<ottidmes>
gchristensen: my big bit of Nix code in that shell function does that on a broader scope, all attributes of the package + all attributes of the meta attribute of the package
jasongrossman has joined #nixos
dermetfan has quit [Ping timeout: 245 seconds]
hamishmack has joined #nixos
<ottidmes>
gchristensen: so actually there more unrelated attributes to the package itself are inside attrset passed to mkDerivation the more likely my function will be able to pinpoint the location, because those attributes will not be reset
carlosdagos has joined #nixos
<gchristensen>
aye
<gchristensen>
cool!
tdbgamer has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
shpx has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
eadwu has joined #nixos
thc202 has quit [Ping timeout: 246 seconds]
shpx has joined #nixos
hamishmack has joined #nixos
shpx has quit [Ping timeout: 268 seconds]
xkapastel has joined #nixos
<duairc>
Wow, this is incredible, not only was I able to import old commits of nixpkgs to get ancient versions of GHC, I was able to add them to my environment.systemPackages and the binary builds are still cached! Didn't have to build anything and they all seem to actually work too
goibhniu1 has quit [Ping timeout: 244 seconds]
<gchristensen>
isn't it magic?
sinner has joined #nixos
<duairc>
And like, *of course they work*, that's just a property of how Nix is built, unless the kernel made some funny change somewhere along the way
sinner is now known as Guest35183
<gchristensen>
=)
<duairc>
I am surprised the binaries are still cached though, do they get cached forever?
<Myrl-saki>
Apparently, builtins.toJSON can not necessarily be reified back with fromJSON.
<Myrl-saki>
error: the string '[{"paths":["/nix/store/q0aw9i16mj12mswj11arir2rb7fr821m-weechat"],"priority":5},{"paths":["/nix/store/p6g25z4da2jcqk551l4zgscflfj2f6m6-weechat-headless"],"priority":5}]' is not allowed to refer to a store path (such as '!out!/nix/store/kja56wqdcjf490b17f61dw8s130i84hk-weechat.drv'), at (string):1:2
<gchristensen>
heh.
<Myrl-saki>
It's a srs problem. ; ~ ;
<Myrl-saki>
It means that there's no way to get the underlying `weechat` without changing the derivation itself.
<gchristensen>
...eh?
<Myrl-saki>
Oh.
<Myrl-saki>
Apparently, a weechat-unwrapped package exists in master now.
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<ottidmes>
is there some easy way to get to the data within a .drv file from a shell?
teto has joined #nixos
Diagon has joined #nixos
<teto>
it seems like I have grub entries that are not valid anymore because the kernels have been garbage-collected. Is that expected ?
<ottidmes>
teto: yeah, only if you do nixos-rebuild boot or nixos-rebuild switch --install-bootloader (I believe that flag is necessary for switch to have the same effect), will it remove those entries
acarrico has quit [Ping timeout: 240 seconds]
iqubic` is now known as iqubic
<teto>
ottidmes: I was kinda surprised, but I guess generating bootloader entries is sensitive enough we don't want to run it everytime ?
<ottidmes>
teto: I do not know the exact reasoning behind it, but it will probably something like that, yes
iqubic has left #nixos ["ERC (IRC client for Emacs 26.1)"]
iqubic has joined #nixos
<gchristensen>
you don't need --install-bootloader
<gchristensen>
the trouble is the bootloader entries are not part of the normal store garbage process, and there is no hook for garbage collection to tell nixos to prune those entries
<ottidmes>
gchristensen: I saw that flag mentioned a few times in similar context, but after checking the logs it seems it was only relevant when changing bootloaders
aanderse_ has quit [Read error: Connection reset by peer]
aanderse_ has joined #nixos
<samueldr>
a good way to reason of that behaviour is how it's not adding or removing bootloader entries, but regenerating the list from what's in the store
Diagon has quit [Quit: Leaving]
chessai has quit [Remote host closed the connection]
<samueldr>
nix itself has no concept of "outside the store" so collecting there can't affect those :/
<gchristensen>
outside ... the ... store ..? :confused:
<ottidmes>
gchristensen: samueldr: do any of you know how to deal with the notation in a .drv, I wish it used JSON instead...
<ottidmes>
samueldr: nix-store --query keeps reporting useless information while the right information is right there in the .drv, so I figure, better get it straight from there
<jasongrossman>
samueldr: LLOL at the allegory of the cave.
<ottidmes>
gchristensen: no wonder I did not remember it reporting it as json, considering the description "show the contents of a store derivation."
<gchristensen>
we cannot free ourselves from phenomenal state but dangit we can try, samueldr
<samueldr>
ouch, no, bad samueldr, no self nerd-sniping
<samueldr>
what if... the bootloader was aware of nix generations innately?
<samueldr>
like, I have a rough idea congealing that makes me think it's a week-end project that's doable... which is scary :)
<jasongrossman>
Or the holographic theory of the universe.
Supersonic has quit [Ping timeout: 252 seconds]
<jasongrossman>
I'm happy GCing my boot partition manually. After all, it's a different partition, and doesn't fill up at the same time as my /nix (or indeed ever).
thadthemad has joined #nixos
strobelight_ has joined #nixos
strobelight is now known as Guest97276
bandali has joined #nixos
<gchristensen>
mine can only fit like 25 :| generations
strobelight_ is now known as strobelight
<teto>
I might open an issue because seeing 3/4 generations in bootloader gave me a false sense of security (like I can fallback 3 generations in the past) while I could only get back one generation
<samueldr>
my idea, would store zero, and considering secure boot, if the drive being opened was encrypted, would have (I think) passable security
<teto>
which was enough to save my ass
Supersonic has joined #nixos
<gchristensen>
aye, it is impossible to GC the current system, or the system which was booted
Guest97276 has quit [Ping timeout: 258 seconds]
<gchristensen>
samueldr: what is your idea? :)
<samueldr>
kexec-based bootloading
<gchristensen>
...oh? :o
<samueldr>
not sure what hurdles are there for real, but there are at least two reall implementation of the idea
jb55 has joined #nixos
<samueldr>
and I know there's a third one I can't remember the name
thadthemad has quit [Ping timeout: 268 seconds]
<samueldr>
though here it would be a specialized one, nixos-only thing
kvda has joined #nixos
<q6AA4FD>
is there any support for nim-lang/nimble a la haskellPackages/cabal2nix
<gchristensen>
samueldr: nice. some hardware doesn't handle kexec very well, but I only know of one chassis which doesn't. so.... I'd ignore that.
<samueldr>
I think the idea is to *not* add the fancy drivers for anything, boot nomodeset and such and then you're pretty fine for 99% of the cases
<jasongrossman>
teto: Good point.
<samueldr>
(knowledge from previous reading sessions)
<q6AA4FD>
samueldr: you could maybe actually make it aware if you built it into the initramfs and used refind
<gchristensen>
cool!
<samueldr>
q6AA4FD: what do you mean?
sanscoeur has quit [Ping timeout: 250 seconds]
<q6AA4FD>
refind will automatically populate its entries based on the kernel.efi files in its boot directory, and use the name to pair with an initramfs
<samueldr>
ah, the idea here is not have pairs of initrd and kernels in the ESP :)
<q6AA4FD>
oh, i see. i didn't realize initramfs were large enough to cause a problem
<samueldr>
q6AA4FD: is this your usual nickname?
<q6AA4FD>
samueldr: yup, why do you ask?
<samueldr>
don't remember interacting with you, and it looks like a placeholder when nickserv boops a nick
<q6AA4FD>
my initramfs is 5.6M, does nixos expand tem?
<samueldr>
(no offense)
d4g has joined #nixos
<ottidmes>
I am trying to figure out whether it is possible to determine whether a given hash is pointing to an older version. I had hoped it was a matter of asking the output (which is reused if the hash stays the same but e.g. url is changed) what its derivation was, but that information seems useless, it point to the very first derivation that produced the output, which tends to be a bogus hash when applying TOFU)
<samueldr>
it depends what you add to them I guess, and how many generations you want to keep
<q6AA4FD>
nah, you're good. i've just got into nix, so i can document my 2k line .org file full of shell scripts in a less ad-hoc way. and 6AA4FD is the hex code for sky blue
<gchristensen>
nice
<jasongrossman>
q6AA4FD: Some people have very small boot partitions, but big /nix partitions, so they're used to keeping lots of generations.
<gchristensen>
my initrd + kernels are 19mb
<samueldr>
around the same here
<jasongrossman>
Same.
<ottidmes>
33.6M
d4g_ has quit [Ping timeout: 245 seconds]
<gchristensen>
dropbear or something?
<ottidmes>
worse, openssh
<gchristensen>
oops
<ottidmes>
by choice, I do not mind a larger initrd, but dont like the idea of having multiple pieces of important software doing the same thing
<samueldr>
gchristensen: which makes me think, wouldn't sshing-in to select a generation be nice?
<gchristensen>
yes!
<gchristensen>
like a P.reboot eXecution Environment but better
<{^_^}>
[nixpkgs] @matthew-piziak opened pull request #53981 → add textlint to nodePackages → https://git.io/fhcXZ
<samueldr>
though I'm not sure how ready it is for x86_64 I think it's mainly geared to be used as firmware for Open Power architecture
<srk>
> builtins.match "^[0-9]+\.$" "100"
<{^_^}>
[ ]
<srk>
doesn't look right
<samueldr>
srk: looks right to me, there's a dot, no?
<samueldr>
well, a period
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<srk>
would expect it to match "100." but not just "100"
<ottidmes>
does anyone know if somehere is being tracked (in the sqlite db of the Nix store maybe) what derivation actually produced an output, rather than just claiming it will produce it?
<ottidmes>
infinisil: in theory yeah, but I am not getting the right answers from there
<ottidmes>
infinisil: I am basically getting the first ever derivation claiming to build the output
<gchristensen>
worth filing a bug over that
<ottidmes>
gchristensen: guess I will just do that, otherwise it makes it useless
<gchristensen>
although it will always be true that many drvs may have produced a given path
aleph- has joined #nixos
<samueldr>
gchristensen: what makes my idea worth looking into as a side-project is how it would be useful for mobile-nixos since there's no way to run a real bootloader easily on the targeted devices
<samueldr>
* (no sane and safe way)
<clever>
samueldr: when i ran linux on my treo 650, the cpu lacked an MMU, so any app could just hijack the cpu and run another OS
<samueldr>
that's cheating
<gchristensen>
LOL
<jackdk>
beautiful. I am reminded of LOADLIN.EXE and GRUB4DOS
strobelight has quit [Ping timeout: 268 seconds]
<clever>
jackdk: the last machine i tried loadlin on, was a 486SX, and it lacked the FPU
<clever>
so it needed a special linux build to even work
<samueldr>
walking uphill both ways
bhipple has joined #nixos
sigmundv_ has quit [Ping timeout: 246 seconds]
<bhipple>
If I run nix copy --to file:///tmp/foo, is there a way I can unpack foo and install it without nix on the machine?
snajpa has quit [Ping timeout: 244 seconds]
<bhipple>
Essentially, I'm looking to create a bootstrap.tar.gz that has a custom nix and a pre-built environment, that can be unpacked and installed on servers to get the nix toolchain on there
<bhipple>
Wondering what the most efficient wayt to do this might be
aleph- has quit [Ping timeout: 258 seconds]
aither has quit [Ping timeout: 252 seconds]
snajpa has joined #nixos
Lingjian has joined #nixos
eadwu has quit [Ping timeout: 268 seconds]
aleph- has joined #nixos
aither has joined #nixos
lassulus_ has joined #nixos
lassulus has quit [Ping timeout: 246 seconds]
lassulus_ is now known as lassulus
jackdk has quit [Ping timeout: 246 seconds]
Lingjian has quit [Ping timeout: 252 seconds]
<wedens>
is there some example of cross-compiling cmake+qt application on linux -> mingw?
<{^_^}>
#53983 (by msteen, 23 seconds ago, open): nix-store --query --deriver does not work as expected
vielmetti has quit [Ping timeout: 250 seconds]
rprije has joined #nixos
<q6AA4FD>
does anyone have a minimal example of a derivation or package where all the major phases are overwritten so a package in an unsupported language can be built?
hamishmack has joined #nixos
vielmetti has joined #nixos
<bhipple>
I'd look at the generic/setup.sh stdenv script to see all of the phases
<bhipple>
These are all the phases you get by default, but if you want to stomp all of them at once you can just set phases = [x, y, z]
<bhipple>
in your package's nix expression
<q6AA4FD>
bhipple: by stomp, you mean limit the phases that are run to the array provided?
<bhipple>
If you don't provide an attribute called phases in your nix expr, you get exactly the ones in that list; you can provide your own to overwrite which phases do or do not get run
<bhipple>
you can also substitute in your own implementation of any given phase. The cmake and waf helpers do this, for instance, with the hook mechanism
<q6AA4FD>
bhipple: i see, thanks for your help!
xkapastel has quit [Quit: Connection closed for inactivity]
<bhipple>
You're welcome. Normally you don't have to replace all of the phases, you likely just have to replace buildPhase and installPhase
<iqubic>
So I'm doing some haskell development and everytime I update my project's cabal file, I need to remember to run cabal2nix. IS there a way to get nix to do that for me?
aleph- has quit [Ping timeout: 268 seconds]
ashkitten has joined #nixos
<iqubic>
It would be best if cabal2nix could be run when I call nix-shell.
fusion809 has joined #nixos
tdbgamer has quit [Ping timeout: 258 seconds]
pie__ has quit [Ping timeout: 268 seconds]
<ghostyy>
i have a question - i am making a custom option for my nixos system configuration, it involves setting boot.initrd.postDeviceCommands to a shell script with some environment variables defined by nix variables (eg in the script i might have $foo, and in the options definition i might have foo = mkOption { type = types.listOf types.str; }
<ghostyy>
how do i define boot.initrd.postDeviceCommands so that it expands those variables based on the nix variables in the options definition?
mtesseract has joined #nixos
<ottidmes>
iqubic: there is callCabal2nix which I believes does that, but I have not tried it yet
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to staging « llvm_6, llvm_7: build all default targets (#53941) »: https://git.io/fhcSe
tzemanovic has joined #nixos
<iqubic>
I'd like to know how that works.
tzemanovic has quit [Ping timeout: 252 seconds]
abathur has quit [Ping timeout: 240 seconds]
mtesseract has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
zgrep has quit [Quit: This is me *really* leaving.]
zgrep has joined #nixos
reinhardt has joined #nixos
ddellacosta has quit [Remote host closed the connection]
zgrep has quit [Quit: This is me *really* leaving.]
zgrep has joined #nixos
<ghostyy>
how do i do shell substitution in builtins.readFile
ddellacosta has joined #nixos
<infinisil>
ghostyy: A bit more specific?
ottidmes has quit [Ping timeout: 258 seconds]
<ghostyy>
i am setting boot.initrd.postDeviceCommands to a shell script in an external file
<ghostyy>
in that file i use variables that are supposed to refer to nix variables with the ${foo.bar} syntax
<infinisil>
Oh, builtins.readFile can't do that
<ghostyy>
ah
<ghostyy>
do you know how to do that then?
<infinisil>
You'd have to define the file as a Nix file, wrap everything in '', and make it a function so you can pass the arguments
<ghostyy>
ah... so its impossible basically
<infinisil>
And then `import ./file.nix { foo = "..."; }'
<ghostyy>
the shell substitution syntax works fine when its inline
<infinisil>
Well, there's some alternatives
<ghostyy>
eg blahblah = "... ${foo.bar} ..."; works
<infinisil>
Yeah, because that's a Nix expression
<infinisil>
Nix can't import non-expressions
<ghostyy>
ah..
<infinisil>
You could use builtins.scopedImport to avoid having to declare the arguments at the top of the file
<infinisil>
Or a more common approach is to use @foo@ where you need the arguments (declare it as a normal file and use those), then run a `substituteAll` over the file in a runCommand to insert the arguments
<ghostyy>
if i just used boot.initrd.postDeviceCommands = ''\n...${foo.bar}..\n'' would that work?
<ghostyy>
ooh
<ghostyy>
substituteall sounds way better
<ghostyy>
since id like to keep it in a seperate file
<infinisil>
You can't have Nix in the file then though, it's gonna be a simple string substitution
<infinisil>
And yes, that example you mentioned does work (try it out)
<ghostyy>
ah i see, thank you!
<infinisil>
:)
* infinisil
goes to sleep now
reinhardt has quit [Quit: Leaving]
Rusty1 has quit [Quit: Konversation terminated!]
vk3wtf has quit [Ping timeout: 252 seconds]
thadthemad has joined #nixos
doyougnu has quit [Ping timeout: 268 seconds]
thadthemad has quit [Client Quit]
doyougnu has joined #nixos
bhipple has quit [Remote host closed the connection]
endformationage has quit [Quit: WeeChat 2.3]
reinhardt has joined #nixos
doyougnu has quit [Ping timeout: 258 seconds]
romanofskiWork has quit [Ping timeout: 250 seconds]
reinhardt has quit [Ping timeout: 268 seconds]
romanofskiWork has joined #nixos
tzemanovic has joined #nixos
tzemanov_ has joined #nixos
<{^_^}>
[nixpkgs] @thoughtpolice pushed 2 commits to master: https://git.io/fhcQG
<{^_^}>
[nixpkgs] @thoughtpolice pushed to master « Revert "ocamlPackages.zarith: use buildOcaml instead of mkDerivation" »: https://git.io/fhcQn
tzemanovic has quit [Ping timeout: 252 seconds]
tzemanov_ has quit [Ping timeout: 252 seconds]
<{^_^}>
[nixpkgs] @thoughtpolice opened pull request #53984 → ocamlPackages.zarith: use buildOcaml instead of mkDerivation → https://git.io/fhcQ0
<{^_^}>
[nixpkgs] @thoughtpolice merged pull request #53984 → ocamlPackages.zarith: use buildOcaml instead of mkDerivation → https://git.io/fhcQ0
<{^_^}>
[nixpkgs] @thoughtpolice pushed to master « ocamlPackages.zarith: use buildOcaml instead of mkDerivation »: https://git.io/fhc5X
thadthemad has quit [Ping timeout: 268 seconds]
freeman42x has joined #nixos
jasongrossman has quit [Ping timeout: 268 seconds]
vk3wtf has quit [Ping timeout: 252 seconds]
ddellacosta has quit [Remote host closed the connection]
rauno has joined #nixos
sinner has joined #nixos
sinner is now known as Guest39954
jasongrossman has joined #nixos
Ariakenom has joined #nixos
goibhniu has joined #nixos
Makaveli7 has joined #nixos
carlosdagos has quit [Quit: Connection closed for inactivity]
Tucky has joined #nixos
freeman42x has quit [Ping timeout: 252 seconds]
civodul has joined #nixos
Guanin has joined #nixos
<fusion809>
Hi, which package provides the Python module pari_instance, because I'm getting this error https://paste2.org/J0FUnHg7 when installing SageMath 8.5 from nixos-unstable.
juhulian has quit [Ping timeout: 258 seconds]
thc202 has joined #nixos
<schmittlauch[m]>
Hi, I want to install Nix on a system I have no root access on -> I cannot create a /nix directory. How do I install Nix with a different nix store location located in my home directory?
<elvishjerricco>
schmittlauch[m]: If your distro has user namespaces, you can use a chroot nix store. Otherwise you'll have to forego the binary cache (which assumes `/nix`), and bootstrap a big compiler toolchain.
<elvishjerricco>
As for how exactly to do those things, I'm not exactly sure TBH
<schmittlauch[m]>
eww, I thought that was easier
<elvishjerricco>
schmittlauch[m]: IIRC, it's not too bad to install it in a custom location if you don't mind losing the binary cache
<schmittlauch[m]>
but I see that RPATHs probably should be absolute
<elvishjerricco>
The nix expressions automate the whole bootstrapping for you
<elvishjerricco>
so that's mostly a CPU time problem, not a complexity problem
<schmittlauch[m]>
elvishjerricco: it's a university machine I don't use that often, so I might not have the CPU time available. let's see how I do it
plakband has joined #nixos
dermetfan has joined #nixos
<schmittlauch[m]>
elvishjerricco: Can you point me to resources describing on how to set up nix in a namespaced chroot?
<elvishjerricco>
schmittlauch[m]: Not sure a resource exists, TBH. I know it's possible, but I'm not sure anyone's ever take the time to document it. You can check the Nix manual: https://nixos.org/nix/manual.
dave_uy has quit [Quit: Ping timeout (120 seconds)]
dave_uy has joined #nixos
erictapen has joined #nixos
<duairc>
So when I try to run ghci I get an error about not being able to find libz.so
<duairc>
It seems to be a fairly common thing, and can be solved by running ghci inside nix-shell -p zlib
<duairc>
But is there a way that I don't have to do that?
Asmadeus has joined #nixos
<duairc>
Also, I can't seem to build anything that depends on postgresql, it keeps not finding libpq-fe.h even though I have postgresql installed
<duairc>
Ah, nix-shell fixed that
<duairc>
I guess I'll just get used to using nix-shell
init_6_ has joined #nixos
<etu>
duairc: direnv is nice if you want to get nix-shell environments automatically when entering directories
<duairc>
etu: Okay, cool, I'll look into that
<etu>
duairc: "use nix" in an .envrc loads the shell.nix in the same directory :)
init_6 has quit [Ping timeout: 268 seconds]
<duairc>
I am kind of stressed at this point though, I've spent the last week setting up and learning NixOS and I'm not able to get any actual work done! Obviously that's my own problem, but I just hope I reach the summit of this learning curve soon...
init_6_ is now known as init_6
dogbread has quit [Quit: Konversation terminated!]
alex`` has joined #nixos
ThatDocsLady_ has joined #nixos
ThatDocsLady has quit [Ping timeout: 250 seconds]
knupfer has joined #nixos
tmaekawa has joined #nixos
<{^_^}>
[nixpkgs] @roberth pushed to master « Revert "nixos/modules/misc/nixpkgs.nix: Use pure Nixpkgs function" »: https://git.io/fhCsn
tmaekawa has quit [Client Quit]
ThatDocsLady_ has quit [Remote host closed the connection]
ThatDocsLady_ has joined #nixos
<plakband>
I posted here yesterday about my mac mini, the live CD runs fine but when booting from disk I get a black screen after GRUB ((no)modeset has no effect).
<duairc>
Actually even with nix-shell I'm still getting the libz.so error
<{^_^}>
[nixpkgs] @primeos pushed to master « swayidle: init at 1.1 »: https://git.io/fhCsr
<plakband>
Today I found out that I can actually boot from disk by first loading rEFInd from my live disk, and then booting normally from there (still going through GRUB)
<plakband>
Anybody have an idea what causes this, and how I can boot without the disk?
silver has joined #nixos
<plakband>
s/the disk/live disk
juliendehos has quit [Remote host closed the connection]
juliendehos has joined #nixos
sakalli_ has joined #nixos
sakalli_ has quit [Excess Flood]
sakalli_ has joined #nixos
periklis has joined #nixos
Jackneilll has joined #nixos
Jackneill has quit [Read error: Connection reset by peer]
sakalli_ has quit [Client Quit]
<Taneb>
Is there a way to make a hydra server use an alternative nix store?
estrom has joined #nixos
<{^_^}>
[nixpkgs] @primeos pushed to master « swaylock: init at 1.2 »: https://git.io/fhCGh
<estrom>
Does supplying and applying boot.kernelPatches also generate a new nixos generation like 'normal' configuration? I.e. will I be able to hop back to an unpatched build of the kernel?
<estrom>
Also, instead of supplying e.g. 'pkgs.kernelPatches.ubuntu_fan_4_4' can I supply my own git patches?
<duairc>
estrom: For your first question, I can't see any reason why not
knupfer has quit [Ping timeout: 250 seconds]
knupfer has joined #nixos
<duairc>
And for your secondly, yes, I also don't see why not, they're just derivations like any other
<duairc>
Take a look at nixpkgs/pkgs/os-specific/linux/kernel/patches.nix to see how they're defined
<duairc>
Sorry, they're not derivations actually, they're {name, patch} attrsets
<duairc>
name is a string and patch is a path
acarrico has joined #nixos
<estrom>
duairc: That's really neat. I like so far how simple the nixos experience has been
<estrom>
Can I find the nixpkgs/pkgs/os-specific/linux/kernel/patches.nix file on my machine? Currently looking at it on github
<hooo>
should I start using nixos for all my needs?
ottidmes has joined #nixos
<etu>
hooo: Sure, if you want to do that :-)
<srhb>
hooo: Maybe, if it suits your needs!
<hooo>
I'm so pissed off again today, Centos networking doesnt work anymore for no apparent reason. There is always something wrong with linux
<hooo>
I should have never accepted any of these system updates
<etu>
hooo: NixOS could hopefully be consistently broken then ;)
<etu>
Since it's reproducible
<hooo>
what about rollback support
<srhb>
hooo: That is indeed one of the main magic tricks. If an upgrade breaks your world, you can usually postpone fixing it by booting into an earlier system.
<hooo>
the only thing that worries me about nixos is all the new stuff I have to learn about
<srhb>
Yep, it's a steep learning curve.
<etu>
hooo: Yeah, the rollbacks have saved me more than once :-)
<rauno>
A question, is there some restrictions by nixos with network namespaces ?
<rauno>
Having a struggle with mesos and it's networking cni
knupfer has quit [Ping timeout: 258 seconds]
<ottidmes>
Is there any existing Nix tool that can answer me this question or is there a way I can query the Nix store DB directly. I tried this, but I get an error about attempting to write a readonly database: sqlite3 /nix/var/nix/db/db.sqlite 'SELECT COUNT(drv) FROM DerivationOutputs WHERE path = "/nix/store/swzxzspyqad5qdbbpiwr20bg8p7nl0ph-test.txt"'
knupfer has joined #nixos
brejoc has joined #nixos
rauno has quit [Read error: Connection reset by peer]
<ar>
hm. i've installed libvirt on nixos (added libvirt to "environment.systemPackages"), and things like virsh and libvirtd binaries are now available, libvirtd systemd unit doesn't show up
<ottidmes>
hmm the readonly error was just simply running it as root
plakband has quit [Quit: WeeChat 2.3]
<mightybyte>
I'm trying to use the NixOS gitlab-runner service to do CI. But I keep getting the following error:
<mightybyte>
"Failed to start process: exec: "su": executable file not found in $PATH"
<mightybyte>
Anyone know how to fix this?
<hooo>
How exactly does nixos ensure that packages do not depend on each other? I dont understand what the config language being functional or whatever has to do with this
brejoc has quit [Ping timeout: 268 seconds]
<mightybyte>
hooo: It stores every package in a unique content-addressed location
<hooo>
so does it make docker obsolete?
<wedens>
depends on what use case you have in mind
<mightybyte>
Not completely, but it can be used in many of the same things you'd want docker for.
<wedens>
btw, you can us nix to create small and reproducible docker images
<ottidmes>
mightybyte: you probably have to add /run/wrappers/bin to your PATH
Makaveli7 has joined #nixos
drgbr has joined #nixos
<mightybyte>
ottidmes: How do I do that? This is all in configuration.nix so my default assumption is that that's already handled.
<gchristensen>
hooo: when a package is built, it is built in a sandbox which contains nothing but the packages explicitly listed as a dependency
<gchristensen>
hooo: so it cannot accidentally depend on something you happened to have installed, because it can't look at your system
<ottidmes>
mightybyte: well if PATH is overwritten at some point, the system can set it properly all it wants, so you probably have to check first where where the error is exactly coming from and what the environment is like at that point
fuzzy-id has joined #nixos
<duairc>
I saw that cyris212 had the same problem as me with zlib.so a few days ago, they said that nuking ~/.ghc worked for them, but it didn't for me
<duairc>
Specifically, snap fails to build for me, looking for zlib.so
<mightybyte>
wedens: Cool, thanks. I'll give that a try
mkoenig has quit [Ping timeout: 240 seconds]
<duairc>
mightybyte: I wasn't responding to you, but it is funny I see you in #nixos at the exact moment when I'm trying (and failing to build) snap. I didn't know you used NixOS. Do you still contribute to snap?
erictapen has quit [Ping timeout: 268 seconds]
mkoenig has joined #nixos
periklis has quit [Ping timeout: 268 seconds]
rauno has joined #nixos
fuzzy-id has quit [Remote host closed the connection]
<mightybyte>
duairc: Heh yes, that is kind of funny. I am still the primary maintainer of snap.
<mightybyte>
duairc: Usually I build snap in a "nix-shell -p zlib" shell.
erictapen has joined #nixos
<wedens>
mightybyte: actuall... I think `services.gitlab-runner.preStart = "export PATH=/run/wrappers/bin:$PATH"` has more chances to work :)
<duairc>
mightybyte: Yeah, I've seen that trick, and it works for other libraries that depend on zlib, but for some reason I still get the zlib.so error when it tries to build snap
leotaku has quit [Remote host closed the connection]
leotaku has joined #nixos
<mightybyte>
duairc: How are you trying to build snap? I think "nix-shell -p haskellPackages.snap" should work.
<duairc>
mightybyte: Well I have a big cabal project with lots of pacakges, and I'm just trying to build the whole thing with cabal... I'm not using the derivations in haskellPackages, because I'm on nixos-unstable and they're all built with GHC 8.6, but a few of my dependencies still need 8.4 so I'm stuck on that
<duairc>
Maybe there's a better way, I'm still very new to this
<ottidmes>
duairc: then just use haskell.packages.ghc844 instead
<duairc>
Oh wow, didn't realise you could just do that
<mightybyte>
duairc: Oh, then you have to declare zlib as a dependency of your package.
<duairc>
mightybyte: But it's snap-core itself which fails to build
<mightybyte>
duairc: If you get snap via nix, it will know that zlib is a dependency. But if you get it via cabal, it will not.
<duairc>
But okay, I'll try using haskell.packages.ghc844
d5[m] has left #nixos ["User left"]
<mightybyte>
duairc: Yeah, s/snap/snap-core/ in all of my above comments should be a valid substitution.
<duairc>
mightybyte: Great, I'll try that out, thanks
<mightybyte>
duairc: That's a nice general default.nix template that should build most haskell packages that don't have version dependencies incommpatible with your current version of nixpkgs.
<mightybyte>
So just put that in your project directory and do nix-build.
<duairc>
I haven't really got the hang of using nix with Haskell development yet... I guess I'll have to take that plunge now, but I haven't gotten any work done in over a week because I've been getting everything on my laptop set up and I was hoping I could just continue with the cabal way I know without having to learn more stuff just to get something done.
<mightybyte>
duairc: Once you have some kind of a nix specification of your project, running "nix-shell" will drop you into a shell where all the dependencies are available. Then you can do normal cabal builds from there.
<duairc>
mightybyte: Thanks, I really appreciate you helping me out here!
<mightybyte>
duairc: You could continue with the cabal way, but you'd have to install all the non-haskell dependencies manually (kind of like you'd have to do with macOS, other Linux distros, etc).
<ottidmes>
mightybyte: to fix your problem I think it should be: systemd.services.gitlab-runner.path = [ "/run/wrappers" ]; instead. I checked, path is defined with an apply of: apply = ps: "${makeBinPath ps}:${makeSearchPathOutput "bin" "sbin" ps}"; which if you test it has to be minus the /bin:
<ottidmes>
> let ps = [ "/run/wrappers" ]; in "${makeBinPath ps}:${makeSearchPathOutput "bin" "sbin" ps}"
<{^_^}>
"/run/wrappers/bin:/run/wrappers/sbin"
jomik has joined #nixos
<mightybyte>
ottidmes: Great thanks! I'll try that now.
myskran has joined #nixos
Dedalo has quit [Ping timeout: 244 seconds]
abathur has quit [Ping timeout: 272 seconds]
<mightybyte>
ottidmes: That seems to have fixed it! Should I open a bug about this? It seems like services.gitlab-runner should just work.
rauno has quit [Ping timeout: 268 seconds]
<wedens>
mightybyte: PR would be great
<wedens>
I may also need to use gitlab runner soon and I'd rather not remember to add this :)
<ottidmes>
mightybyte: I think it suffices to test in on your own NixOS version, just take the master version of the gitlab-runner module, place it in your configuration directory, add disabledModules = [ "services/continuous-integration/gitlab-runner.nix" ]; to it to use it instead of the one in <nixpkgs/nixos>, and import it
<duairc>
mightybyte: The default.nix file you gave me, if I have a big folder with lots of cabal packages managed by a cabal.project file, does that work for that? Or do I to have .nix files in each .cabal project?
<wirew0rm>
hooo: because it is autogenerated by nixos-generate config from the currently mounted devices and overwritten on subsequent runs. you can find it in /etc/nixos/hardware-configuration.nix which gets imported by configuration.nix
<jomik1>
I guess I need an "unstable" channel then, ye?
<ottidmes>
or if unstable is not in NIX_PATH but defined in an overlay as unstable = ...; then you would do: imports = [ (pkgs.unstable.path + "/nixos/modules/services/x11/display-managers/lightdm-greeters/enso-os.nix") ];
<wedens>
`imports = [ unstable + "/nixos/modules/services/x11/display-managers/lightdm-greeters/enso-os.nix" ]` if you don't use channels
<jomik1>
Ye okay.
<jomik1>
I normally use a fetchFromGitHub thingy
jomik has quit [Ping timeout: 244 seconds]
<wedens>
I've forgotten about braces inside expressions in lists in nix. again.
<wedens>
* around
<jomik1>
Btw, if I have an <unstable> channel, how can I then inherit from that in an overlay? If that makes sense.
<duairc>
How do I convert a string like "../package" into an absolute path?
<romildo>
I am getting the following messaging when running gnome-boxes from the terminal:
<romildo>
(gnome-boxes:5243): Boxes-WARNING **: 12:23:38.938: libvirt-broker.vala:66: Failed to update domain 'boxes-unknown-2': Failed to set domain configuration: Cannot check QEMU binary /nix/store/cxbnr3h07d527d3lp16yjc2a9l9d6zk2-qemu-3.1.0/bin/qemu-system-x86_64: No such file or directory
<jomik1>
I have `unstable = import (fetchFromGitHub {...} {}`, how would I replace that to just track `<unstable>` ?
<wedens>
import <unstable> {}
<jomik1>
Thanks :)
<duairc>
wedens: I figured it actually, thanks!
<romildo>
It seems that gnome-boxes is trying to use a qemu binary that is not installed anymore.
<romildo>
Any clues?
jomik1 has quit [Quit: WeeChat 2.2]
knupfer has quit [Remote host closed the connection]
<Taneb>
Is there a nice way to say "Someone must type the password in to decrypt this drive during boot"?
<gchristensen>
in code? or in words?
<gchristensen>
because in words, that sounds nice
<Taneb>
In code, ideally that I can put in my configuration.nix
<gchristensen>
sure, is it LUKS?
<Taneb>
No, ZFS
<gchristensen>
no idea about encrypted zfs
<gchristensen>
but with luks, if it for a password, it just ... prompts
<gchristensen>
does it not work already?
<duairc>
I'm using encrypted ZFS and it "just works" for me
<duairc>
Well, I guess that's for /, I don't know how to do it for arbitrary drives
<Taneb>
duairc: this isn't for /
<rauno>
damn got dc before, has anyone idea about that network namespace question i asked before? as is there any limitation by nixos itself ?
<duairc>
Taneb: I'd be interested to know if you figure this one out
<Taneb>
I'm configuring a fileserver, the OS is on an SSD but the files are on a bunch of HDDs with encrypted ZFS
<duairc>
I guess this is slightly different, but I think I want to make a LUKS partition that has within it two partitions: one for swap, and one containing a key, which is then used to decrypt a ZFS /
<gchristensen>
Taneb: and it doesn't already work?
morr has quit [Quit: WeeChat 2.3]
<cransom>
rauno: no, no arbitary network namespace i've ever ran into.
<ottidmes>
it seems like the sqlite3 executable wants to always open the DB in read/write mode, I tried: sqlite3 file:///nix/var/nix/db/db.sqlite?mode=ro 'SELECT COUNT(*) FROM DerivationOutputs WHERE path = "/nix/store/swzxzspyqad5qdbbpiwr20bg8p7nl0ph-test.txt";' but I still get the error: attempt to write a readonly database, guess I will try the python bindings and see if it too complains
<gchristensen>
duairc: correct. I didn't pick ZFS encryption because it isn't stable yet.
tommyangelo has joined #nixos
brejoc has quit [Client Quit]
<{^_^}>
[nixpkgs] @bjornfor opened pull request #53989 → nixos/pam: add pam_nologin.so by default → https://git.io/fhCEQ
<tommyangelo>
`nixos-rebuild` is silently failing after a couple of downloads. There are no errors, it just stops. I can't find any related issues on github. Has anyone seen this behaviour before?
<ottidmes>
tommyangelo: are you on NixOS? what version of NixOS are you on? or what version of Nix are you using?
<tommyangelo>
wedens: tried that, it just stops after a successful curl call. No errors are output
<duairc>
tommyangelo: You haven't enabled nerdfonts, have you? I was experiencing behaviour like that when I did, basically it turns out it was downloading a 2GB tarball in the background, it wasn't actually stopped
<gchristensen>
is it stopped, or is it still executing?
<tommyangelo>
ottidmes: nixos 18.03 on a Vultr VM. nix is 2.0.4
<tommyangelo>
duairc: no, this is a very minimal config on a server to run one service.
<duairc>
In the source-overrides of haskell.lib.developPackage, is there a way to get them to apply transitively to dependencies as well?
<{^_^}>
[nixpkgs] @dotlambda opened pull request #53990 → [WIP] esphomeyaml: init at 1.10.0 → https://git.io/fhCua
<tommyangelo>
gchristensen: stopped afaict. The command exits, and top doesn't show any thing
<ottidmes>
tommyangelo: then we can rule out the Nix 1.x being a problem, because a common problem when updating from 17.09 also had these symptoms, but it is apparantly something else
<teto>
trying to fetch msmtp master since it fixes a bug. I override src with fetchFromGitLab https://paste.ubuntu.com/p/p8rt7YpWsV/ yet I can't generate the proper url. It's an hosted gitlab instance. Anyone knows what's wrong ?
<jabranham>
teto: not sure, but you could just use fetchGit
samrose has quit [Read error: Connection reset by peer]
<wedens>
it's probably some old gitlab version that doesn't have api that fetchFromGitLab uses
<Taneb>
Whenever I do a nixos-rebuild switch I get "cp: cannot stat '/run/keys/initrd-ssh-key': No such filre or directory"
<{^_^}>
[nixpkgs] @FRidh merged pull request #53972 → openblas: 0.3.4 -> 0.3.5, rework a bit → https://git.io/fhc0E
<{^_^}>
[nixpkgs] @FRidh pushed commit from @dtzWill to staging « openblas: 0.3.4 -> 0.3.5, rework a bit (#53972) »: https://git.io/fhCaJ
<Taneb>
This file used to be necessary when I was trying to get an initrd ssh server, but I removed mention of that from the nixos config files and yet this error persists
<gchristensen>
sounds like you didn't completely remove it. are you sure you saved?
Guest74 has quit [Quit: Guest74]
<gchristensen>
:thinking: I wonder if you have to delete the system profile for that generation
<clever>
gchristensen: and for extra points, can you identify what this command does?
<clever>
NIX_REMOTE=local?root=/mnt/img/ nix-env --profile /mnt/img/nix/var/nix/profiles/system --set -f '<nixpkgs/nixos>' -A system -I nixos-config=/mnt/img/etc/nixos/configuration.nix --option system aarch64-linux
iqubic has quit [Ping timeout: 268 seconds]
<gchristensen>
uhhh builds a nixos system in that "remote"store? :thinking~
<clever>
gchristensen: thats 80% of it, but what about the last 3 flags?
<ottidmes>
clever: do you perhaps know if you can query the Nix store db.sqlite file without root?
<gchristensen>
I assume it uses a remote builder to make it an aarch64 system in that remote store
<clever>
gchristensen: nope, --option system convinces nix that the "local" machine is aarch64
<clever>
so builtins.currentSystem claims aarch64
<gchristensen>
oh jeeze
<gchristensen>
ok, so binfmt magic?
<clever>
exactly!
<gchristensen>
!!
<clever>
that was used to modify an rpi image to have ssh enabled by default, because the user had no way to hook up a monitor
<gchristensen>
omg!
<ottidmes>
clever: I doubt you mean literally --something, I checked the nix-store manpage and tried the options listed, but unless I missed something you cannot query for all derivers for some output, like: sqlite3 /nix/var/nix/db/db.sqlite 'SELECT drv FROM DerivationOutputs WHERE path = "/nix/store/swzxzspyqad5qdbbpiwr20bg8p7nl0ph-test.txt";'
<gchristensen>
clever: nix is sooo cool
<clever>
and without ssh, you cant do anything, and cant use it as a remote builder
<clever>
ottidmes: you can, one sec
tzemanovic has quit [Ping timeout: 252 seconds]
<teto>
wedens: I feared so. I fell back on fetchurl which works. Thanks for the help !
<clever>
cat: /nix/store/w9qswqwxbwymi1v068m8ha6yqz4cl64f-glibc-2.27.drv: No such file or directory
<clever>
ottidmes: dang, only works if the given drv exists in the store
<ottidmes>
clever: does not work, --deriver uses the deriver field, which points to _a_ deriver, different info
<clever>
yeah, and its the wrong drv
halfbit has joined #nixos
erasmas has joined #nixos
<clever>
you probably need root then, or blindly loop over every .drv, or a new command in the daemon protocol
<clever>
gchristensen: that SD card also needed a `switch-to-configuration boot` as well, but thats easy enough
<ottidmes>
clever: I inspected the database itself and --deriver is useless, it actually refers to a old wrong hash deriver that just happened to be the first to mention the output, not the one actually building the output
<clever>
gchristensen: the bigger pain-point, is that the fresh img from hydra, lacks /etc/NIXOS, and /nix/var/nix/profiles/system, so nixos-enter really doesnt like it
<clever>
gchristensen: also, due to entirely lacking a /etc, the configuration.nix for above was missing, so we had to nixos-generate-config one, and then fix it to not use grub
<gchristensen>
I don't have an /etc on boot either, what is the trouble?
cyounkins has joined #nixos
<gchristensen>
and I think nixos-generate-config doesn'at use grub for rpi?
<clever>
nixos-rebuild and friends, on a fresh aarch64 image
<clever>
gchristensen: no ssh, had to run nixos-generate-config on an x86 machine
<gchristensen>
ooh right
<samueldr>
the sd image hasn't had an activation made, AFAIUI, it's basically a copy of the closure + script to initialise the nix db at boot and activate
<clever>
samueldr: switch-to-configuration must have been ran though, for /boot/ to be configured?
<tommyangelo>
I found out what the problem was. The nix process is being OOMed
<Taneb>
OK, on booting it dumps me in a maintenance terminal and tells me I need to fix things. "zfs load-key -a; zfs mount storage/encrypted" fixes things adequately
orivej has quit [Ping timeout: 250 seconds]
<{^_^}>
[nixpkgs] @arianvp opened pull request #53991 → build-support/make-initrd: Don't derive derivation name from file name → https://git.io/fhCrY
<Taneb>
"boot.zfs.requestEncryptionCredentials = true" seemed to make no difference, and neither did "boot.zfs.forceImportAll = false"
<Taneb>
gchristensen: any clue what I might be missing?
<gchristensen>
no idea, I've never used ZFS encryption
<Taneb>
Ah, no worries
<Mic92>
Taneb: do you have one or multiple pools?
ThatDocsLady_ has joined #nixos
<Taneb>
Mic92: one pool, with one sub-dataset
<Mic92>
Taneb: is the pool your filesystem?
<Mic92>
*root fs
<Taneb>
No
<Mic92>
Taneb: is your root fs encrypted?
<Taneb>
No
<Taneb>
Although it's not the end of the world to make it so, just a bit of effort
<{^_^}>
[nixpkgs] @fgaz opened pull request #53992 → Fix some dead links → https://git.io/fhCoG
<gchristensen>
ohhhh duairc has the key on a partition
<Ankhers>
Anyone having issues building blender-2.79b on NixOS?
tzemanovic has quit [Remote host closed the connection]
shibboleth has quit [Remote host closed the connection]
tzemanovic has joined #nixos
shibboleth has joined #nixos
tzemanovic has quit [Remote host closed the connection]
<Mic92>
zfs-import-partition.serviceConfig.RequiresMountsFor= should then assigned the key location
tzemanovic has joined #nixos
<Taneb>
Mic92: thanks for your help, I'll give that a go
<Mic92>
this makes sure that systemd mounts the cryptdevice partition before starting the zfs import service
tzemanovic has quit [Remote host closed the connection]
<simpson>
Ankhers: I was able to build /nix/store/brja0pngfibd92g675qi3zwk8i3jwx1f-blender-2.79b just now. How is it not working?
<rnhmjoj>
do builder.sh scripts need a shebang?
<Mic92>
rnhmjoj: as far as I know nix starts the builder with builder.sh as an argument
<Ankhers>
simpson: A lot of undefined reference errors. One sec. I didn't realize I had limited scrollback. Let me run it again and I will post the log.
<rnhmjoj>
mic92: so it should be safe to remove...
<rnhmjoj>
it seems so
linkrage has quit [Quit: Leaving]
<duairc>
Is it possible to overlay the set of haskell packages available from inside developPackage?
Lears has quit [Remote host closed the connection]
<duairc>
overrides (in Haskell developPackage) seems to not do anything, or at least only apply to direct dependencies of the project, not transitive ones
<clever>
duairc: you want haskellPackages.override { overrides = ...
<duairc>
clever: I don't have to duplicate the override logic in multiple places? I mean technically the pkgs that will result from what you have there, pkgs.haskell.packages.ghc843 won't have that new stuff, only pkgs.haskellPackages
<clever>
duairc: you can also skip the nixpkgs config and just directly to myHaskellPackages = pkgs.haskell.packages.ghc843.override { overrides = ....
<babic>
clever: I'm not sure that's the same use case, my shared.nix is more like the lib.nix from that repo. But thanks for introducing me to that repo, it will help with other stuff I'll want to do.
<duairc>
clever: I guess I'm missing some sort of bigger picture here. Two weeks ago, (at least from a Haskell point of view), I knew nothing about Nix. I have a big repo with a bunch of cabal packages and a cabal.project file that ties them all together. I just want to be able to build and test it, but there seem to be so many different ways to do that. Cabal itself doesn't work anymore because it can't find
<duairc>
zlib.so no matter what I do. I don't know all the different approaches to Haskell in Nix fit together. I've read many things but they all seem to make each other obsolete (or not?). with your myHaskellPackages thing, where do I put that, and how do I use that to build my project?
philippD has joined #nixos
<clever>
duairc: nix can only provide a shell suitable for a single cabal file by default
<clever>
duairc: so most cabal files must be built by nix in /nix/store/ and then you get a shell suitable for developing one
<__monty__>
You can manually create an environment with a ghcWithPackages that has *all* the required packages though. As long as there's no conflicts.
<duairc>
So if I have a bunch of cabal packages that aren't on Hackage or known by Nix at all, how do I build a project that where all these packages depend on each other without repeating myself? Do I make a default.nix for each package?
<clever>
duairc: make a foo.nix for foo.cabal, using cabal2nix, then callPackage all of those inside the overrides = ... block of a haskell package override
<clever>
so when foo.cabal depends on bar, it will look for a bar attr in that set
Makaveli7 has quit [Ping timeout: 244 seconds]
<duairc>
clever: Okay, but I the thing I'm missing is, where does the "haskell package override" go, how do I actually use it? You have it in config.nix, how does that get used by nix-build or nix-shell or whatever?
<{^_^}>
[nixpkgs] @Mic92 pushed 3 commits to release-18.09: https://git.io/fhCXJ
<clever>
duairc: or you can just directly use the return value of something like pkgs.haskell.packages.ghc843.override { overrides = ....
<duairc>
Okay, well one thing I wasn't getting before was that this gets passed to nixpkgs as the config parameter, that makes sense, I think
<duairc>
I'm a little bit closer now
<clever>
duairc: if the config param of nixpkgs is missing, it will automatically load ~/.nixpkgs/config.nix
<duairc>
Wow, I never knew that
<duairc>
Is that not ~/.config/nixpkgs instead of ~/.nixpkgs? Or are they interchangable?
tzemanovic has joined #nixos
<clever>
duairc: ~/.nixpkgs is the old one, it has since moved to ~/.config/nixpkgs/
<clever>
but it will look for both
<duairc>
That makes sense, just checking
<clever>
it will even look for $NIXPKGS_CONFIG which has a higher priority
knupfer has quit [Remote host closed the connection]
<clever>
which defaults to the non-existant file /etc/nix/nixpkgs-config.nix
cyounkins has quit [Remote host closed the connection]
cyounkins has joined #nixos
<duairc>
clever: I think I've just about put all these pieces together now... or at least I can see how to get something that will work, it might not be pretty at least it's a start
<duairc>
Thanks, I really appreciate it
tommyangelo has quit [Quit: Leaving]
pSub has joined #nixos
pSub has joined #nixos
pSub has quit [Changing host]
<{^_^}>
[nixops] @PsyanticY opened pull request #1079 → EC2_backend: Fix mispelled function → https://git.io/fhC1J
<{^_^}>
[nixops] @AmineChikhaoui pushed 2 commits to master: https://git.io/fhC1Z
<Mic92>
Taneb: the root fs should be mounted first in any case.
ddellacosta has joined #nixos
<Mic92>
Taneb: I will mention the deficiency in the wiki so.
sinner has joined #nixos
sinner is now known as Guest39310
hph^ has quit []
Makaveli7 has joined #nixos
nst^ has joined #nixos
Guest39954 has quit [Ping timeout: 268 seconds]
Jackneill has joined #nixos
<wedens>
clever: I've noticed there are two stack.yaml files in cardano-shell repo, and they use two different `shell-file` values, what's the point?
orivej has quit [Ping timeout: 268 seconds]
techieAgnostic has quit [Quit: ciao]
tA- has joined #nixos
<{^_^}>
[nixpkgs] @pmahoney opened pull request #53996 → openfst: relax platform requirement to include darwin → https://git.io/fhCD1
dsiypl4 has joined #nixos
orivej has joined #nixos
steve-chavez has joined #nixos
<steve-chavez>
Hello, how can I get the sha256 of a .deb file?
humanoyd has joined #nixos
waleee has joined #nixos
humanoyd is now known as Guest14411
<wedens>
,tofu steve-chavez
<{^_^}>
steve-chavez: 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.
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
orivej has quit [Ping timeout: 272 seconds]
<steve-chavez>
wedens: Thank you
cyounkins has joined #nixos
Guest14411 has left #nixos ["WeeChat 2.3"]
<Taneb>
Mic92++
<{^_^}>
Mic92's karma got increased to 7
<Taneb>
Mic92: you've helped me figure this out, thank you
<__monty__>
Is nix-lang untyped or dynamically typed?
waleee has quit [Ping timeout: 272 seconds]
nst^ has quit []
<gchristensen>
is it not strictly typed, since you have to toString things?
<infinisil>
__monty__: dynamically
kisonecat has joined #nixos
<infinisil>
Only once the code gets evaluated, the types are checked, so it's not static
<gchristensen>
huh?
winem_ has quit [Ping timeout: 272 seconds]
<Mic92>
that's true.
<infinisil>
> if true then 1 + 2 else 1 + "foo"
<{^_^}>
3
<__monty__>
And what types are there? String, function, list, record?
<infinisil>
and a bunch of other primitives
<infinisil>
int, float, null
<infinisil>
> builtins.typeOf null
<{^_^}>
"null"
<gchristensen>
how does laziness make it dynamic?
<__monty__>
gchristensen: In a statically typed lazy language that wouldn't be legal.
<clever>
gchristensen: there are some nixos options that accept true, false, and "relaxed" for example
<gchristensen>
so what does it make it?
<__monty__>
Dynamically typed : )
<Mic92>
gchristensen: you can assign types to function
<Mic92>
*cannot
<gchristensen>
what would it be if you don't have to use toString?
<simpson>
Mic92, __monty__: Careful not to conflate Nix's type system, which assigns types to values, with any underlying type theory for Nix, which surely is degenerate.
<clever>
there is a single global table, that maps the strings to ints
<clever>
so all set lookups are based on int
<__monty__>
Huh, unexpected.
<__monty__>
Kinda surprised that works for nixpkgs without problems. I guess most keys repeat though.
<clever>
certain keys like "name" are also pre-allocated early in the nixexpr startup
<clever>
and the int behind it is stored into a global variable
<clever>
so you can then lookup the "name" of a derivation, without ever doing strcmp("name",foo)
<simpson>
__monty__: Eh. Sort of; we might demand that the relationship from the syntactic type system to the semantic type theory is functorial. But that's no problem here; all we're really saying, then, is that Nix's type system has features and restrictions which don't manifest in the type theory. (This dovetails nicely with the observation that Nix's values are only distinctly-typed inside Nix, and not outside Nix
<simpson>
in the interpreter.)
<clever>
__monty__: and also, all lists require the key/value pairs to be sorted by the key as an int
<Taneb>
Plan for tomorrow is to migrate my work's Hydra server onto the fileserver I was setting up today and last week
<__monty__>
clever: But they're not array-backed?
olto has joined #nixos
olto has quit [Remote host closed the connection]
olto has joined #nixos
<Mic92>
__monty__: lists is stored as a c++ vector if that is your question.
<__monty__>
Mic92: Then what did clever mean by key/value pairs are sorted by key? Arrays/vectors are always "sorted" by key/index, no?
olto has quit [Remote host closed the connection]
<Mic92>
__monty__: I think attribute sets are sorted by key
olto has joined #nixos
slack1256 has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
boredom101 has quit [Ping timeout: 256 seconds]
o1lo01ol1o has joined #nixos
<__monty__>
Ah, ok.
ThatDocsLady_ has quit [Ping timeout: 258 seconds]
<rauno>
Hum, anyone here has experience with running mesos on nixos ?
freeman42x has joined #nixos
dermetfan has joined #nixos
<slack1256>
What is the command to download paths and derivations to /nix/store?
<{^_^}>
[nixpkgs] @Profpatsch pushed commit from @NinjaTrappeur to master « bazel: fix python stub paths. »: https://git.io/fhCQZ
<gchristensen>
WhittlesJr: probably none
<gchristensen>
ZFSoL isn't going anywhere. too many billions of dollars worth of data stored on it.
samrose has joined #nixos
<clever>
WhittlesJr: as for why FPU stuff is special in kernel, most of the context-switching in the kernel assumes the FPU is not in use, so it doesnt need to backup and restore the FPU registers (huge) when context switching
<clever>
so you need a special function to save&restore it, to not clobber whatever userland proc was using the FPU last
<gchristensen>
the current thinking is adding a shim to SPL
<WhittlesJr>
That's reassuring, thank you. People have been saying things like "well I'd rather switch OSs than switch file systems" so I got spooked.
witchof0x20 has joined #nixos
elgoosy has joined #nixos
juliendehos has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @veprbl pushed commit from @fgaz to master « maintainers: update fgaz's email address »: https://git.io/fhC5k
orivej has joined #nixos
<simpson>
WhittlesJr: And yes, people are stubborn and hate admitting that they chose poorly when it comes to filesystems.
<__monty__>
simpson: Are you saying people shouldn't run ZFS? (Not trying to start a flamewar. Just want to find out why if I understand you correctly.)
<simpson>
__monty__: I'm saying that people should pay attention to the licenses for the code that they're using, and also that people who chose ZFS have been constantly warned that the license on the existing code is problematic.
<simpson>
"every filesystem sucks", to borrow a phrase.
dsiypl4 has joined #nixos
bsima has joined #nixos
romildo has joined #nixos
<WhittlesJr>
simpson: Yeah it's too bad about all the politics involved... I've been really happy with ZFS from a features+performance standpoint. I agonized for quite a while trying to choose between zfs and btrfs and finally went with zfs because it seemed more stable and featureful. I hope everything works out but if I end up having to switch filesystems then that's what I'll do.
<ivan>
why does my services.nix-serve error with nix-daemon[8915]: error processing connection: user 'nix-serve' is not allowed to connect to the Nix daemon
<ivan>
I don't need to make it a trusted user, do i?
<gchristensen>
simpson: the license incompatabilty is only in distribution, though.
<__monty__>
gchristensen: Doesn't that mean nixpkgs could no longer provide zfs packages?
<simpson>
gchristensen: Right.
<gchristensen>
no, it can't provide a combined kernel + zfs package
<nek0>
My dear bretheren, I am in peril. I am trying to update monodevelop, because I might need it in the foreseeable future. My build process comes to a halt with errors: "Error: NameResolutionFailure". Does somebody have some experience in the field?
Dagger2 has joined #nixos
<rauno>
how does substituteInPlace --subst-var-by mount work with @mount@ vars in source files?
<gchristensen>
__monty__: we can distribute a compiled kernel, and a compiled zfs, as long as you mush them together on your own.
<simpson>
WhittlesJr: Meh. The ZFS-on-Linux meme-pushers have had a good decade to document the ZFS core, which supposedly was only "10kloc" of "simple" code, and reimplement it in-kernel on top of VFS with a GPL license. They just didn't think they'd ever have to, apparently.
Mateon3 has joined #nixos
<simpson>
Of course, it'd be imprudent to ignore that many ZFS fans *like* the idea of skipping VFS, md, etc.
<gchristensen>
meme
<samueldr>
might I suggest moving the ZFS licensing discussion to ↘️ #nixos-chat, to ensure questions aren't lost through it?
<__monty__>
rauno: The @ wrapping is just the cue for what things get substituted.
<samueldr>
rauno: the patch contents has no way to know where the resulting binaries will be
<samueldr>
and it needs to (should) point to an absolute path
sanscoeur has quit [Read error: Connection reset by peer]
<__monty__>
It's like ${these} for string interpolation.
sanscoeur has joined #nixos
<rauno>
ah okay :)
<symphorien>
nek0: in the build sandbox, network access is forbidden
<nek0>
symphorien: ahh. thanks.
MP2E has joined #nixos
grumble is now known as AMENDMENT_EFFFFF
AMENDMENT_EFFFFF is now known as grumble
mtesseract has joined #nixos
shibboleth has quit [Quit: shibboleth]
ubert has quit [Quit: Leaving]
dermetfan has quit [Quit: WeeChat 2.3]
__red__ has joined #nixos
<__red__>
I have an application that has to be compiled using gcc-4.9 - is there a way to get nix-shell to give me a -p stdenv that includes that compiler?
<__red__>
or how else would you recommend I approach it>?
mtesseract has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
kvda has joined #nixos
kvda has quit [Client Quit]
steve-chavez has quit [Quit: Page closed]
shibboleth has joined #nixos
kisonecat has quit [Ping timeout: 268 seconds]
_kwstas has joined #nixos
kisonecat has joined #nixos
halfbit has quit [Ping timeout: 250 seconds]
<__red__>
simpson, symphorien: Is there a more generic version of overideCC? I need to also override texinfo
<symphorien>
how ?
halfbit has joined #nixos
<__red__>
I'm compiling old software, the the latter versions of the texi files contain stuff that's considered a syntax error in modern versions
doyougnu has joined #nixos
<__red__>
this kind of stuff: "warning: @itemx should not begin @table"
<symphorien>
so what would you like to override ?
Guanin has joined #nixos
<__red__>
the version of texinfo in stdenv
<symphorien>
does texinfo belong to stdenv ?
<symphorien>
otherwise, nixpkgs has texinfo4
<__red__>
lemme dig and double-check
<symphorien>
so add texinfo4 to buildInputs
TheAppleMan has joined #nixos
<__red__>
ah, ioh - this is pollution from my environment
<symphorien>
you can run nix-shell --pure
<__red__>
oh man, I wish I knew about that years ago
<__red__>
thank yo :-)
<__red__>
yoU
elgoosy has quit [Remote host closed the connection]
yayforj has joined #nixos
_kwstas has quit [Quit: _kwstas]
_d0t has joined #nixos
kisonecat has quit [Ping timeout: 268 seconds]
knupfer has joined #nixos
WhittlesJr has quit [Quit: Page closed]
Ariakenom has joined #nixos
shibboleth has quit [Quit: shibboleth]
civodul has joined #nixos
TheAppleMan has quit [Remote host closed the connection]
TheAppleMan has joined #nixos
TheAppleMan has quit [Remote host closed the connection]
TheAppleMan has joined #nixos
TheAppleMan has quit [Remote host closed the connection]
TheAppleMan has joined #nixos
fendor has joined #nixos
TheAppleMan has quit [Remote host closed the connection]
fusion809 has quit [Remote host closed the connection]
THFKA4 has quit [Ping timeout: 252 seconds]
TheAppleMan has joined #nixos
myskran has quit [Quit: myskran]
THFKA4 has joined #nixos
pSub has quit [Quit: pSub]
TheAppleMan has quit [Remote host closed the connection]
lord| has quit [Read error: Connection reset by peer]
lord| has joined #nixos
slack1256 has quit [Remote host closed the connection]
Celmor has joined #nixos
<Celmor>
is it possible to install a package in a livecd session?
<gchristensen>
sure
doyougnu has quit [Ping timeout: 258 seconds]
<Celmor>
`nix-env -qa` didn't list what I looked for, so I run `nix-channel --update`, another nix-env errors with 'This version of Nixpkgs requires Nix >= 2.0, please upgrade' so I tried `nixos-rebuild switch` which then errors with ' The ‘fileSystems’ option does not specify your root file system.'
<infinisil>
Oh, that's pretty bad
<gchristensen>
when did you get this livecd?
olto has quit [Quit: switching offline]
<{^_^}>
[nixpkgs] @8084 opened pull request #54000 → init: create_ap at 0.4.6 → https://git.io/fhWea
<Celmor>
2018-02-02
<gchristensen>
aye, probably best to fetch a new one
<gagbo>
I'm wondering, do you guys have a separate partition for /home or do you just put everything in /, assuming that anyway /nix/store separates everything non-user from the rest ?
<hyper_ch>
boot.zfs.enableUnstable = true;
<hyper_ch>
gagbo: I don't run sepearte /home
<hyper_ch>
but could easily create a dataset in zfs for it if I wanted
<gagbo>
infinisil: hyper_ch ok thanks for the input guys
<iqubic>
Is there a way to update a package with nix, while not waitin for the 6 month stable release? For some reason the chat application Discord has decided that it wants me to update right now, before I can use this application.
<gchristensen>
iqubic: the discord update is already backported, going through the release process
babic has quit [Ping timeout: 268 seconds]
<hyper_ch>
Celmor: you need to remove line 8 adn lines 84 - 100
<iqubic>
Ah. Good. I didn't know that.
<hyper_ch>
Celmor: I use that to add some custom files to the live iso /root folder for almost automated zfs install
<iqubic>
How long do you think it will be before it hits stable?
<iqubic>
Or can I just fetch it from unstable right now?
_kwstas has quit [Client Quit]
<hyper_ch>
Celmor: also line 21 set to your keyboard layout
<infinisil>
Yeah, placeholder is a Nix 2.0 feature.. And people *really* wanted to use this in nixpkgs asap, so now it's incompatible with too old nix versions..
<infinisil>
gchristensen: Could you possibly run ofborg with the minimum Nix version as specified in lib/minver.nix such that we can prevent such breakages before they get merged?
o1lo01ol1o has quit [Remote host closed the connection]
<dmj`>
hey johnw :)
freeman42x has quit [Ping timeout: 252 seconds]
<infinisil>
Celmor: Try `-A nixUnstable` instead of `-A nix`, worth a try, maybe the newer version is better
<Celmor>
infinisil, you mean for tmpfs or does it actually require memory? couldn't I mount a network share?
samrose has quit [Ping timeout: 245 seconds]
<infinisil>
Yeah it needs memory, you could use swap though
samrose_ has joined #nixos
o1lo01ol1o has joined #nixos
<symphorien>
it can be that $TMPDIR is in /run and is too small
<dmj`>
johnw: long time no talk :)
_kwstas has joined #nixos
freedan42x has joined #nixos
ma27 has quit [Quit: WeeChat 2.2]
Judson has joined #nixos
vyorkin has joined #nixos
vyorkin has quit [Client Quit]
ma27 has joined #nixos
<Judson>
I know already that I got up on the wrong side of the bed today, but: nix-env -iA nixpkgs.jira-cli #-> error: jira-2.0.0 not supported for interpreter python2.7
<Judson>
My gut reaction is that this is what comes of repo-scoped mirroring of the upstream package repos.
* Judson
grumps
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<{^_^}>
[nixpkgs] @dywedir pushed commit from @P-E-Meunier to master « toml2nix: init at 0.1.1 (#53883) »: https://git.io/fhWTS
<infinisil>
Judson: My guess is that something just got updated in the dependency tree of jira, and nobody bothered to check whether jira still works with it
<infinisil>
Judson: `nix-env -iE 'pkgs: pkgs.jira-cli.override { pythonPackages = pkgs.python3Packages; }'` should work
<infinisil>
(as in, some dependency deprecated support for python 2)
ddellacosta has quit [Remote host closed the connection]
Dedalo has joined #nixos
protocol has quit [Quit: WeeChat 2.2]
ng0 has quit [Quit: Alexa, when is the end of world?]
LogicAside has quit [Quit: ZNC 1.6.5+deb1+deb9u1 - http://znc.in]
vk3wtf has joined #nixos
<infinisil>
Celmor: Looks like it worked :)
<Dedalo>
I have a strange issue. I have a new X1 Carbon 6th and the battery charge doesn't go over 79%. It used to charge completely till two days ago, suddenly it doens't anymore. I have tried, today. to fully discharge the battery, but it's still not charghing more than 79%. I can't understand why. Is there a way to fix it? I know that some of you are using the same laptop, it may happened to you as well.
<benley>
Dedalo: I have one of those too. Have you updated the firmware on it?
<infinisil>
That can just happen when the battery gets a bit old
<benley>
iirc they changed some power related EC things in a recent firmware release
<benley>
it can, but these laptops only came into existence like a year ago
<infinisil>
Oh
<Dedalo>
benley: yes I did, I have used Windows just for that, I have updated the firmware, then I have removed Windows and installed NixOS.
knupfer has quit [Remote host closed the connection]
<benley>
k cool
tzemanovic has quit [Remote host closed the connection]
<Dedalo>
tilpner: I'm using that already, it's imported in my configuration.nix
Ariakenom has quit [Read error: Connection reset by peer]
<benley>
tilpner: wow, I've never seen that handful of parameters before
<Celmor>
infinisil, and bootable, thanks for your help
<Dedalo>
benley: the laptop is new, it has about 3 weeks
<tilpner>
So you told it to stop at 80%, and now wonder it does what you said? I don't use tlp anymore, but I think that's what that means
<tilpner>
*wonder why it does
<{^_^}>
[nixpkgs] @Infinisil opened pull request #54006 → hackage2nix: Add megaparsec < 7.0 for idris → https://git.io/fhWIE
<Celmor>
though I haven't been able to getting a generated nixos livecd to boot after writing to a flash drive
<gchristensen>
infinisil: it is complicated by needing to support the previous stable too
eadwu has joined #nixos
<tilpner>
Read the linked article, it seems to be a measure of improving lifetime
<infinisil>
gchristensen: Context?
tzemanov_ has joined #nixos
<benley>
those tlp thresholds are cool but holy cow, I don't think I would set those as defaults for anybody without making them understand what was happening first
<gchristensen>
21:24 <infinisil> gchristensen: Could you possibly run ofborg with the minimum Nix version as specified in lib/minver.nix such that we can prevent such breakages before they get merged?
<Dedalo>
tilpner: oh shit, I didn't notice that, I have included because I thought it was necessary
<tilpner>
You should probably read things you put in your config :)
<infinisil>
gchristensen: Hmm I see, I have this idea of putting some expression into nixpkgs that builds you a Nix of the minimal version, so you can say `(import <nixpkgs> {}).nixMinimal`. This should optimally be compatible with a lot of previous Nix versions
<Dedalo>
tilpner: you right, I should have read it. Indeed, do you think is going to save battery life?
<infinisil>
So the dependency graph shouldn't use placeholder, etc.
<tilpner>
Dedalo: Read the linked article. I don't know, I have no thinkpads
anderslundstedt has joined #nixos
<infinisil>
gchristensen: Then tools like ofborg could use this instead of their own Nix
<infinisil>
More purity!
<gchristensen>
ofborg couldn't use that
<gchristensen>
it is easy to imagine a PR which changes nixMinimal to my-cool-malware
<Celmor>
is zfs version of 0.7.0 normal in a newly built image?
<Dedalo>
I heard that kernel 5.0 is dropping ZFS support, is that true?
<gchristensen>
no
<clever>
Dedalo: the problem feature is only used to make checksums run faster, and the optional thing can just be turned off
<gchristensen>
the kernels has never supported ZFS
<simpson>
Dedalo: It's not even wrong; Linux didn't have ZFS support before.
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #53996 → openfst: relax platform requirement to include darwin → https://git.io/fhCD1
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fhWIx
shpx has joined #nixos
<__monty__>
Better add an entry to the knowledge base. Can imagine that question'll come up more : )
LogicAside has joined #nixos
dermetfan has joined #nixos
<gchristensen>
,zfsol = Linux Kernel 5 made a minor API change which people are saying is breaking ZFS. It'll be fine. ZFS isn't going anywhere. https://github.com/zfsonlinux/zfs/issues/8259
<{^_^}>
zfsonlinux/zfs#8259 (by tonyhutter, 6 days ago, open): Linux 5.0: _kernel_fpu{begin,end} no longer exported
<Dedalo>
clever: how do I turn it off?
<gchristensen>
Dedalo: youdon't have to do anything.
<clever>
Dedalo: i expect the nixpkgs expression for zfs to do that for you, when the update comes around
yayforj has quit [Ping timeout: 268 seconds]
<infinisil>
gchristensen: The usual abbreviation is ZoL btw
<Dedalo>
clever: good to know because I would like to use ZFS even with the new kernel :-)
<infinisil>
Ohhh
<infinisil>
Yeah I forgot I didn't fix that yet..
* infinisil
goes to fix it
<thoughtpolice>
Hrm, is there an easy way to get the closure of all the things that will be inside nix-shell? Like, if I do 'nix-shell shell.nix' or 'nix-shell -A foo', how can I get the closure of all the build deps?
nst^ has joined #nixos
<{^_^}>
[nixpkgs] @jtojnar opened pull request #54007 → Darwin GSettings support → https://git.io/fhWLl
gagbo has quit [Quit: I'm out !]
<thoughtpolice>
It's sort of a weird case. I don't actually have a default.nix, just a shell.nix I use to do some iterative development. So the expression isn't using stdenv.mkDerivation, it uses pkgs.runCommand
gagbo has joined #nixos
<thoughtpolice>
So for right now I just have this weird hack that just takes all the buildInputs and writes them to $out/nix-support/propagated-build-inputs to keep things clean
<thoughtpolice>
Then `nix path-info -r $(nix-build --no-link shell.nix)` will do the trick. Is there an easier way?
<thoughtpolice>
My intent is basically that I want to put a cache for my shell.nix somewhere
<thoughtpolice>
And people can just use the trusted-substituters option at nix-shell time
<thoughtpolice>
So it just pulls in the *closure* of things for building, but the actual "environment" is still created by the user
<thoughtpolice>
(This is because my shell.nix contains stuff out of tree, so it's to avoid building)
<clever>
thoughtpolice: whats wrong with just running a hydra that hosts the result of the full build (and anything hydra had to dl to do so)
<jabranham>
In the systemd journal ("journalctl -b -p 3"), I have "systemd-modules-load[534]: Failed to find module 'snd_pcm_oss'". Is this cause for alarm?
<Guanin>
That seems audio related - if everything works related to sound, it's probably fine I'd guess
<LnL>
gchristensen: thoughtpolice: put env-vars in the output, that should keep about everything alive
<{^_^}>
[nixpkgs] @palfrey opened pull request #54008 → libnetfilter_acct: init at 1.0.3 → https://git.io/fhWt2
<Guanin>
jabranham, I just noticed that I have the same error, but never noticed it
<jabranham>
Guanin: yeah, my sound is working just fine. I just don't like having weird log messages :-)
<LnL>
gchristensen: thoughtpolice: alltho that's kind of an implementation detail, runCommand "foo" {} "cp env-vars $out"
<Guanin>
jabranham, it might be related to https://nixos.org/nixos/options.html#sound.enableossemulation ? At least, snd_pcm_oss seems to be a module that provides OSS emulation
<Guanin>
jabranham, but it seems like the default kernel config (at least for unstable.linuxPackages_4_20) does not have OSS emulation enabled, so the module is not compiled (as far as I understand)
<jabranham>
Guanin: ah, so this likely fails for everyone, not just us
<Celmor>
gchristensen, what's the difference in version?
<Guanin>
jabranham, I'd guess, but it would be nice to know whether other have the same problem :)
<gchristensen>
sorry. NixOS's isos ship ZFS Stable by default. the config you pasted uses ZFS Unstable.
<gchristensen>
what are the differences? I dunno
winem_ has quit [Ping timeout: 268 seconds]
dsiypl4 has quit [Ping timeout: 268 seconds]
<Celmor>
seems the version reported by modinfo before and after rebuild (about a year later) is the same
<rain1>
To compile C code I go nix-shell -p autoreconfHook
<rain1>
but how can I also get zlib available?
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<monokrome>
w/ a 500GB M.2 PCIe Samsung and a 1TB SATA SSD
<monokrome>
and 32GB RAM
<monokrome>
(if that matters?)
tzemanov_ has quit [Remote host closed the connection]
<clever>
freedan42x: can you pastebin the full output?
<freedan42x>
clever: actually it's the full output
tzemanovic has joined #nixos
<clever>
freedan42x: did you give the right path to adga?
tzemanovic has quit [Remote host closed the connection]
<freedan42x>
clever: yes, as `whereis agda` says
tzemanovic has joined #nixos
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
<{^_^}>
[nixpkgs] @benley opened pull request #54014 → pythonPackages.jq: init at 0.1.6 → https://git.io/fhWOd
tzemanovic has quit [Client Quit]
halfbit has quit [Ping timeout: 272 seconds]
<clever>
freedan42x: run realpath on that, then try again
jackdk has joined #nixos
<clever>
freedan42x: it must be the path that had adga in it, not the user-environment
<freedan42x>
clever: ugh, it's path to agda: /nix/store/...-user-environment/bin/agda
<clever>
freedan42x: thats just the user-environment, run realpath on that
gagbo has quit [Ping timeout: 258 seconds]
<freedan42x>
clever: Oh, sorry, I didn't understand what you meant, here's the realpath: "/nix/agda"
<clever>
that doesnt look like a valid path
<symphorien>
Isn't agda installed in configuration.nix ?
<freedan42x>
symphorien: no
ddellacosta has joined #nixos
gagbo has joined #nixos
ddellaco_ has joined #nixos
<freedan42x>
clever: thank you, anyway
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<clever>
freedan42x: can you give the full path without censoring out the hashes?
<freedan42x>
clever: "/nix/store/a53ja2a8jflrhzygdxw8gmrpl3mxfy1r-user-environment/bin/agda", also, when I'm trying to CD to that folder it says that it's not a directory
<clever>
freedan42x: and "realpath /nix/store/a53ja2a8jflrhzygdxw8gmrpl3mxfy1r-user-environment/bin/agda" prints what?
<freedan42x>
clever: now nix-env -e Agda; nix-store --delete /nix/store/nl7rc8mnavz6j6a955c28p5krx4hj758-Agda-2.5.4.1 ?
<clever>
nix-env -e was already done
<freedan42x>
clever: it have deleted, thank you!!
<{^_^}>
[nixpkgs] @tbenst opened pull request #54016 → cmtk: init at 3.3.1 → https://git.io/fhWsf
worldofpeace has joined #nixos
<kini>
Dumb question, but a comment in the autogenerated /etc/nixos/configuration.nix says "You should change this [referring to system.stateVersion] only after NixOS release notes say you should." What exactly should I be looking for in the release notes? Just the fact that there are already release notes for the version I want to set system.stateVersion to?
janders223 has joined #nixos
<jasongrossman>
kini: It's not a dumb question. The naming of that variable causes very frequent questions here. The answer is that you shouldn't change it.
freedan42x has quit [Quit: Leaving]
<simpson>
kini: If you don't see anything interesting to you in the release notes, then don't change `stateVersion`. That's it.
<simpson>
It's a field for doing stuff that could break folks. By *not* changing it, folks have less broken stuff.
<janders223>
after frustrations of _not_ getting multi-user to work on macOS 10.14 I reverted back to a single user install. I installed nix-darwin and home-manager, now any calls to nix-env result in `error: cannot import '/nix/store/7kn4vp8nfpa3qrqbfdmw2mnyz05pnd5h-source', since path '/nix/store/v1p99faf8p9m6ksyrdich95qcv3sp8c2-source.drv' is not valid, at
<kini>
OK. I feel like the wording of that comment could be improved, then. How does this sound? "You should not change this unless a Nixos release notes entry explicitly tells you to."
<kini>
*NixOS
erasmas has quit [Quit: leaving]
rfold has joined #nixos
<rfold>
Hi, I'm using builtins.listToAttrs, and it seems to add override and overrideDerivation keys to the attrset returned. Is this expected behavior?
<janders223>
and I'm an idiot, forgot to update, solved everything
<jasongrossman>
,stateversion
<{^_^}>
jasongrossman: Did you mean stateVersion?
<{^_^}>
Setting stateVersion to the latest release doesn't upgrade anything and can only break your setup at best. To actually upgrade NixOS see https://nixos.org/nixos/manual/#sec-upgrading. If you want to update the stateVersion option regardless, Ctrl-F for "stateVersion" in https://nixos.org/nixos/manual/release-notes.html to see things that need to be manually migrated with the new value.
<simpson>
kini: I suppose that the notes could say "don't change this" or something less ambiguous, but then we'd be in the position of having a rule that people know better than to follow.
<jasongrossman>
samueldr: IIRC you had a really good idea about what to rename stateVersion to.
<c00w>
dataFormatVersion?
<samueldr>
stateEpoch, with numbers unrelated to the version
<c00w>
+1 to unrelated numbers.
<samueldr>
the idea was 100 so 18.09 is still smaller than 100
<kini>
well, all I'm saying is to change it from "you should change this only when ____" to "you should not change this except when ____"; semantically it's the same, but it foregrounds that you usually don't want to be doing it
romanofskiWork has quit [Ping timeout: 245 seconds]
rfold has quit [Quit: leaving]
Denommus has left #nixos ["ERC Version 5.3 (IRC client for Emacs)"]
romanofskiWork has joined #nixos
shpx has quit [Quit: shpx]
Guanin has quit [Remote host closed the connection]
shpx has joined #nixos
<jasongrossman>
samueldr: Oh, yes. Thanks. I don't know how to do a PR for this. Maybe I should learn how.
<samueldr>
I have one opened already
<jasongrossman>
samueldr: Oh, fantastic.
<samueldr>
(too bad, you should learn how to!)
<jasongrossman>
kini: I hear you. I think that's a good idea as well. I don't think it's enough on its own, but the process will decide.
<jasongrossman>
samueldr: I need to learn to write derivations. It's on my list. Not at the top of the list, I'm afraid!
eadwu has quit [Ping timeout: 252 seconds]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
jabranham has quit [Quit: ERC (IRC client for Emacs 27.0.50)]