a cache, ended up creating local binary cache and sync'ing it with some S3 tool I found that ran in parallel .. and crashed periodically, lol :(
<manveru>
Myrl-saki: you can use ${nix-build --no-out-link -A llvmPackages.clang-unwrapped}
<Myrl-saki>
manveru: True.
<Myrl-saki>
Thanks.
<dtz>
so it's likely cmake is just being bad at finding libclang
<dtz>
llvm's cmake bits are tricky and often misused
<dtz>
and our splitting into multiple outputs doesn't help much
<pallav>
dtz: Even with an s3 tool, how do you know what you need to download (what urls need to be synced). Is there a way to recursively clone the s3 bucket?
oida has quit [Ping timeout: 265 seconds]
<Myrl-saki>
Wtf
<Myrl-saki>
Oh, -A, not -p
<Myrl-saki>
echo $(nix-build --no-out-link -A hello)
<sphalerite_>
pallav: what about making it a caching http proxy?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] shlevy opened pull request #1863: Allow includes from nix.conf (master...conf-includes) https://git.io/vAOw2
NixOS_GitHub has left #nixos [#nixos]
xcmw has joined #nixos
<genesis>
i stuck with nix-shell, if i try : nix-shell '<nixpkgs>' -A dosbox for example, it's working , but same with my own derivation fail
<pallav>
spharelite_: I am setting this up for our institute mirror server. The traditional way the rest of the mirrors are maintained is using rsync, so I hadn't thought of this. I will consider this if I am unable to find a better way.
<TonyTheLion>
can you build for ARM on hydra?
<pallav>
The number of users is fairly low, caching proxy would still be slow the first time a package is downloaded by a user, so this is not my primary preference
<genesis>
(i've only my git nixpkgs , no channel on this profile)
vaninwagen has quit [Ping timeout: 240 seconds]
<sphalerite_>
TonyTheLion: Hydra produces builds for aarch64. armv7 is a different story
<genesis>
seems i've to provide a -I option :)
<sphalerite_>
pallav: yeah… If you really want a complete mirror you will need a *lot* of storage since the official binary cache never gets GCd
<Myrl-saki>
symphorien: Sorry, turns out that the build script I was using required patching of llvm.
<samueldr>
last known estimate was 60TB
<sphalerite_>
pallav: there's a project to get nix working with IPFS which seems to be everyone's ideal solution to the problem, but I'm not sure what the state of that is
raynold has quit [Quit: Connection closed for inactivity]
<TonyTheLion>
sphalerite_: it appears I might have needed ArmV7. Guessing its not been implemented?
<sphalerite_>
genesis: you can also just do nix-shell ~/nixpkgs -A foo, or if it's your cwd nix-shell -A foo
<sphalerite_>
TonyTheLion: it does work, there's just no binary cache
<sphalerite_>
TonyTheLion: well, there are actually some unofficial binary caches
<pallav>
spharelite_: Can't the mirror be of the packages required by the latest stable and the latest unstable? That should cover 90% of all cases.
<TonyTheLion>
oh that might not be such a big issue
<NixOS_GitHub>
[nixpkgs] abbradar opened pull request #34930: Fixes for extra flags handling in nixos-install (master...nixos-install-caches) https://git.io/vAOoY
NixOS_GitHub has left #nixos [#nixos]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] FlorentBecker opened pull request #34931: Ocaml Janestreet Packages: fix installation path of stub dlls. (master...fixjanestreetstubs) https://git.io/vAOo3
NixOS_GitHub has left #nixos [#nixos]
bennofs1 has joined #nixos
xcmw has quit [Ping timeout: 268 seconds]
mrkgnao has joined #nixos
Drakonis has joined #nixos
coot has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] binarin opened pull request #34932: zsh-autoenv: init at 2017-12-16 (master...zsh-autoenv) https://git.io/vAOKP
NixOS_GitHub has left #nixos [#nixos]
oida has joined #nixos
romildo has quit [Quit: Leaving]
<lewo>
grahamc: are you here?
<gchristensen>
I am
<gchristensen>
lewo: grahamc is my alter-ego. how can I help?
<lewo>
grahamc: yes :)
<lewo>
grahamc: I've tryed to call the bot, but nohting is happening
<gchristensen>
hmm
<gchristensen>
link?
<gchristensen>
(also, please message me at gchristensen, every time grahamc is pinged it buzzes the phone in my pocket :P )
<gchristensen>
I forgot to deploy your addition at first D: sorry
mkoenig has quit [Remote host closed the connection]
tertleeltret has quit [Quit: Connection closed for inactivity]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 5371811 Benjamin Hipple: grpc: 1.8.3 -> 1.9.1...
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
[nixpkgs] shlevy pushed 2 new commits to master: https://git.io/vAOiU
<NixOS_GitHub>
nixpkgs/master d6023f0 Shea Levy: Merge branch 'grpc-upgrade' of git://github.com/bhipple/nixpkgs
Ross has quit [Ping timeout: 248 seconds]
zzamboni has quit [Quit: Leaving.]
mkoenig has joined #nixos
<joepie91>
gchristensen: never publicly disclose your notification triggers! :P
zzamboni has joined #nixos
<Rovanion>
So I installed xarchiver, by adding it to systemPackages, but its unable to open zip files due to it not finding zipinfo. So my question is: Is it so that packages in nixos has optional dependencies like on debian systems and I need to do something special in order to install them?
<lewo>
gchristensen: no problem. To be hoonest, I tryied hard to wirte @grahamcofborg case insensitive and github renders it camelcase :/
<joepie91>
Rovanion: nope, it should either include the dependency by default or let you enable it as a flag to the package, and the xarchiver package does neither, so it seems it's just a bug :)
<joepie91>
Rovanion: best file a bug on nixpkgs about it
pkill9 has joined #nixos
<gchristensen>
that isn't a problem, it is case-insensitive. if you manage to type it @GrAhAmCoFbOrG it'd be fine
<sphalerite_>
is there an easy way to add an extra symlink to the system profile dir?
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
xcmw has joined #nixos
freeman42x]NixOS has quit [Remote host closed the connection]
__Sander__ has joined #nixos
freeman42x]NixOS has joined #nixos
freeman42x]NixOS has quit [Remote host closed the connection]
<genesis>
calling buildPhase in the nix-shell should have the same output than in nix-env -i ? or i've to source some makeFlags and so ?
<srcCoon>
Is there a difference between 'default' and 'system' profiles? Somehow my root's '.nix-profile' is pointing at the non-existent 'default', while 'system' ones seem to be in perfect synchrony with the system.
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dezgeg pushed 1 new commit to master: https://git.io/vAO1S
<NixOS_GitHub>
nixpkgs/master 68052b5 Tuomas Tynkkynen: python: Pull ensureNewerSourcesHook call to all-packages.nix...
NixOS_GitHub has left #nixos [#nixos]
codygman has quit [Ping timeout: 260 seconds]
sigmundv has quit [Ping timeout: 276 seconds]
ssmike has joined #nixos
<infinisil>
srcCoon: that might be the case for users that never ran nix-env
<__monty__>
Does nix-env evaluate the entire nixpkgs when you do nix-env -u?
<sphalerite_>
genesis: I think you actually need to do "eval buildPhase"
<sphalerite_>
__monty__: yes, in the same way that nix-env -i without -A does
<__monty__>
sphalerite_: Is there a way round this similar to -iA?
ssmike has quit [Ping timeout: 256 seconds]
<__monty__>
Ah, thanks.
<__monty__>
Hmm, so no simple flag as a workaround?
<genesis>
sphalerite_ it seems produce different than buildPhase, that's a bit tricky
<sphalerite_>
unfortunately not
<genesis>
gnu/stubs-32.h : No such file or directory # include <gnu/stubs-32.h>
adfaure has joined #nixos
<genesis>
i don't have that using nix-env hum
<adfaure>
Hello, I want to use sed in a package to replace a bash variable formatted as ${var}. but nix try to interpret it; so I use the syntax $${var} to tell nix to escape it... hence sed does not find it anymore. Any tip please ?
<sphalerite_>
adfaure: how are you quoting the nix string?
<srcCoon>
infinisil: i was just wondering what's the difference between them conceptually. I'm guessing, different use-cases but the rest eludes me.
kiloreux has quit [Remote host closed the connection]
<__monty__>
adfaure: Do you need sed rather than just "substituteInPlace "string" --replace "string""?
<adfaure>
sphalerite_: Sorry I dont understand your question,
<sphalerite_>
adfaure: if it's double quotes, like "foo", you'll want to use \$. If it's doubled single quotes you'll want to use the really ugly ''$
<__monty__>
Not sure how you'd do the equivalent of your find.
<adfaure>
Thank you.
<adfaure>
It should do with an ls
<__monty__>
I think you could use builtins.readDir
<LnL>
dezgeg: that diff doesn’t look right
zzamboni has joined #nixos
<__monty__>
I'd still consider only changing the definition of SPARK_HOME rather than changing it in place everywhere unless there's a reason. The less you patch the easier it'll be to understand later.
johs has quit []
johs has joined #nixos
<Dezgeg>
yeah I may have messed up the conflict resolution
<Dezgeg>
but it did build
tmaekawa has quit [Quit: tmaekawa]
<Dezgeg>
note that there was a package upgrade in master that presumably made it build with LLVM 5 now
<adfaure>
__monty__ : I need to patch only ${spark_home}/sbin otherwise it wont work (however I am not sure it will anyway)
ssmike has joined #nixos
<__monty__>
Oh ok, so you do have a reason, then nvm my comments.
<gchristensen>
omg mrkgnao
<hyper_ch2>
anyone knows of a nice small notebook for linux iwth 2x nvme slots?
<adfaure>
Ok, thank you for the help :)
<mrkgnao>
gchristensen: o/
pietranera has joined #nixos
<gchristensen>
I can't top that
<mrkgnao>
so Konsole gives me notifications for IRC pings *and* renders ligatures? sweet
<mrkgnao>
well, if you say so
<mrkgnao>
:)
<mrkgnao>
say, I have a NixOS problem, you free?
<gchristensen>
sure
<pietranera>
Hello, I am on NixOS and have the unstable channel installed on my normal user, how can I ensure that that channel is picked up by nix-shell? (it seems that the root channel is used instead).
<mrkgnao>
so a bunch of operations are failing with "imported archive of /nix/store/$LONG_PATH lacks a signature"
<pietranera>
(specifically I am trying to use Python with pandas 0.22 but I end up with pandas 0.20)
<mrkgnao>
I'm on unstable
<mrkgnao>
likely relevant: when I switched to unstable, I also added a local channel for home-manager purposes
<gchristensen>
do you have any extra binary caches or remote builders defined?
<mrkgnao>
and there's a name collision there
<mrkgnao>
no
<mrkgnao>
just the cache.nixos.org one
<mrkgnao>
so, e.g. if I do `home-manager switch` the "lacks a..." problem comes up
<mrkgnao>
and doing nix-env -iA foo gives me "warning: name collision in input Nix expressions..."
<gchristensen>
the collision is probably not relevant to the unsigned path one. is the $LONG_PATH constant or variable?
<mrkgnao>
I tried renaming the local channel, but IIRC it broke nix-env, nix-shell etc and I had to revert
<LnL>
that means you have both a system and user nixpkgs channel, but that's not related to the signature problem
ssmike has quit [Ping timeout: 256 seconds]
<srhb>
pietranera: nix-shell uses whichever nixpkgs is on the NIX_PATH, so you'll need to feed in a different version explicitly when using -p
<mrkgnao>
gchristensen: it's the path to the nixpkgs version that home-manager uses
<mrkgnao>
sorry, that's wrong
<mrkgnao>
it's the path to the home-manager .tar.gz, which is fixed in ~/.config/nixpkgs/home.nix
<pietranera>
srhb I tried to override nix-shell with "-I nixpkgs=/path/to/unstable-channel", shouldn't that work too? I have the package definition in a shell.nix file because I need to access the environment.
<LnL>
do the paths it's complaining about exist or not?
<srhb>
pietranera: Yes, that should work
<pietranera>
OK, I am clearly doing something wrong then because I end up with the wrong version of pandas :(
<mrkgnao>
they don't
<pietranera>
I'm going to give it another go.
<srhb>
pietranera: Perhaps you can tell us exactly what you did. :)
<LnL>
then nix is trying to substitute them from somewhere
<LnL>
do you have any caches / builders or special NIX_PATH configured?
<gchristensen>
mrkgnao: do you have a channel defined for home-manager?
<mrkgnao>
LnL: I ran `home-manager switch --option ..." and there's no change, not sure if it's picking up the switch
<mrkgnao>
gchristensen: as far as I understand, no
erasmas has joined #nixos
<mrkgnao>
LnL: I edited the home-manager source to always add that switch to nix-build, but it didn't change anything
<LnL>
I don't think that would be applied until the next switch
<mrkgnao>
I installed it with nix-shell /path/to/hm -A install
<mrkgnao>
and then tried rerunning `home-manager switch`
<srhb>
pietranera: You can test that it works easily with something like: nix-shell -I nixpkgs=https://github.com/NixOS/nixpkgs-channels/archive/nixos-unstable.tar.gz -p 'python.withPackages (ps: [ ps.pandas ])'
<adfaure>
re __monty__ I am sorry, but somehow I dont understand the behaviour of substituteInPlace, I do : substituteInPlace $out/sbin/start-master.sh --replace "''${SPARK_HOME}/sbin" "$out/sbin"
<adfaure>
And as a result I get: "${SPARK_HOME}/nix/store/6a65wkiaipfldff5kpss80pyj58pycfm-sparkSt/sbin/spark-config.sh"
mgttlinger has quit []
mgttlinger has joined #nixos
mgttlinger has quit [Max SendQ exceeded]
<infinisil>
gchristensen: you motivated me to get irc push notifications, got it working \o/
mgttlinger has joined #nixos
mgttlinger has quit [Max SendQ exceeded]
<gchristensen>
nice!
mgttlinger has joined #nixos
<__monty__>
adfaure: I think your escape's just wrong "\${..." or drop the quotes entirely and use '' instead of \
<adfaure>
__monty__ I already use '' :)
<srhb>
It just needs _more_ escaping.
<__monty__>
adfaure: '' only works in a ''string''.
<srhb>
once for nix, once for bash.
<srhb>
:-)
<__monty__>
At least that's what I thought. Listen to srhb over me though.
<adfaure>
I use this: --replace "''${SPARK_HOME}/sbin" "$out/sbin"
<srhb>
I was just thinking that if it's a multiline string that's interpreted by a shell, and you want the literal string ${foo}, you need to not have nix interpret it, so that's ''${foo}, and then you need to have bash not interpret it, so that's \''${foo} -- right?
<srhb>
(I may be wrong, escaping always trips me up)
<srhb>
Better yet, use single quotes. :-)
<gchristensen>
mrkgnao: I have no idea ... it sounds like LnL has a better idea? :$
<adfaure>
srhb: And just like that ... it worked. I still magic to me, but I'll take it.
<mrkgnao>
is there any way I can trace this?
<mrkgnao>
more verbosity, as it were?
<adfaure>
srhb: __monty__ : thanks
pie__ has joined #nixos
<gchristensen>
rycee: you around?
eacameron has quit []
<LnL>
yeah, can you run the nix-build command that home-manager would run with -vvvvvv
pie_ has quit [Remote host closed the connection]
<LnL>
mrkgnao: ^
<mrkgnao>
one moment
eacameron has joined #nixos
<infinisil>
mrkgnao: your problem is that home-manager isn't picking up changes to home.nix?
<mrkgnao>
infinisil: it is, but it's failing saying that the .tar.gz doesn't have a signature
<LnL>
oh!
<LnL>
the tar.gz not other paths?
<mrkgnao>
yes, just that tar.gz, afaict
<__monty__>
srhb: Wait, where does the shell come into it? Aren't the phases just evaluated by nix?
<srhb>
__monty__: No, that'll be bash :)
<mrkgnao>
okay, so with -vvvvvv I didn't get anything interesting, just a bunch of extra curl logging
<mrkgnao>
LnL: ^
<srhb>
__monty__: substituteInPlace is just a shell function
<srhb>
__monty__: You have be gifted with a "how the hell does this work so well?!" feeling.
<srhb>
been*
szicari has joined #nixos
ssmike has joined #nixos
betaboon has joined #nixos
<mrkgnao>
srhb: alternatively, "I'm terrified it'll suddenly stop working one day and I'll be none the wiser"
<srhb>
:-)
cement has joined #nixos
<srhb>
It's pretty impressive.
leat has quit [Ping timeout: 256 seconds]
<mrkgnao>
trve
<mrkgnao>
true*
<mrkgnao>
god, those black metal influences never leave do they
<srhb>
<3
betaboon has quit [Read error: Connection reset by peer]
<srhb>
Okay, the single quote idea was terrible though.
<srhb>
It _should_ be '''${foo}' for the literal string '${foo} I think, but...
<srhb>
literal '${foo}' even
<srhb>
Is there a bug in the parser or am I misunderstanding?
ssmike has quit [Ping timeout: 260 seconds]
<samueldr>
''', the first two escapes to ' since it's not followed by $ afaik
<srhb>
Glorious
<__monty__>
Are single quoted strings even allowed?
<mrkgnao>
can anyone help me with the channel name collision? I have a workaround for home-manager now, but the other one's worrying
<samueldr>
there may be a better suited escape than '' ${"'"}''${foo}' ''
<srhb>
__monty__: That was to avoid bash interpreting it.
<mrkgnao>
just go full zalgo
<srhb>
samueldr: Ow :D
<samueldr>
(I'd like someone that actually knows nix to answer you though)
<srhb>
I think the answer is "what the hell are you doing the rules weren't designed for this"
<samueldr>
(I looked at the output results in nix-repl to see what was going on)
<__monty__>
mrkgnao: How about specifying NIX_PATH?
<samueldr>
yeah, escaping is hell in most situations :)
<mrkgnao>
while that would probably work, I'd just like to rename the channels so that they aren't ambiguous any more
<srhb>
samueldr: It's easy to produce a string with two single quotes in front though
<__monty__>
mrkgnao: And why not do that through NIX_PATH?
<srhb>
'''''''${foo}' ''
<srhb>
Glorious
<srhb>
Can't get one though. :-)
<__monty__>
srhb: I'm lost are we talking in the context of a multiline string?
<mrkgnao>
LnL: gchristensen: I also get "lacks a signature" failures on doing nix-channel --update
<Myrl-saki>
What day is 18.03?
<srhb>
__monty__: that and bash. The whole point was to avoid 1) nix interpretation and 2) bash interpretation. bash will interpret "${foo}" but no '${foo}'
<Myrl-saki>
Or are stable releases always on the first day of the month?
<srhb>
__monty__: But how do you produce '${foo}' in a multiline nix string?
<gchristensen>
Myrl-saki: probably at the very very end of March
<Myrl-saki>
gchristensen: Oh.
<Myrl-saki>
gchristensen: Thanks.
<gchristensen>
yeah
<__monty__>
Is it not '''${foo}'''?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] andir pushed 5 new commits to master: https://git.io/vAOdV
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master f80b233 Andreas Rammhold: libyaml-cpp: 0.5.3 -> 0.6.1
<NixOS_GitHub>
nixpkgs/master 85d0eb7 Andreas Rammhold: interception-tools: removed custom libyamlcppWithoutBoost...
<NixOS_GitHub>
nixpkgs/master 77dba48 Andreas Rammhold: openxcom: 1.0.0 -> 1.0.0.2018.01.28...
<infinisil>
I should make a guide on escaping strings in nix
<infinisil>
or how about a wiki page
<srhb>
__monty__: Nope! Two quotes to escape ${
<__monty__>
Is it not '''''${foo}'''?
<srhb>
__monty__: Nope, that's a syntax error!
<srhb>
Oh wait
<srhb>
I miscounted.
<srhb>
That's, uh...
<clever>
!tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
<srhb>
That's a syntax error!
<srhb>
>_>
<srhb>
(I had to try that one)
<samueldr>
sequences of single-quotes are... weird... in how they're interpreted by nix
<samueldr>
though, it all makes sense in a parser POV
<__monty__>
How does this make sense from the parser's pov?
<gchristensen>
clever: tofu-ing to get the 0s? :)
<clever>
gchristensen: yeah, lol
<LnL>
rofl
<gchristensen>
I PM {^_^} all the time to get the 0's haha
<Myrl-saki>
clever: That doesn't work for a huge download, right?
<srhb>
clever: If you practice hard you can hold the zero key for exactly the right amount of time.
<clever>
wasnt sure it accepted it in a PM
<clever>
Myrl-saki: it would result it in downloading twice
<srhb>
Bonus: It works for other keys too!
<Myrl-saki>
clever: RIght.
<clever>
srhb: i have done that when holding backspace to delete large blocks, lol
<srhb>
:D
<Myrl-saki>
I use nix-prefetch-url, but I have no idea what the other prefetches are caled.
<srhb>
nix-prefetch-git
<LnL>
I just remember it's 52 characters and use vim
<srhb>
There are no others...
<srhb>
:-)
<srhb>
LnL: Well, that's just not very creative.
<clever>
srhb: and i have messed with the repeat rate to cheat at some web-games, and the shell is practically unusable if you set the repeat rate to max, lol
<Myrl-saki>
srhb: I don't know what package has that though. ; n ;
<srhb>
clever: This is true :-P
<clever>
srhb: let me see...
<Myrl-saki>
Oh, it's literally called nix-prefetch-git
<samueldr>
depending on the amount of single quotes, it can treat them as escapes for double-quotes, or escapes for double-quotes with a finished string, starting next an unfinished string
pie__ has quit [Ping timeout: 260 seconds]
<samueldr>
escapes for double-singlequotes ** (confusing myself here)
<srhb>
samueldr: I think this means we cannot possibly do it.
<clever>
srhb: defaults: auto repeat delay: 500 repeat rate: 20
xcmw has joined #nixos
<samueldr>
yeah, why ${"''''"} for single quotes lenghts :(
<srhb>
samueldr: We can get n*2 single quotes before a ${ for any n though.
<tnks>
I'm having a hard time finding with Google the difference between trusted-binary-caches and extra-binary-caches.
ssmike has quit [Ping timeout: 276 seconds]
asuryawanshi has joined #nixos
<kakuzade>
Hi. I need to place a script.sh in the /etc/profile.d folder. How do I do this in NixOS? Can it be done with an expression in /etc/nixos/configuration.nix?
kim0 has quit [Quit: Connection closed for inactivity]
<NixOS_GitHub>
nix/master b8739f2 Shea Levy: Enable specifying directories in plugin-files.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] nlewo opened pull request #34939: dockerTools.buildImage: do not add /nix/store in the tar stream (master...fix-34137) https://git.io/vA3eR
<clever>
gchristensen: any ETA on nix 1.13 in nixpkgs?, i'm looking to use builtins.fetchTarball with a sha256
<LnL>
1.13?
<gchristensen>
2.0?
<clever>
wait, i think i misread something, let me test some more...
Itkovian has joined #nixos
<clever>
gchristensen: ah, i think i see the problem
Arcaelyx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
gchristensen: builtins.compareVersions says that 1.12pre5873_b76e282d is older then "1.12", so this nix expression doesnt try to use builtins.fetchTarball with a sha256
<dtz>
clever: IIRC Nix 2 is slated for 18.03, at least optimistically
<clever>
gchristensen: and if i just comment out the check, the 1.11 nix-shell fails, before the 1.12pre nix is ran
<clever>
and that led to my confusion
<dtz>
haha d'oh
<gchristensen>
oops
<clever>
and its the host nix 1.11 that has to eval the expressions to provide nixUnstable
<clever>
so there is no way to hard-code around compareVersions being dumb
<chreekat>
Looks like I can't pass a 'dontPatchELF' attr to the haskell-specific mkDerivation (haskell-modules/generic-builder complains "unexpected argument"). Is that intended? What if I really *want* dontPatchELF?
<LnL>
yeah, but only if your client matches the daemon exactly
<clever>
chreekat: use .overrideAttrs to insert it after the haskell wrapper has finished its job
<clever>
LnL: i dont think that counts when using hydra-eval-jobs, which sets restricted mode
<chreekat>
clever: thx. I assume I add that to the result of the mkDerivation call?
<clever>
LnL: the builtins.storePath is rejecting paths that you just pulled out of nowhere, because you didnt purely define how they got made
adfaure has quit [Ping timeout: 260 seconds]
ambro718 has joined #nixos
<clever>
chreekat: the entire mkDerivation call returns something that is both an attrset, and a derivation, and you can index into the .overrideAttrs attribute to override things
<NixOS_GitHub>
[nixpkgs] timokau opened pull request #34940: sage: Don't build docs by default (master...sage-no-docs) https://git.io/vA3Jh
NixOS_GitHub has left #nixos [#nixos]
<chreekat>
clever: thx again. I found the docs for overrideAttrs, too, so I should be set. :)
<ambro718>
I am having a problem with Chromium with nixos-unstable: moving the mouse and scrolling within Chromium causes interminnent hangs (tens of milliseconds long), seemingly system-wide as the hangs cause: mouse to stop moving then jump, sound from video in another tab to skip, and also sound from video playing in Firefox to skip.
<clever>
LnL: ok, so builtins.fetchurl { url=; sha256=; } fails with error: imported archive of ‘/nix/store/jcmcikcs1f69aj6d8415blcjpdlz6f4d-5c09cdc187b014143302551e62d4973b5bf52814.tar.gz’ lacks a signature
<clever>
but import <nix/fetchurl.nix> under 1.11 can pass, and produce the exact same storepath
<clever>
and if that path exists, builtins.fetchurl { url=; sha256=; } doesnt fail!
teej has joined #nixos
ssmike has quit [Remote host closed the connection]
ssmike has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] shlevy pushed 1 new commit to master: https://git.io/vA3I0
<NixOS_GitHub>
[nixpkgs] nlewo pushed 2 new commits to master: https://git.io/vA33X
<NixOS_GitHub>
nixpkgs/master 7f454c0 Vincent Demeester: gomodifytags: init at 2017-12-14...
<NixOS_GitHub>
nixpkgs/master 596a03e lewo: Merge pull request #33429 from vdemeester/init-gomodifytags...
NixOS_GitHub has left #nixos [#nixos]
kelleyNif has quit [Quit: Leaving.]
mkoenig has joined #nixos
<eacameron>
On my new install of nixos for some reason vim doesn't accept the `End` key or the `Delete` key. It either does nothing or does something else.
<romildo>
clever, eacameron, the version is explicitly split in major version and minor version, and the major version alone is used as a subdirectory in the url.
<eacameron>
romildo: I see. As long as the versions don't do anything wonky you can use `lib.concatStringsSep "." (lib.take 2 (lib.splitString "." ver))`
<romildo>
clever, eacameron, update-source-version does not seem to support that style of setting the version. So I am thinking about defining only version, without a explicit split and major and minor version, and determine the major version with a nix subexpression in the url.
<Dezgeg>
maybe " builtins.match '^([0-9]+)\.([0-9]+)' ver " would also work
Lisanna has joined #nixos
ma271 has joined #nixos
<romildo>
Would it be a nice idea to have builtin functions for splitting version strings in major and minor? That would make the nix expression for the derivations less clutter.
<eacameron>
romildo: Trouble is the definition of "major" and "minor" is not the same for all versioning schemes
<Dezgeg>
well I guess a problem is that I don't think there is a good convention whether 'major' means in say, 1.2.3; is it '1' or '1.2'
stephenjudkins has quit [Remote host closed the connection]
<rycee>
gchristensen: I am now :-)
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
Dezgeg: i generally think of 1.2.3 as major version 1, minor version 2, patch version 3
<Dezgeg>
yeah that's the most common I think
<clever>
where 1 is incremented for massive changes, 2 more for abi level changes, and 3 for minor things where you can mix the binaries between versions
<Dezgeg>
that's semver actually, I think
<clever>
not that nix allows such mixing, lol
<Dezgeg>
or not quite
<rycee>
adisbladis: Yeah, it does seem likely that they aren't running tests. Perhaps nobody except we are? :-)
xcmw has joined #nixos
<romildo>
My second option would be to define this function locally. Would it be better to define it in pkgs/desktop/mate/default.nix, and inherit it in every derivation underneath mate, or to define it in each derivation. Or even not defining it and calculating the major version directly in the url in each derivation. What do you think?
<michalrus>
Hey, so… when I do `nixos-rebuild vm` and in that QEMU try to run `nix-shell -p htop`, it fails with EACCES. Is this expected? :o
<mpickering>
Are there any tools which will generate "shell.nix" files with the current version of nixpkgs pinned?
<mpickering>
I feel like there should be a very convenient way to pin nixpkgs as your build breaking because you update nixpkgs is one of the worst things about nix imo
<mpickering>
it should also be way more convenient to "re-pin" a project if you want to upgrade
<michalrus>
s/EACCES/EPERM huh
roberth has joined #nixos
<srhb>
mpickering: fwiw I find pkgs ? import (fetchTarball ...) {} to be pretty ergonomic.
<rycee>
mrkgnao: Hmm, I've never seen the "imported archive of …" error before. For Home Manager you can use `-I nixpkgs=…` to point out a nixpkgs that you'd like to use. I've never actually used a per-user channel for Home Manager.
<mpickering>
srhb: I'm sorry but I don't think it is at all. You have to know the special "magic" to use fetchTarball or import <nixpkgs> and use fetchGithub and so on. Then you have to find the right commit hash you want or the correct source location and then you need to nix-prefetch xxx to get the right sha
<LnL>
pretty sure I know what's going on now, but he's not here anymore
<mpickering>
This is in the realm as "undefined variable" if you type the wrong thing for "nix-shell -p adasd"
<rycee>
mrkgnao: Actually, I don't think I use any channel anywhere. Just local checkouts or NIX_PATH having nixpkgs=http://… in it.
<clever>
mpickering: and if fetchTarball complains about signatures, its because your nix-daemon is 1.11
<clever>
rycee: matrix is doing weird things again, i dont see a mrkgnao on irc
<clever>
from my view point, your talking to somebody that doesnt exist :P
Neo-- has quit [Ping timeout: 265 seconds]
smichel17 has joined #nixos
<rycee>
clever: Hmm, maybe so. Riot is tab expanding his name so I assumed he was online. In any case, I'm online at the moment so hopefully it'll show up on botbot :-D
<srhb>
mpickering: I'm assuming your complaint isn't pinning at a specific revision (url, hash...) but rather that you'd have those incantations created for you at some arbitrary point in time when you type in the command?
<clever>
rycee: if 2 matrix users are in the same irc channel, it can show their matrix username, which differs from the irc nickname
<clever>
rycee: which then heavily confuses everybody on irc, because irc doesnt show the matrix names
humanoyd has joined #nixos
stephenjudkins has joined #nixos
pie__ has joined #nixos
<srhb>
mpickering: (I guess I'm asking what would you *like* to do)
stephenjudkins has quit [Remote host closed the connection]
<mpickering>
srhb: Yes essentially but I feel like it is such an important thing to do in projects that it should be a very very easy thing to do. With special syntax or support via config files or something to make it specifying a point in time you knew your environment worked as painless as possible.
stephenjudkins has joined #nixos
<rycee>
clever: Ya, I can imagine. He's showing up as @freenode_mrkgnao:matrix.org for me so I guess not a Matrix user, though. Maybe matrix just remembers users for some time?
<srhb>
mpickering: Right. Agreed.
<clever>
rycee: what has he said most recently?, that should persist
<mpickering>
I was hoping someone would have pointed me to the relevant issue ;) but it seems this isn't a burning issue for other users
<srhb>
Personally I like 1) not shelling out to non-nix configgy things and 2) just using nix to do the nix things. I would like autogeneration over special support though, which could be easily implemented into cabal2nix etc.
<srhb>
I also tend to just carry a nixpkgs.nix around which Does The Thing, so that's what I import instead of <nixpkgs>
<srhb>
I personally don't find it very clunky, but I totally get your viewpoint.
<srhb>
mpickering: If only dir env hooks were standardized and reliable, we could just set NIX_PATH... :-)
<rycee>
Btw, mrkgnao's config looks fine to me but I'm not sure how well Home Manager works with nixUnstable. I've never tried it…
<srhb>
rycee: It works fine.
<rycee>
srhb: Cool, good to know! :-)
<srhb>
wait what, home manager has a nixos module now :o
<mpickering>
A tool which generated your "nixpkgs.nix" thing by looking at what your "<nixpkgs>" variable referred to might be good?
<srhb>
mpickering: Yeah, that sounds like an excellent approach!
<srhb>
It's not always possible to determine the source of <nixpkgs> though
<clever>
rycee: aha, he last spoke 3-ish hours ago, and ping timeout'd about an hour ago
<rycee>
srhb: Yeah, and it even sort of works!
<srhb>
But something can probably be hacked out
<mpickering>
right...
<clever>
rycee: thats why i dont see him having spoken recently, and cant tab-complete him
<rycee>
srhb: I'm using it for declarative containers and nixops deploys.
<srhb>
rycee: Exciting. I'll have to convert mine... Soonish. Any day now.
<srhb>
Right behind the "swap out all gpg and ssh keys" project.
<srhb>
Probably.
<srhb>
:-)
<rycee>
clever: Ya, I guess matrix is slow in dropping people from the tab completion.
<mpickering>
<nixpkgs> just points to a directory right? Is this always a git repo or is it sometimes just a normal directory?
<mpickering>
how do channels work?
<srhb>
mpickering: It can be many things, including an url to a tarball.
<clever>
mpickering: nix-channel manages unpacked tarballs in /nix/store/
<srhb>
mpickering: Mine points at a normal git checkout that I manage, usually.
ssmike has joined #nixos
<rycee>
srhb: It's a bit beta at the moment and I hope to change the way packages are installed to support `nixos-rebuild build-vm`. This doesn't work right now because HM wants to change the Nix store when it does the activation.
<clever>
mpickering: it will download the latest tar over http, unpack it within /nix/store/, and your set of active channels is managed with nix-env, the same as your profile
<srhb>
rycee: Ah, right. Okay :)
jtojnar has joined #nixos
<infinisil>
srhb: the gpg + ssh keys stuff is on my list too..
<infinisil>
I should really make security higher priority
<infinisil>
Hey does anybody know what's up with this: Every time i do suspend, the screen turns off, as it should, but then after a couple seconds it turns on again, as if i pressed a key, but I didn't
fendor has joined #nixos
rihards has joined #nixos
asuryawa_ has joined #nixos
asuryawanshi has quit [Read error: Connection reset by peer]
kwork has quit [Ping timeout: 268 seconds]
<srhb>
infinisil: I've had this happen with faulty usb stuff
<srhb>
infinisil: Check dmesg
sigmundv has joined #nixos
<srhb>
infinisil: I remember having to forcibly unload some usb module to get rid of it, but it wasn't on this laptop so it's not in my config.
vidbina has joined #nixos
<srhb>
s/forcibly/manually
<infinisil>
srhb: it was because of some usb device connected or just the internal usb driver/hub thing?
<srhb>
Internal thing.
<srhb>
You can try disabling all of the ones in /proc/acpi/wakeup (make sure you have a way to input things or make the test unattended) to see if that's it
<infinisil>
i have nothing plugged in, so it's probably the same problem
<clever>
one of my older laptops has a USB smart-card reader, directly under the pcmcia slot, but it never had working linux drivers
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
ive also heard that the mouse in apple laptops is partially usb based
<infinisil>
srhb: thanks for the tips, will take a look
<infinisil>
I can't really make a whole lot of sense of the dmesg output though
<srhb>
infinisil: Sure thing, hope it works out :)
<srhb>
OK, just try disabling the wakeups
<srhb>
Easier to test that way
<srhb>
I think you just echo EHC1 > /proc/acpi/wakeup to toggle the EHC1 wakeup, etc.
* infinisil
tries
<infinisil>
disabling each one one-by-one and trying
fendor has quit [Read error: Connection reset by peer]
<infinisil>
Aha!
<infinisil>
It's the lid!
<srhb>
Well, that's rather annoying :P
<infinisil>
I've actually had problems with the lid before
<infinisil>
But that's perfect now, I don't want it to un-suspend anyways when opening the lid
<srhb>
Two birds... :)
<infinisil>
well okay maybe I do, but it's not important
ssmike has quit [Remote host closed the connection]
<infinisil>
This is the first time I'm trying out hibernation, which might be essential, because when suspended my battery seems to be drained
<infinisil>
And it works!
isidore has quit [Quit: WeeChat 2.0.1]
<srhb>
Great!
<srhb>
I don't have swap, so I guess I can't hibernate. :P
<infinisil>
I didn't really need swap before, but now I'm glad I have it
<infinisil>
Btw, there's something similar to hibernation, but on a per-application basis: https://criu.org/Main_Page
hamishmack has quit [Quit: hamishmack]
<infinisil>
I've tried to compile a kernel with the required options for it to work once, but couldn't be bothered to wait for it to finish
<clever>
srhb: when i was setting up my new laptop with 32gig of ram, i configured my justdoit.nix script to make a 40gig swap partition, just for hibernation
<clever>
srhb: i forgot, that my script works in mb
arjen-jonathan has joined #nixos
<srhb>
It's fine, you can still hibernate if you just shut down most applications! :D
<infinisil>
lol
<clever>
srhb: i'm not sure if the unpacked kernel would even fit inside 40mb
<srhb>
Indeed.
<infinisil>
needs more types
<clever>
infinisil: heh, but how do you deal with mb and gig both being int based?
<infinisil>
i swear, eventually I'll abandon bash and use a nice haskell interface for everything
<clever>
infinisil: it would need to support 40G as a valid number
<srhb>
Plenty of newtypes :)
<clever>
infinisil: i dont think even haskell can do that, enless you just do `Gig 40`
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] shlevy opened pull request #1867: Allow plugins to define new settings. (master...register-settings) https://git.io/vA30D
NixOS_GitHub has left #nixos [#nixos]
<infinisil>
clever: yeah newtypes can do that
<srhb>
clever: Usually you plumb it with some operator to do the logic you need
<srhb>
clever: So you end up with 40 # someunitmagic
spietz has quit [Ping timeout: 248 seconds]
<srhb>
See for instance the units package.
<infinisil>
they wrap a type (e.g. int) with zero cost, but can't be combined with normal int, a totally separate type
Myrl-saki has quit [Ping timeout: 276 seconds]
<clever>
infinisil: but what if i want to accept both mb and gig?, and just convert it all down to bytes
<srhb>
That's fine.
<srhb>
The internal representation can be whatever you like.
spietz has joined #nixos
<srhb>
40 # GiB would just be some sort of smart constructor that is the only way to construct a Bytes value.
<srhb>
(For instance)
<clever>
ah
<infinisil>
I actually wrote a unit library for Swift! There to get 40MB, you'd do `40 * MB` (just a multiplication), while MB is defined as 1000000 bytes, so you'll end up with 40000000 bytes
<clever>
infinisil: but what if i want 1024 based units? :D
<isHavvy>
MiB
<clever>
i can never keep those 2 straight, lol
<infinisil>
It's very flexible :)
<infinisil>
same
<clever>
i just always work in 1024, and always call it mb or MB
<srhb>
I remember it like this: Gibi sounds cooler than Giga. 1024 is righter than 1000.
* srhb
is very smarts
kiloreux_ has quit [Remote host closed the connection]
<srhb>
>_>
Myrl-saki has joined #nixos
kiloreux_ has joined #nixos
<Lisanna>
I thought Nix scanned output paths to compute runtime dependencies... but I just did a test where I echo'd a /nix/store path to the output file, and nix-store -qR isn't showing that path as a runtime dependency
<clever>
Lisanna: nix only checks for your inputs
<johnw>
srhb: or just: bi(inary) is a power of 2
<infinisil>
clever: what do you mean by "1024"? What are these last two symbols "2" and "4"?
<clever>
Lisanna: so if you hard-code a storepath into your derivation, its not an input, so nix doesnt know to look for it
<Lisanna>
clever Ok, that actually makes things a lot clearer
<srhb>
johnw: Oh I like that one!
<Lisanna>
I was wondering how nix scanned for paths since they can be different lengths...
<srhb>
johnw: ... That's why they're named like that. TIL.
<infinisil>
heh
orivej has joined #nixos
<hodapp>
are there still articles being published about how ternary/trinary is gonna take over and it's better because 3 is closer to e?
<infinisil>
srhb: follow-up question: You said you solved your suspend problem by unloading some module, would that work for me too now that i found out it's this lid thing in wakeup?
<Dezgeg>
I think nix actually just looks for the hash followed by a dash, i.e.: zzy8hrky147hdp9x994kpp8jg49mg2ad-
<srhb>
infinisil: Presumably, but I'm not sure if it's fine grained enough to not take something else with it. The righter approach would be to ensure that that wakeup is disabled.
<infinisil>
sooo, how do they get there anyways?
<srhb>
mumblemumble kernel mumble driver.
<infinisil>
if you happen to know
<srhb>
(so no, I don't) :-)
<infinisil>
heh
<infinisil>
I shall take a google
<clever>
infinisil: oh, that reminds me, my laptop goes into standby if i close the lid, and wakes up when i open the lid
<srhb>
I'd probably just throw it in a unit.
<infinisil>
srhb: I'm inclined to, but it feels so wrong
<clever>
infinisil: i fixed discovered that, when i was trying to get behind a desktop, while using a voice chat on the laptop, and the lcd panel was in the way, so i just closed it
<infinisil>
srhb: Or just activationScripts probably
<clever>
infinisil: then the voice chat went silent, lol
<srhb>
Woops :P
<infinisil>
clever: i think something is preventing mine from going to sleep when closing the lid when an external screen is plugged it
<infinisil>
in*
<clever>
infinisil: one sec
<clever>
[root@system76:~]# cat /proc/acpi/wakeup | grep -i lid
<clever>
LID0 S3*enabled platform:PNP0C0D:00
<clever>
infinisil: i can see that LID0 is an enabled wakeup source on my system
<clever>
there should be a way to disable that
Ross has quit [Ping timeout: 276 seconds]
<infinisil>
There is, srhb just showed me, LID0 > wakeup
<infinisil>
but I'd like a permanent solution
<clever>
infinisil: and that echo appears to toggle the option!
<clever>
so you cant just jam it into an activation script
<clever>
it will toggle every time you `nixos-rebuild switch`
<infinisil>
clever: oh damn
<clever>
so you need to grep, and then conditionally echo
<infinisil>
ugh
<srhb>
Isn't this sort of thing configurable via sysctl anyway?
<infinisil>
clever: I'll do that if nothing else works
<infinisil>
srhb: taking a look at sysctl -a..
<clever>
infinisil: also, id use a systemd one-shot service, rather then an activation script
Ubercow has joined #nixos
<clever>
infinisil: its far too easy to brick the entire machine with activation scripts
Ross has joined #nixos
<clever>
infinisil: a couple months ago, i was helping somebody that tried to do network in his activation script, and it intially worked, via nixos-rebuild switch
<clever>
infinisil: and then months after putting it in, he rebooted
<clever>
infinisil: the failure to reach the network, causes stage-2 to abort in a weird place, and then $PATH isnt configured properly
<clever>
infinisil: and that leads to nixos being unable to find systemd, lol
<infinisil>
whew
<Myrl-saki>
How do I have a clang which supports these flags but still have it wrapped? -cc1 -fsyntax-only -x c -code-completion-at
<infinisil>
Well I wouldn't put *networking* stuff in activationScripts, since that's very fragile
<infinisil>
But I think acpi stuff would probably be fine
<clever>
infinisil: just make sure to test it with a full reboot before you let garbage collection eat your rollbacks
<Myrl-saki>
clang-unwrapped is fine with it.
<lejonet>
Hmm, what is the proper way of ensuring stuff from linux headers being present in a build-env? I'm having troubles with a package that wants asm-generic/errno.h which seems to belong to the kernel headers (nixos.linuxHeaders ? )
<clever>
lejonet: have you tried adding linuxHeaders to the buildInputs?
<Dezgeg>
I'd think glibc already provides enough of the headers to have asm-generic/errno.h
<srhb>
infinisil: Oh, here. It outputs the sysfs on the far right side
<srhb>
infinisil: For instance, platform:PNP0C0D:00
<srhb>
infinisil: So.. echo disabled > /sys/devices/platform/PNP0C0D:00/power/wakeup should work I think
<srhb>
infinisil: Which makes it at least idempotent :)
<lejonet>
clever: that is what I'm thinking about doing, I tried doing a nix-shell '<nixpkgs>' -A linuxHeaders, and didn't get anything in /usr/include, but I probably just don't know how to use nix-shell
<clever>
lejonet: first, that gives you a shell suitable for building the linuxHeaders, not using them
<infinisil>
srhb: Nice!
<srhb>
I haven't tested though!
<clever>
lejonet: second, it never puts them in /usr/include, it sets up some env variables so gcc will just find them
<clever>
srhb: writing to the wakeup "file" in the right /sys path doesnt have an effect on the /proc/acpi/wakeup output, but it might be several layers of filtering
<srhb>
Hm, darn...
<srhb>
Too naive :-)
<clever>
srhb: would need to actually suspend to confirm if it has any impact
<clever>
srhb: and my system also has a glitch, where it sometimes locks up while going into suspend
<clever>
it just freezes solid, while leaving the keyboard and lcd backlights on
<clever>
so it didnt finish going into suspend
<infinisil>
srhb: clever: Actually the entries in /sys/devices and /proc/acpi have different values right now (disabled and enabled), so that's probably not it
<clever>
infinisil: same
<srhb>
It does not work, indeed. Sorry.
mekeor has quit [Ping timeout: 265 seconds]
<Myrl-saki>
Is there a clang wrapper flag that says pass all these to the underlying binary?
<Myrl-saki>
clang -cc1 -fsyntax-only -x c -code-completion-at -:62:48 -
<Myrl-saki>
Gives me unknown argument
<Myrl-saki>
While using unwrapped gives me standard header errors.
adisbladis has joined #nixos
<infinisil>
srhb: I have an idea, get the contents of every file in /proc and snapshot this, toggle the option, and snapshot again, then do a diff -r :D
<srhb>
yikes :P
<clever>
infinisil: several of the files in /proc are magic and do fun things when read
<infinisil>
oh no
ssmike has joined #nixos
<infinisil>
just when read though??
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/master 1f18a7c Robert Helgesson: perl-Parallel-ForkManager: init at 1.19
<NixOS_GitHub>
[nixpkgs] rycee pushed 2 new commits to master: https://git.io/vA3gE
<NixOS_GitHub>
nixpkgs/master 49a24af Robert Helgesson: cloc: 1.74 -> 1.76
NixOS_GitHub has left #nixos [#nixos]
<clever>
infinisil: for example, /proc/kmsg removes bytes from a buffer as you read it
<clever>
infinisil: so next time you read it, you only get new messages
<infinisil>
oh
<infinisil>
eh it's gonna be fine, here i go
Ross has quit [Ping timeout: 248 seconds]
<clever>
and if you write to /proc/kmsg, it appears in the main dmesg buffer
<infinisil>
so it's like a queue
<clever>
so you can insert helpfull markers into the real dmesg logs
<infinisil>
or named pipe
<clever>
i think /proc/kmsg has its own read-pointer within the dmesg loop
<clever>
and that persists between opens of the handle
<clever>
infinisil: also, /proc/self is a symlink to the pid that is trying to read the symlink
<clever>
infinisil: and all of the numbered "directories" exist for every pid, and will vary
<infinisil>
didn't work indeed, cpu spinning and doesn't seem to progress
<clever>
infinisil: there is also a root symlink under every one of those, that points to /
<infinisil>
yeah it got stuck on the very first one even
<infinisil>
well it was worth a try
<clever>
infinisil: while its hung, find the pid of the process that is hung
<clever>
infinisil: then do ls -l /proc/PID/fd/
<clever>
what files does it have open?
Ross has joined #nixos
<infinisil>
/proc/1/task/1/pagemap
<clever>
infinisil: thats a binary file, a bitmap of pages and some state
<clever>
and the task directory contains symlinks for every thread in the process
<clever>
i suspect its generating a list with the state of every single page in the entire 64bit address space
hamishmack has joined #nixos
rauno has joined #nixos
<lejonet>
clever: hmm, odd, it seems like the package has linuxHeaders in its buildInputs, behind a optionals stdev.isLinux, so linuxHeaders should be added to buildInputs, weird
dan_b has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nix] catern opened pull request #1869: fix the nix-profile scripts to error on bad profile ownership (master...master) https://git.io/vA32d
<romildo>
When writing a derivation that depends on some packages from a group of packages like gnome3.eog and gnome3.evince, is there any difference between naming them individually (like {pkgs, eog, evince}) or naming just the group (like {pkgs, gnome3}) and then referring to them as attributes in the group (like gnome3.eog) in the derivation?
<clever>
romildo: you can only do { eog, evince }: if your file is loaded via something like gnome3.callPackage
<romildo>
Is one of these two styles preferable?
<romildo>
clever, suppose it is loaded like that.
<clever>
romildo: if your going to be using a lot of gnome3 packages, and/or your thing is part of gnome3, then using gnome3.callPackage is usually better
<Myrl-saki>
What the hell
<Myrl-saki>
I have no idea how the clang wrapper works.
<lejonet>
clever: meeh, it seems like the actual include is #include "/usr/include/asm-generic/errno.h" -.- someone was stoopid when they coded that
<clever>
Myrl-saki: have you done `which clang` and then just read the bash script it points to?
<Myrl-saki>
clever: Yep.
<Myrl-saki>
clever: Not extensively, ofc.
tmaekawa has quit [Client Quit]
<Myrl-saki>
I'm just doing greps rn.
<clever>
lejonet: heh, yeah, nix cant deal with that, all nix does is slap on a ton of -I's
<lejonet>
clever: yeah, I would argue that its bad practice to do that, includes _should_ be relative, so that you don't have this stupid problem
<Myrl-saki>
Okay. I understand what's happening now.
<Myrl-saki>
This is going to be a stupid question.
<Myrl-saki>
Is there a way to make a constant variable in bash? lol
<sphalerite>
Myrl-saki: readonly foo=bar iirc
<Myrl-saki>
sphalerite: Yeah, that's what I tried, but I guess it doesn't work on exec?
<lejonet>
clever: hmm, how would I use nix-shell to step, by step, build a package, so that I can modify some stuff in between phases to see if a workaround works before making a patch?
<clever>
Myrl-saki: lol, the man page for bash uses both read-only and readonly to refer to read only variables
Ubercow has quit [Quit: Ubercow]
freeman42x]NixOS has joined #nixos
<Myrl-saki>
Basically, what happens is that the clang wrapper wants to link if it finds a filename.
<infinisil>
clever: How dare they
<sphalerite>
Myrl-saki: oh yeah at that point they're just environment variables and the child process can do what it wants
<sphalerite>
Wait
<sphalerite>
No... I don't get it
<sphalerite>
What's the problem?
<Myrl-saki>
sphalerite: It mangles the flags.
<clever>
lejonet: oh, do you already know what you want to edit?, it might be simpler to use another trick i use
<clever>
lejonet: nix-shell '<nixpkgs>' -A hello, then run unpackPhase, and then duplicate the entire source tree with `cp -r`, so you have a new and -orig copy
<clever>
lejonet: then edit the new one, and use `diff -ru` to make a patch
<Myrl-saki>
That doesn't work, because it passes `-Wl...`, which instructs the linker, which breaks the successive flags.
<lejonet>
clever: that is kindof what I wanted to do
<Myrl-saki>
Actually, maybe the clang wrapper itself should be patched.
<Myrl-saki>
Nah.
<Myrl-saki>
I'll just use libclang.
<infinisil>
clever: nice!
<Myrl-saki>
llvmPackages.libcxxClang
<Myrl-saki>
That's not libclang, I believe?
freeman42x]NixOS has quit [Ping timeout: 248 seconds]
Ross has quit [Ping timeout: 255 seconds]
Ross has joined #nixos
<Myrl-saki>
TIL
<Myrl-saki>
clang-unwrapped seemed to work.
andrewrk has joined #nixos
MP2E has joined #nixos
<sphalerite_>
Myrl-saki: my guess is that's libc++
<Myrl-saki>
How do I register a root without having it in my user directory?
<Myrl-saki>
Maybe I should RTFM first lol
<Lisanna>
>wrote a string find/replace function in Nix
<Lisanna>
> find builtins.replaceStrings after I get it working
chrisbar1 has joined #nixos
asuryawa_ has quit [Remote host closed the connection]
<Myrl-saki>
I just made a .roots directory, so that I can manage it in userspace. This is funnily becoming somewhat imperative.
asuryawanshi has joined #nixos
smichel17 has quit [Remote host closed the connection]
goibhniu has quit [Ping timeout: 248 seconds]
<Myrl-saki>
Okay, I need another help now.
<Lisanna>
Myrl-saki IMO there should be a NixOS module which lets you give a list of drvs to register as roots or something
<Myrl-saki>
If I use `[ clang llvmPackages.clang-unwrapped ]`, which is first in PATH?
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] bachp opened pull request #34942: Update minio client and server (master...feat/minio-update) https://git.io/vA3XZ
<Myrl-saki>
One of my biggest gripe about NixOS is probably that configuration.nix doesn't do some things that nix-shell does.
<gchristensen>
like what?
<pie__>
dtz, i hear musl stuff merged, nice \o/
<Myrl-saki>
gchristensen: Libraries for one.
<gchristensen>
ah
<Myrl-saki>
It's not so much a problem, really. It's pretty pebcaky.
<Myrl-saki>
Who here uses Emacs and actually uses emacsWithPackages? I need advice.
ssmike has quit [Ping timeout: 256 seconds]
<Myrl-saki>
There's actually an irony-server package. That's surprising.
<Myrl-saki>
Actually, I'm all-in for that. For Emacs packages that require a binary, use Nix for the binary, while emacs donwloads the .el, I can see desync of the API/ABI though.
romildo has quit [Quit: Leaving]
<Myrl-saki>
The irony. Double irony even. I'm installing a package called irony, and I got jinxed by what I said earlier.
ambro718 has joined #nixos
<infinisil>
Myrl-saki: i am using emacswithpackages
<sphalerite_>
So it seems I can work around the `nix copy` out-of-memory issue by using nix-store --export | ssh root@foo nix-store --import … it's not pretty but it doesn't seem to be running out of memory. This seems like a serious deficiency in nix copy though
<Myrl-saki>
infinisil: Can you send me your .emacs and your configuration.nix?
<infinisil>
Because if so, it should also work on ubuntu (will have to extract it from configuration.nix though)
etu has quit [Read error: Connection reset by peer]
<cement>
I've got the vimrc
<cement>
I'm not confident in the packages that go with that
etu has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill opened pull request #34943: nginx/modules: use eustas' brotli module, google one is abandoned (staging...feature/ngx_brotli-eustas) https://git.io/vA3Sy
NixOS_GitHub has left #nixos [#nixos]
<sphalerite_>
cement: are your bosses going to require you to use ubuntu without installing nix on it?
humanoyd has quit [Quit: WeeChat 2.0.1]
<cement>
no, that's only on the NPMs I have to deploy to
<cement>
this laptop I can install nix on just fine
<wavewave>
how can i check if a directory exists inside a package in /nix/store?
<infinisil>
wavewave: huh? I don't understand
<cement>
extra, possibly relevant info: I'm using nix itself to manage the vim packages I have installed
<wavewave>
I am trying to use 'pathExists' but it complained '/nix/store...' cannot refer to other paths
<wavewave>
infinisil: in nix expression.
<wavewave>
if I have a nix variable for the package.
<wavewave>
to explain further, I am making a function that takes pkg as an argument and check whether "${pkg}/lib" exist.
<sphalerite_>
that's a form of import-from-derivation
<lejonet>
lol, always fun when nixops deploy stops the network, the small heartattack you get :P
<infinisil>
wavewave: Ah, in that case you can check for the existence by doing it at build time instead: use bash to iterate over the build inputs and set this env var according to the result
<infinisil>
But it won't be nice
<Profpatsch>
lewo: But from what I saw skopeo needs root?
<infinisil>
wavewave: what does "effectively" break a build mean? Because warning messages shouldn't really break anything
<Profpatsch>
Not sure why tbh
<Profpatsch>
Let me merge the PR3
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Profpatsch pushed 1 new commit to master: https://git.io/vA3Qc
NixOS_GitHub has left #nixos [#nixos]
<NixOS_GitHub>
nixpkgs/master eef9d36 Profpatsch: buildah: init at 0.11
<wavewave>
infinisil: it's just our internal thing. disallow any warnings :-P
<Profpatsch>
Ah, of course I meant buildah.
knupfer has joined #nixos
<infinisil>
wavewave: well then you're not gonna like nixpkgs, there's loads of warnings alllll over the place..
chreekat has quit [Quit: quitting]
<wavewave>
infinisil: it's not that bad. only Haskell built by stack. anyway thank you for your suggestion!
<wavewave>
at least, i confirm that using pathExists is a dead end.
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
nixpkgs/staging 193ebca Vladimír Čunát: libmpc: 1.0.3 -> 1.1.0...
<NixOS_GitHub>
nixpkgs/staging f8865b1 Vladimír Čunát: mpfr: 3.1.3 -> 3.1.6 (bugfix)...
<NixOS_GitHub>
[nixpkgs] vcunat pushed 2 new commits to staging: https://git.io/vA37I
NixOS_GitHub has left #nixos [#nixos]
wip_ has quit [Quit: brb]
rihards has quit [Quit: rihards]
chreekat has joined #nixos
alexteves_ has quit [Ping timeout: 240 seconds]
pkill9 has quit [Ping timeout: 264 seconds]
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] Lassulus opened pull request #34945: gnunet service: set private TMPDIR (master...gnunet-tmpdir) https://git.io/vA3d3
<clever>
troydm: and the patches themselves, add #if statements, that check the versions
<troydm>
clever: because I'm on 4.9 kernel and my wl driver connection is not stable and I have tons of TX errors, apparently some of those patches fix this problem but I'm still experiencing unstable connection on 4.9 kernel
<clever>
so the compiler will conditionally apply things at compile time, based on the version of the kernel source
<troydm>
I'm currently trying to switch to 4.13 to see if it will help or not
<troydm>
but I think I should have gone for 4.12 instead
kelleyNif has quit [Quit: Leaving.]
<troydm>
also how do I enable parallel compilation because currently kernel is compiled using only 1 processor
freeman42x]NixOS has quit [Ping timeout: 248 seconds]
<samueldr>
(next time please use a pasting service for anything bigger than two lines (though at this time there wasn't any other conversation going on) messages get throttled)
<gchristensen>
can Nix tell me the cryptographic sum of a store path? ie: sha256 of the $out, not the hash used to calculate the $out
<samueldr>
sorry, no experience with wl/broadcom hardware, though I would have loved to say I did, only to know how to help
<troydm>
so I'm assuming those are related to TX errors and some ppl suggest cfg patches
Leon[m] is now known as Leon[m]1
goibhniu has joined #nixos
NixOS_GitHub has joined #nixos
<NixOS_GitHub>
[nixpkgs] dtzWill closed pull request #34916: xz: set CONFIG_SHELL to /bin/sh, fix retained reference to bootstrap (staging...fix/xz-invalid-references) https://git.io/vAYSk
NixOS_GitHub has left #nixos [#nixos]
coot has joined #nixos
leons is now known as leons_legacy
Leon[m]1 is now known as leons
<samueldr>
though, if you have good suggestions for your macbook hardware from other distros, and don't know how it maps to nixos, pretty sure everyone here is glad to help :)
<troydm>
apparently 4.12 related patch among those is cfg related
<samueldr>
you're on 4.14 yet or haven't finished rebuilding/booting?
<troydm>
samueldr: it's still building 4.13, just want it to finish to check if it'll work with 4.13 or not, will rebuild for 4.14 anyway
<Profpatsch>
gchristensen: Can ofborg also run VM tests?
<samueldr>
ah, 4.14 should already be built though (iirc)
<leons>
samueldr: I'm actually having the exact same wifi issues under every linux distro with my Macbook. NixOS actually has the best support out of the box apart from that :)
<Profpatsch>
gchristensen: Hm, some tests are not in the release.nix, e.g. the rabbitmq.nix
<gchristensen>
hmm why not?
<Profpatsch>
Probably forgot them.
<gchristensen>
duh, they're not release blockers...
<Profpatsch>
You can run tests directly with nix-build, nix-build nixos/tests/rabbitmq
<Profpatsch>
.nix
<gchristensen>
hmm no
<gchristensen>
Profpatsch: I won't do that. they should be added to release.nix
<gchristensen>
release.nix doesn't make it a release blocker
<Profpatsch>
Can’t ofborg execute arbitrary nix files in the repo?
<gchristensen>
absolutely, most certain, no, it can't execute arbitrary Nix. it can only call ./default.nix and ./nixos/releas.nix
<troydm>
leons: do u have them fixed on NixOS? which kernel are u running?
<Profpatsch>
Well, if it can call default.nix it practically can execute arbitrary nix, right? :)
<gchristensen>
nope
<gchristensen>
because ./default.nix can't import ../default.nix and arbitrary nix could allow bugs for referencing ../default.nix
<Profpatsch>
Of course it should only be able to execute files in the repo.
kelleyNif has joined #nixos
<gchristensen>
I depend on Nix's purity checks to enforce it
<Profpatsch>
Do you containerize the checkout path?
<gchristensen>
no
<Profpatsch>
Oh, okay.
ma271 has quit [Ping timeout: 276 seconds]
<leons>
troydm: Sorry, I haven't fixed them. I tried various driver and kernel versions, it was always horrible. I've now switched to another notebook entirely tough checking regularly if the macbook has better support. I remember that using connman somehow worked pretty reliably under arch
<gchristensen>
it really depends on Nix's security being good
ditadi has joined #nixos
<Profpatsch>
But nix sandboxing is enabled, yes?
<gchristensen>
yeah :)
<infinisil>
troydm: leons: Well I'm using a MacBook Air 2012 and don't have any problems fwiw
<troydm>
on a side note I have xterm/urxvt related cursor theme issue, I've set default cursor theme in Xresources but apparently nor xterm nor rxvt don't want to use it, tried every possible thing, problem is that screen on this Macbook is Retina HiDPI so default xterm cursor is very tiny
<troydm>
Xresouces fixes this issue for other apps
<leons>
infinisil: I've got a friend who, if I remember correctly, has that exact macbook model. Works perfectly for him under Arch too, must have a different wifi chip...