LarryTheCow has left #nixos ["ERC (IRC client for Emacs 26.3)"]
ssk has joined #nixos
<ssk>
I'm trying to debug a problem with the liblapack derivation and I seemed to have borked my /nix/store or something?I do nix-shell -E '(import <nixpkgs> {}).pkgs.liblapack' and then in the shell I do genericBuildand I get a bunch of these type of errorscp: cannot create regular file
<ssk>
'source/cncz072a1rm5vhbcpm5fl85i4gz82dk2-source/meson.build': Permission deniedcp: cannot create regular file 'source/cncz072a1rm5vhbcpm5fl85i4gz82dk2-source/meson_options.txt': Permission denieddo not know how to unpack source archive /nix/store/cncz072a1rm5vhbcpm5fl85i4gz82dk2-sourceand I swear this wasn't happening all of 5 minutes agoshould I
<ssk>
just rm that source in the nix store?will it mess more things up?
<ssk>
I'm trying to debug a problem with the liblapack derivation and I seemed to have borked my /nix/store or something? I do `nix-shell -E '(import <nixpkgs> {}).pkgs.liblapack'` and then in the shell I do `genericBuild` and I get https://gist.github.com/satshabad-cr/cf32be7d2f3d164e2c43c0734645dc20
<ssk>
i swear this wasn't happening 5 minutes ago
<ssk>
Should I remove that source from the nix store manually?
<cole-h>
bbigras: btw, waybar PR looks good now. I see "(process:1623): playerctl-WARNING **: 17:37:47.497: Playback status not set on track change; getting status from interface instead" in the output, dunno if that's something you can deal with or not
<morgrimm_>
Has anyone figured out any Github Actions to kick off Hydra jobs yet, that anyone knows of?
<bbigras>
cole-h: ok thanks, because spotifyd doesn't support sending tracks changes. I started a crappy PR for that. I'm at 4 pr just to get Spotify in my waybar. Thanks I'll take a look
<cole-h>
bbigras: And it's only a warning, so if you can't fix it, I don't think that's a big deal.
<gchristensen>
ConditionACPower=true is great to prevent a service from starting while on battery. is there a way to make a service *stop* when ConditionACPower=false?
dsx_loves_you has quit [Quit: dsx_loves_you]
cosimone has quit [Quit: Quit.]
pareidolia has quit [Ping timeout: 256 seconds]
<infinisil>
gchristensen: My only idea is to have another oneshot service for the false condition that stops it :/
xcmw has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
simukis_ has quit [Quit: No Ping reply in 180 seconds.]
<gchristensen>
hrm :/ I wish znapzend ran on a schedule and not a daemon
pareidolia has joined #nixos
simukis_ has joined #nixos
reallymemorable has joined #nixos
rogue_koder has quit [Ping timeout: 260 seconds]
rogue_koder has joined #nixos
xcmw has joined #nixos
<Gaelan>
Does the NixOS install ISO include the Goddamn Broadcom Wifi Drivers(TM)?
<{^_^}>
[nixpkgs] @eraserhd opened pull request #88113 → gambit: disable use of epoll() on Darwin → https://git.io/Jfuce
proofofkeags has quit [Remote host closed the connection]
morgrimm_ has joined #nixos
morgrimm has joined #nixos
proofofkeags has joined #nixos
morgrimm_ has quit [Ping timeout: 246 seconds]
morgrimm has quit [Ping timeout: 246 seconds]
<hpfr[m]>
I'm still getting the p11-kit cert error with flatpak. I'm on latest nixos-unstable, so I think this shouldn't be happening? How can I debug this?
thc202 has quit [Ping timeout: 256 seconds]
proofofk_ has joined #nixos
willghatch[m] has joined #nixos
proofofkeags has quit [Ping timeout: 258 seconds]
proofofk_ has quit [Remote host closed the connection]
andreas303 has quit [Remote host closed the connection]
felixfoertsch has quit [Ping timeout: 272 seconds]
felixfoertsch23 is now known as felixfoertsch
andreas303 has joined #nixos
tomberek has joined #nixos
<tomberek>
c00w: hey, can you clarify your comments about issue #84826, i want to make sure i understand what you mean wrt using buildGoModule and private modules
<pistache>
is there a reason why packages that use wrapProgram prefer to --prefix the PATH rather than --suffix ?
<pistache>
I want to submit a PR for the mpv package to allow it to use a youtube-dl from the user's path (it's currently not possible because of --prefix)
<pistache>
but from a quick look in nixpkgs, most PATH wrappers use --prefix rather than --suffix, so I fear that there's something that I'm missing
<pjt_014>
If I want to suggest that a package be replaced with a fork of itself, which kind of issue do I file? It's not a new package req, and it's not a bug really...
<pjt_014>
hyper_ch: It's p7zip. The original on sourceforge has been abandoned for 3+ years and has security vulns, while this fork has fixed them and was last updated this week.
<quinn>
pjt_014: i mean an 'emacs mode' is basically just an emacs plugin that is a little bit more cleverly hooked and scoped.
<pjt_014>
I assume they typically aren't made by including the entirety of emacs as a dep?
<quinn>
pjt_014: generally a program having an emacs mode means it has an emacs mode so emacs can properly syntax highlight and indent whatever file format it uses. some of them also have full blown UIs written in emacs you can't refactor out
<pjt_014>
hmm
<pjt_014>
can you easily tell how factor-outable it'd be by looking at it?
<quinn>
pjt_014: yeah, probably
<etu>
yeah, there's minor modes and major modes. Major modes are things like syntax highlighting and minor modes are... pretty much everything else...
<etu>
It's usually functions that can be keybinded if it's a minor mode
<pjt_014>
but the packaging ought to be changed too
<quinn>
pjt_014: elisp is really only compiled after by emacs or with JIT iirc. that's irrelevant. something some (crazy) devs do is they make the primary 'ui' of their application in emacs. org-mode is an example, so is notmuch iirc. it looks to me like that might be the case. are you able to work with recutils files without emacs?
orivej has quit [Ping timeout: 265 seconds]
kvda has quit [Ping timeout: 265 seconds]
orivej has joined #nixos
<quinn>
pjt_014: on further examination, it seems like there are some binaries accompanying it. i would say as long as you can use the package for most of its intended features w/o emacs it's reasonable to make a PR and remove the dep.
<pjt_014>
It's a plain-text database, so yes. Maybe the lack of emacs could make it croak while trying to do something contrived, but I haven't come across anything that comes close to that.
<quinn>
pjt_014: no, it shouldn't. i don't want to look at the source any more but i would be SHOCKED if the C integrates emacs as a library because emacs gets very breaky when you use it that way. gnu developers are crazy but not that crazy (i hope)
ilikeheaps has quit [Ping timeout: 260 seconds]
kvda has joined #nixos
<quinn>
i haven't used nix for packaging emacs, i used a special plugin, but that looks right.
chloekek has quit [Ping timeout: 260 seconds]
smatting has joined #nixos
ilikeheaps has joined #nixos
<pjt_014>
quinn: I do have an idea for a compromise
lord| has quit [Ping timeout: 256 seconds]
<pjt_014>
either removing the emacs from the recutils package and then introducing a recutilsFull one with the emacs stuff, or doing the reverse and having a recutilsMinimal.
<pjt_014>
that'd be if they can't be cleanly separated
<quinn>
pjt_014: they can definitely be cleanly separated dude. just remove the dependency, it's not necessary for building or the functioning of the binaries as far as i can tell.
knupfer has joined #nixos
<pjt_014>
okay, non-crazy gnu devs then. That's good.
<pistache>
I've seen a few users bit by this problem use the override to remove emacs, so maybe (just a suggestion, I don't really know) it would be good to still accept the emacs argument if you remove the dependency
<quinn>
pjt_014: you should review the source code yourself, use the test suite (if it has one), or write your own imo. if you don't want to do the due-diligence you should leave it a personal overlay
<quinn>
pistache: that's interesting. maybe the binaries to call emacs *shudders*. i think you are right about still taking the argument.
dingenskirchen has quit [Read error: Connection reset by peer]
dingenskirchen1 has joined #nixos
orivej_ has joined #nixos
<angerman>
why would my nix-prefetch-git not produce any json?
orivej has quit [Ping timeout: 272 seconds]
dingenskirchen1 is now known as dingenskirchen
chiastre has quit [Ping timeout: 240 seconds]
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pistache>
quinn: hmm apparently recutils does import emacs code (for example an implementation of free() in alloc.c), but it looks like it's optional anyway
jgeerds has joined #nixos
<pjt_014>
what frustrates me is that there's a vim plugin too, but it's separate >_<
<quinn>
pistache: i can't find that, do you mind telling me the full path?
<pjt_014>
angerman: what's the exact command you're trying?
chiastre has joined #nixos
<quinn>
pjt_014: some people really go crazy for emacs, i don't really understand it (despite using it for a year).
<pistache>
quinn: ./lib/alloca.c
<bqv>
Guys, this is hardly an obnoxious PR and actually fixes some behaviour #85462 - if anyone here can, I'd appreciate it merged, otherwise I'm going to close it in a day or so and you can cherry-pick the commit later
<pistache>
quinn: don't worry, it didn't sound confrontational at all
<pjt_014>
angerman: It could be caused by the specific repo maybe? I just tried nix-prefetch-git https://github.com/rxi/lite and it worked fine. Gave me a bunch of other output but there's --quiet for that.
<angerman>
pjt_014: on my mac it just works, on the linux machine it just gives me the hash.
<pjt_014>
hm. a version issue maybe? This thing has no --version flag though, which is frustrating.
<quinn>
pistache: WhAAAAAt. dude that is BAFFLING. fd can't even find any c code with 'alloc' in the git checkout.
<raboof>
having a quick look at nixops, it looks like that provides a way to easily _update_ remote/cloud NixOS machines. I'm looking for something that can easily _replace_ nix-built images. would nixops work for that as well, or is there something else I could look at?
<quinn>
pistache: (oh i looked at the alloca.c, you are absolutely right)
<pistache>
quinn: yeah it's weird, I'm still trying to understand what's going on
<raboof>
(the deployed images I want to update/replace are built with nix but not full nixos machines themselves)
doyougnu` has quit [Remote host closed the connection]
kvda has joined #nixos
<quinn>
pistache: may god protect you
<pistache>
quinn: EMACS_FREE is not even present in the Git history.. there are some 'ifdef emacs' for petty things that were removed in 2013
kvda has quit [Client Quit]
orivej_ has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
thc202 has joined #nixos
o1lo01ol1o has joined #nixos
<quinn>
pistache: maybe check the CVS? it looks like savannah offers checkouts for both or something
virus_dave has joined #nixos
<quinn>
actually that might just be for the website
sigmundv__ has joined #nixos
virus_dave has quit [Client Quit]
sigmundv__ has quit [Remote host closed the connection]
<jtojnar>
angerman you can also run it in debug mode `bash -x (which nix-prefetch-git) url`
<quinn>
pjt_014: yeah that's the url pistache is using, however why TF it's not in the git or in the git history is a little concerning. it means that the source code they hand over at releases is (significantly) different from the ones in VCS
<pistache>
quinn: it's because they bundle gnulib in the releases
<pistache>
recutils itself doesn't seem to have much relation to emacs (except the mode it provides), but gnulib does use some code from emacs when it's available
agsdheidjd has quit [Ping timeout: 244 seconds]
agsdheidjd has joined #nixos
<quinn>
pistache: that is very spooky. thanks for figuring it out, pretty interesting
orivej has quit [Quit: No Ping reply in 180 seconds.]
<{^_^}>
[nixpkgs] @bqv closed pull request #85462 → nixos/nat: substitute iptables for compat under nftables → https://git.io/JfJ2q
<unclechu>
hey, about x-server, i have this option `services.xserver.deviceSection = '' Option "TearFree" "true" '';` but it doesn't work, i realized that in `xserver.conf` there are no `Device`-sections for some reason
<unclechu>
any ideas why there might be no `Device` sections in the `xserver.conf` for my X-session?
<pjt_014>
pistache, quinn: well, I gotta go for now. You both cool if I include some of the chat long when I do make the gh issue?
tareluerlz2 has joined #nixos
orivej has joined #nixos
__monty__ has joined #nixos
<tareluerlz2>
I do muliboot and I was wondering ,is it hard to not mess that up ? chose a already formatted , not wreck , efi partition , just asking
doyougnu has joined #nixos
cfricke has joined #nixos
<clever>
tareluerlz2: i think the "proper" way to do multiboot on efi, is that each OS registers itself in the efi vars, and the bios lets you pick one
maddo has joined #nixos
<unclechu>
i realized that it happens only for `i915` video driver, i added `modesetting` to `services.xserver.videoDrivers` list (there was only `i915` there) and that `TearFree` option appeared there
<tareluerlz2>
I do have mulitboot system , Can you easy use an alreadly made partition (root) , swap so
leotaku has quit [Ping timeout: 256 seconds]
justanotheruser has quit [Ping timeout: 260 seconds]
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
justanotheruser has joined #nixos
<clever>
tareluerlz2: nixos tends to break if other things are in /etc, /bin, and /lib, so its best to give nixos its own rootfs, zfs and lvm make that a lot simpler to manage
<tareluerlz2>
seem easy enough to chose its own root, zfs ( used) that with Freebsd it not works so well , why lvm?
<unclechu>
i actually had to name the video driver as `intel`, not as `i915`
<pistache_>
tareluerlz2: sorry, had some problems with my client there
<pistache_>
tareluerlz2: I was asking for the type of the filesystem you want to resize (ext4 ? ext3 ? btrfs ?)
<alj[m]>
Hey guys! How can I override (part of) a nixos module? (nixos/modules/system/boot/loader in this case)
<tareluerlz2>
ext4 , oh and thanks for the help . If you at work , I don't need to know this stuff
fendor has joined #nixos
<typetetris>
Is there something besides `enableIPv6` I need to do, to get SLAAC working?
dermetfan has joined #nixos
<pistache_>
tareluerlz2: MS-DOS partition tables are limited to four primary partitions
<pistache_>
tareluerlz2: but you can have more "logical" partition, or you can have a GPT partition table, which is limited to 128 partitions if I remember correctly
<pistache_>
tareluerlz2: so after resizing the partition itself, you'd need to use the resize2fs tool to resize the filesystem partition
<pistache_>
to resize the filesystem, I mean
o1lo01ol1o has quit [Ping timeout: 264 seconds]
<azazel>
alj: there's a section about declaring/overriding/loading nixos modules in the manual
<b42>
typetetris: i haven't played with this but i'm guessing you need to either enable dhcp on the interface (and let dhcpcd handle slaac) or set boot.kernel.sysctl."net.ipv6.conf.IFACE.accept_ra" to 1 (or 2)
orivej has quit [Quit: No Ping reply in 180 seconds.]
orivej has joined #nixos
drakonis_ has quit [Ping timeout: 240 seconds]
o1lo01ol1o has joined #nixos
drakonis_ has joined #nixos
<typetetris>
b42: thanks
Guest16883 has quit [Quit: leaving]
<typetetris>
b42: dhcp is activated but i only get a link local address. But the host has a globally reachable (whatever that means, i am out of my depth here) ipv6 address. I configured it statically, and also set `networking.defaultGateway6` but `ip r l` doesn't show any ipv6 related stuff though. `networking.enableDHCP=true` still.
<typetetris>
b42: `useDHCP` on the interface not `networking.enableDHCP`.
<pistache_>
typetetris: "ip -6 r l" ?
domogled has joined #nixos
iyzsong has joined #nixos
<ramses_>
jtojnar: I might get some inspiration there indeed, thanks!
<typetetris>
pistache_: Thanks, that shows something!
syd has joined #nixos
<syd>
Hi peeps
<syd>
I'm trying to override the Service.ExecStopPost value for each systemd.user.services
zupo has quit [Ping timeout: 264 seconds]
kvda has joined #nixos
<syd>
but when I try this, I get an infinite recursion
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<clever>
syd: nix must first know what attrs config.systemd.user.services contains, before it can give you the value of config.systemd.user.services
<clever>
syd: so it has to call mapAttrs before it can know what the args to mapAttrs are
<alj[m]>
Why does replacing `generic-extlinux-compatible` in system/boot/loader prompt my whole system to rebuild half its binaries? Is there any way to stop that?
orivej has quit [Ping timeout: 246 seconds]
<syd>
clever: that makes sense
<syd>
so how would I work around that then?
wnklmnn has quit [Ping timeout: 260 seconds]
<clever>
alj[m]: find the .drv from before and after, and run nix-diff on both of them
orivej has joined #nixos
<clever>
syd: its not really possible to override that, youll need to just edit the module that created the option to begin with
<typetetris>
the nixos firewall logs some connections, where can I take a look at those logs? `journalctl -f` only shows me ipv4 stuff, does that mean ipv6 doesn't reach that host and it is blocked before?
<clever>
syd: it would likely be simpler for you to just watch the journal for things exiting, and then act on that instead
<syd>
Ah it's just not possible. That's too bad. I figured maybe there was a way
<{^_^}>
[nixos-search] @garbas merged pull request #35 → Use a custom attrname analyzer → https://git.io/JfuMT
<{^_^}>
[nixos-search] @garbas pushed commit from @adisbladis to master « Use a custom attrname analyzer (#35) »: https://git.io/JfuH7
<pistache_>
syd: as suggested by clever, you can try to override the value of systemd.units in in system/boot/systemd.nix (or the equivalent for home-manager), but yes watching the logs might be an easier solution
<b42>
typetetris: possibly ... you can pass -n to ip6tables to see how many packets matched which rule, that is sometimes useful
<syd>
pistache_ clever: what do you mean by 'watching' logs?
<syd>
I'd have to write some separate program for that right
<typetetris>
if using nixops is a `system.stateVersion` in the configuration for a host still necessary? (using nixops for a nixos host)
<clever>
typetetris: the first time nixops deploys, it will query the remote machine for its stateVersion, and then put that into the internal sqlite database
<clever>
typetetris: but its then set using mkDefault, so your config can still override things
<clever>
what you set it to isnt that important, its more important that it doesnt change without you planning for that change
orivej has quit [Ping timeout: 264 seconds]
orivej_ has joined #nixos
<typetetris>
clever: Okay, is there some manual, what I have to pay attention to, when it should be changed? (Just read about for example postgresql default version.)
<clever>
typetetris: its usually in the nixos release notes
<b42>
,stateVersion
<{^_^}>
Changing stateVersion 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 need to change stateVersion for some odd reason 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
<manveru>
https://hydra.nixos.org/build/119042848 fails because of some strange compiler-rt issue... tried finding the cause but since it seems to never have been built before it's kinda hard
<alj[m]>
clever: I've been trying to get the information you've asked for, but I'm getting tilted trying to find the right drv... The real question is why replacing the generic extlinux module warrants rebuilding all of this: https://termbin.com/9gaq
orivej has joined #nixos
<clever>
alj[m]: you just want the top-level drv for all of nixos
<clever>
alj[m]: nix-instantiate '<nixpkgs/nixos>' -A system, before and after the change
kvda has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
waleee-cl has quit [Quit: Connection closed for inactivity]
<alj[m]>
clever: Is there some way to not use nix-diff for troubleshooting this? I cant install it because ghc doesnt work on armvl7 for some reason...
<alj[m]>
(one of the reasons for tilt)
<clever>
alj[m]: use nix-copy-closure to copy both drv files to an x86 machine
<clever>
then you can use x86 nix-diff
lunik1 has joined #nixos
vidbina has joined #nixos
<hyper_ch>
trying to build samba 4.12.3 that was just released :)
potong has joined #nixos
ilikeheaps has quit [Quit: Konversation terminated!]
<alj[m]>
clever:
<alj[m]>
* clever: its saying its pulling 1500 paths, is it going to pull the whole store over now? i dont have that much disk space
<clever>
alj[m]: it will only be copying tiny .drv files and any paths from foo = ./.;
<alj[m]>
its saying its copying 1500 paths, is it gonna fill up my disk now?
<alj[m]>
Also apologies, my client is being wonky as fuck
jkarni_ has joined #nixos
rauno has joined #nixos
<alj[m]>
It says the only difference is the file i actually changed aka the patch in the extlinux-conf-build.sh. https://termbin.com/fou2
<clever>
alj[m]: if you undo the change and nixos-rebuild, what happens?
<clever>
pistache_: i dont think ive ever had problems deleting a snapshot while full
<hyper_ch>
so, PR done for samba 4.12.3... my problem with shadow copy2 still remains though :(
<clever>
pistache_: i have however had problems deleting a file while full, because deleting a file modifies the dir, but `echo -n > file` truncates without updating the dir
<alj[m]>
clever: is there any way i can prevent a rebuild? I'm literally changing one line in that stupid script and I dont want to waste all this time rebuilding my binaries...
<clever>
alj[m]: if you take the 2 drv's you gave to nix-diff, and run `nix-build --dry-run` on each, what is the results?
<typetetris>
do flake references allow specifying the sha? If not, why not?
<alj[m]>
on my x86 machine or the arm?
Maxdamantus has quit [Ping timeout: 256 seconds]
<clever>
alj[m]: on the arm
Darkmatter66_ has joined #nixos
Maxdamantus has joined #nixos
Darkmatter66 has quit [Ping timeout: 264 seconds]
<pistache_>
clever: it's still an open issue, that's why I advise being careful with refreservation (always manually leave some space for ZFS itself)
morph000 has joined #nixos
<pistache_>
see zfs#9895 and zfs#9849 for more info
morph000 has quit [Client Quit]
<alj[m]>
> clever: I've been trying to get the information you've asked for, but I'm getting tilted trying to find the right drv... The real question is why replacing the generic extlinux module warrants rebuilding all of this: https://termbin.com/9gaq
<alj[m]>
clever: no output on the "before changes".drv and the same output as in that termbin for the "after changes".drv
<{^_^}>
error: syntax error, unexpected ',', expecting ')', at (string):310:65
<pistache_>
I *think* that this problem appeared with 0.8.0, because I don't remember this happenning with 0.7.X
m1sosoup has quit [Ping timeout: 256 seconds]
m0rph has quit [Ping timeout: 256 seconds]
<clever>
alj[m]: i think the problem, is that the gcc used to build everything may not be in the cache?
<clever>
alj[m]: so you need to build gcc just to get the ability to build anything
<clever>
pistache_: ah, i'll read those tickets for more info
jkarni_ has quit [Ping timeout: 246 seconds]
<pistache_>
it was quite an experience, breaking a whole server with just a snapshot
<clever>
pistache_: there is also the new checkpointing feature, which might be able to undo such a thing
jkarni_ has joined #nixos
<clever>
pistache_: basically, a checkpoint is just a pool-wide snapshot, there can only be one, and it can even undo zpool feature upgrades
<clever>
but reverting to a checkpoint will undo everything on the entire pool
<alj[m]>
clever: looking at the all the paths containing "gcc" i find that gcc 9 (the one that is also trying to be built) is not installed but gcc 7.4 is... but why? i'm so confused right now
CRTified has quit [Quit: Gateway shutdown]
<pistache_>
clever: yep, but that would only help if there was a checkpoint that had been made
<hyper_ch>
how does "github" know whom to contact as reviewers for a PR?
<pistache_>
however, I was able to recover the pool by rebooting and importing an earlier TXG
<pistache_>
but had to made a full remote backup before that because at that point I wasn't even sure I was going to be able to import the pool again
<clever>
alj[m]: has this machine ever built something from source before?
<alj[m]>
it has, yes
<clever>
not sure whats up then
kaliumxyz has quit [Ping timeout: 256 seconds]
zaeph1 has quit [Ping timeout: 260 seconds]
reallymemorable has joined #nixos
<alj[m]>
nice.
<alj[m]>
guess I'll just let it run then..
<tareluerlz2>
well, I hope to install nixos today
lucksus has joined #nixos
maxdevjs has quit [Remote host closed the connection]
cosimone has joined #nixos
zaeph has joined #nixos
<alj[m]>
clever: Thanks for the awesome help again, you rock!
<lucksus>
Hey folks, I'm trying to cross-compile my Rust project to Android using a nix-shell (on Ubuntu host) with `crossSystem`, and I'm running into the exact same problem described and patched here: https://github.com/NixOS/nixpkgs/issues/56156. I tried to apply the same patch using an overly but no matter what I do, it seems the overlay doesn't apply to the LLVM pulled in by crossSystem. And I can't find comprehensive documentation for
<{^_^}>
#56156 (by nomeata, 1 year ago, closed): Cannot build LLVM-6 for Android
<lucksus>
crossSystem, like how to set the LLVM version used etc. Is there anybody on who could help me understanding und configuring crossSystem?
<alj[m]>
clever++
<{^_^}>
clever's karma got increased to 428
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pistache_>
tareluerlz2: NixOS's installation process was one of the first things that hooked me up
<pistache_>
tareluerlz2: it's very.. straightforward. you setup your partition, a proper configuration file and run "nixos-install", and you have a working system
maxdevjs has quit [Remote host closed the connection]
<tareluerlz2>
the partition part is what I wonder about , I have not used parted in sometime, I just don't want to delete my other systems
morgrimm has joined #nixos
<betawaffle>
what's the correct way to get nixos to configure overrides for a template unit instance? ie. getty@tty2
<betawaffle>
i want to add environment variables to just that unit
tareluerlz2 has quit [Remote host closed the connection]
<pistache_>
betawaffle: systemd.services."getty@tty2".serviceConfig.Environment, I suppose
<betawaffle>
as opposed to `systemd.services."getty@tty2".environment?
<pistache_>
right, you can do it with .environment as well
<betawaffle>
the problem is (i think) nixos will just try to create a getty@tty2.service, rather than an overrides.conf because the existing service is getty@.service (a template), and nixos doesn't understand they are related
Rusty1 has joined #nixos
<betawaffle>
at which point it's not an override, and it doesn't have all the existing bits of getty@
<pistache_>
tarelerulz2: mhh what I wrote was wrong, that partition with "nixos" label is not the installer itself, you either created a filesystem yourself or it was existing beforehand
<{^_^}>
[nixpkgs] @nlewo merged pull request #87154 → Some performance optimizations to dockerTools.build{,Layered}Image → https://git.io/JfnGZ
<cransom>
gchristensen: question on hydras `one_at_a_time`. I run the default (4) evaluators. i have it turned on for 3 of my projects, but judging from the log and the time since last evaluation, they don't seem to run in parallel. rather, i don't see any other evaluations happening while one of the one-at-a-times are evaluating.
knupfer has quit [Quit: knupfer]
<gchristensen>
oh it isn't about concurrency
<gchristensen>
it won't evaluate the project while the current latest evalution has any pending jobs
<cransom>
hrm. what i see is that with it enabled, the time since last evaluation is 10x higher (usually around 120s, was getting into 1500 seconds). it was a quiet time, there were no new/existing jobs or work being done.
<silviogutierrez>
Hi all, huge Nix fan here and I'm trying to really grasp it. So thanks in advance to all those who contribute to it. I'd like to better understand channel updates. Let me know if this is the right place or not. I've "pinned" to 20.03, but see some packages aggressively update and others are very far behind. Is there a system behind this?For
<silviogutierrez>
example, on 20.03, GitHub CLI tool was bumped a few days ago to the very latest 0.8.0. See here:https://github.com/NixOS/nixpkgs-channels/blob/nixos-20.03/pkgs/applications/version-management/git-and-tools/gh/default.nixBut on the same release, 20.03, the Google Cloud SDK is lagging months
<pingiun>
but now I'm trying to add unstable plex to my package overrides, but nix complains that I havent enabled unfree packages
<pingiun>
even though I have enabled it and it works when I'm on the regular channel for plex
LambdaDuck has quit [Ping timeout: 265 seconds]
<Bryan[m]>
Hi silviogutierrez , my understanding is that only minor version bumps are included in the stable channel, modulo some backporting decisions I'm not familiar with. Unstable gets all the changes. Plus, it's up to the contributor's own initiative to add and update most packages, so you can imagine that creates some variability in how quickly things get updates
<silviogutierrez>
so Bryan[m], any package maintainer can decide to back port these changes into the release channels?
<Bryan[m]>
Back ports might just be security patches? I'm actually just speculating, sorry. Ah, was the sdk in master before the 20.03 branch was created though? That was some months back. And it's also possible that it was marked as broken if there was some reason it was failing to build before the time cutoff
<silviogutierrez>
ah alright, I may try the forums as well
<silviogutierrez>
thanks
<Bryan[m]>
Cool, good luck
<tarelerulz2>
I'm looking at nano /mnt/etc/nixos/configuration.nix ,but I don't see the boot.grub.useOSProber
silviogutierrez has quit [Remote host closed the connection]
waleee-cl has joined #nixos
LambdaDuck has joined #nixos
<tarelerulz2>
building the configuration in /mnt/etc/nixos/configuration.nix...error: syntax error, unexpected '=', expecting $end, at /mnt/etc/nixos/configuration.nix:6:25(use '--show-trace' to show detailed location information)
magnetophon has quit [Ping timeout: 260 seconds]
<clever>
tarelerulz2: can you pastebin the contents of the file?
<evanjs>
closest I can see there is install-iso-hyperv IIUC
<evanjs>
ping lassulus I guess?
<pingiun>
how can I set nixpkgs.config.allowUnfree on an unstable nixexprs that I downloaded? I can't seem to evaluate plex when i have set that option in my configuration.nix
<infinisil>
clever: They're already doing config = config.nixpkgs.config in the post
<infinisil>
pingiun: So it should just be `nixpkgs.config.allowUnfree = true` in some line in your configuration.nix
<lassulus>
evanjs: uhm, never did anything with hyper-v, so not sure. the make-disk-image.nix I'm using from nixpkgs is supporting these formats: qcow2, qcow2-compressed, vpc, raw
<pingiun>
ah I understand now
<pingiun>
thanks
cosimone has quit [Remote host closed the connection]
<evanjs>
lassulus: I keep seeing "raw" and "vmdk" next to each other so I wonder if that works lol
cosimone has joined #nixos
<lassulus>
evanjs: if `qemu-img convert` can convert to it, I could work
<lassulus>
but maybe just booting a raw image works?
<lucksus>
Hey, anybody who can help me getting an overlay to work inside crossSystem?
<evanjs>
yeah that's what I'm wondering. hrm
drakonis has quit [Ping timeout: 260 seconds]
<evanjs>
time for a raw hello world I guess :D
<lucksus>
Or explain to me how to packages inside crossSystem (like LLVM etc.) are configured?
<evanjs>
"Oh, packages, that's not a big---"
<evanjs>
Like LLVM -- OH
<evanjs>
Those are a bit more complicated on crossPkgs AFAIK
<lucksus>
Ah! So instead of `llvm = super.llvm.override {...}` I would have to do sometyhing like `crossPkgs.llvm = super.crossPkgs.llvm.override {...}` in the overlay?
<tarelerulz2>
No idea what I'm doing wrong
<evanjs>
lucksus: eh afaik it should override cross if you do the normal package but I could be wrong
<evanjs>
I thought cross just refers to the initial package in the end
jkarni_ has quit [Remote host closed the connection]
jkarni_ has joined #nixos
mallox has quit [Quit: WeeChat 2.8]
simba1 has joined #nixos
teto has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from @edolstra to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/JfzUF
tarelerulz2 has left #nixos [#nixos]
tarelerulz2 has joined #nixos
<lucksus>
evanjs: well, I tried everything I could think of to somehow alter LLVM in my overlay but it never affected the derivation that was kicked off by crossSystem. Will try overriding crossPkgs.llvm_6 explicitly.
mog has quit [Ping timeout: 246 seconds]
mog has joined #nixos
ixxie has joined #nixos
<ramses_>
Can I have node scripts produced by building a node2nix package, be wrapped? It seems that they only work when I cd to the store path so that node_modules is in the same directory
KeiraT has quit [Remote host closed the connection]
<evanjs>
rednaZ5: I would initially say this is due to not everything being backported, but hrm.... perhaps there was a CVE that made backporting an update for nodejs to 19.09 more feasible than 20.03, and since 12.16.1 was the newest, that's what was used?
<tarelerulz2>
building the configuration in /mnt/etc/nixos/configuration.nix...error: syntax error, unexpected '=', expecting $end, at /mnt/etc/nixos/configuration.nix:15:11(use '--show-trace' to show detailed location information)
<clever>
tarelerulz2: can you pastebin the new config file?
<clever>
tarelerulz2: you have an extra } that marks the end of the file, in the middle of the file
<clever>
tarelerulz2: every { must be matched to another }
<pingiun>
if I define overlays in nixpkgs.overlays, they should be picked up by nixos modules right?
<evanjs>
bah. a 'x86_64-linux' with features {kvm} is required to build '/nix/store/airn0g1s2vm2igfay8fbsqivq57pglas-nixos-disk-image.drv', but I am a 'x86_64-linux' with features {benchmark, big-parallel, nixos-test}
<{^_^}>
[nixos-search] @garbas pushed 2 commits to bool-query: https://git.io/Jfzkj
<evanjs>
I have to do this on my real linux computer don't I lassulus lol
<evanjs>
(Doing it on Ubuntu via Windows WSL atm)
<evanjs>
WSL 1 **
<evanjs>
not sure if WSL2 includes KVM support or etc... presumably not as it's still a guest machine
<lassulus>
uhm, it needs to be able to run kvm
<lassulus>
I never did anything with WSL also :D
<evanjs>
now I am really curious if the kernel for wsl2 supports kvm
<tarelerulz2>
{ config, pkgs, ... }: then { all the setting , then at the very end of the file }
<clever>
tarelerulz2: yes
<tarelerulz2>
nope I just did that and it did not work
<evanjs>
rednaZ5: so security issues, fixing broken packages that made it into stable release branches, fixing broken download links, etc
meck has joined #nixos
<evanjs>
those are a _few_ examples of why backports might be desirable
<tarelerulz2>
unexpected ']'
<simpson>
rednaZ5: Another is continued compatibility, which can occur at levels of linkage which Nix doesn't understand or handle.
<pingiun>
is there some way nixpkgs overlays are ignored from configuration.nix or am I applying the overlay wrong?
<rednaZ5>
I guess, I should have asked why are channels a thing.
<infinisil>
pingiun: Can't reproduce, this config: https://paste.infinisil.com/toV533mQmw.nix with `nix-instantiate '<nixpkgs/nixos>' --arg configuration ./file.nix -A pkgs.plex` throws an error (as it should)
growpotkin has joined #nixos
numkem_ has joined #nixos
<evanjs>
rednaZ5: that's so we can continue to support various releases for a bit so people aren't always on the bleeding edge if they don't want to be. Otherwise we'd just have "master", unless I misread your question
<pingiun>
infinisil: I was doing nixpkgs.config.overlays instead of nixpkgs.overlays, that seems to be the crux
<gustavderdrache>
my guess is that the nix-build session/process has a pseudo-terminal hooked up to it
<evanjs>
there it is, ty. Code nav isn't on for this repo and I am le tired :D
<evanjs>
Yeah probably
<evanjs>
I should open an issue on... nixpkgs? for that
<evanjs>
idk, the workaround is hacky and it doesn't work everywhere
<evanjs>
it only seems to fix it in `nixos-rebuild` and `nix-env` afaict
<leungbk>
I wrote a recipe for a manual Emacs package just now and found myself repeating the same inputs in buildInputs and in propagatedUserEnvPkgs. Is there a way to only write them once? http://paste.debian.net/1147656
<leungbk>
proofofkeags has quit [Remote host closed the connection]
<gustavderdrache>
probably safer to use the more targeted analytics-disabling env var since that check may or may not remain in angular cli as it matures and learns what pipes are
<gustavderdrache>
yeah i went there
<leungbk>
When I put the inputs only in buildInputs, the package wouldn't work at runtime
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
<leungbk>
evanjs: Thanks!
user_0x58 has joined #nixos
MmeQuignon has joined #nixos
<adisbladis>
leungbk: lpy is in melpa, so why make your own package?
rednaZ5 has quit [Quit: ERC (IRC client for Emacs 26.3)]
<evanjs>
you can check for them with nix-env -qaPA nixos.emacsPackages (or nixpkgs.emacsPackages if you're not running NixOS)
<leungbk>
adisbladis: Oh, I didn't realize that. I had wanted to write a Nix recipe for it a while ago back when it wasn't in melpa but didn't get around to testing the recipe till today.
<evanjs>
as they don't appear in `nix search` results unless you tweak some things in your configuration...
floscr has joined #nixos
o1lo01ol1o has quit [Ping timeout: 240 seconds]
<T0pH4t>
if i have a custom pkg that does a fetch from git, does nix need to always hit that git repository regardless of whether that binary is in a binary cache?
<dsal>
How can I debug this? -- warning: warning: unable to download 'https://nixos.org/channels/nixos-20.03': Problem with the SSL CA cert (path? access rights?) (77); using cached result
alp has quit [Remote host closed the connection]
<dsal>
`nix-channel --update` fails for me with the above error.
<adisbladis>
leungbk: Fair enough :)
alp has joined #nixos
turion has joined #nixos
<floscr>
When I start my session an xterm window opens, when I close that, the session get's killed and I'm back on the login screen?
<floscr>
I've changed some setting while refactoring a while ago, anyone an idea what it was?
orivej_ has joined #nixos
orivej has quit [Read error: Connection reset by peer]
<dsal>
Ah, thanks. I don't know what the system uses, but I had a bundle lying around that got it moving
duckfullstop has joined #nixos
<dsal>
Where does nix-shell get its environment? I can't figure out how it could be getting a newline at the beginning of PS1.
<T0pH4t>
is there a way to print out when a derivation will be built and not used form cache?
<T0pH4t>
print out why*
<srhb>
T0pH4t: Not in general. The answer is always the generic "because it's not in the cache"
<T0pH4t>
well it doesn't even say that
<srhb>
T0pH4t: The way nix determines what to try and fetch is to compute the final hash of the package, and then check if the cache has it.
<T0pH4t>
i mean i know its there cause i put it there
<srhb>
T0pH4t: Ah, it's your cache?
<T0pH4t>
but i built on one system and want to use it on another. They should be compatible, but apparently are not? So i'd like to triage why
<T0pH4t>
yeah a private cache in s3 that i pushe bins to and then use in vms
<srhb>
T0pH4t: nix-instantiate the expression on each system
<srhb>
T0pH4t: use nix-diff to compare the resulting drv files
<T0pH4t>
that just givs me drv file
<srhb>
Each diffence is the cause of the mismatch.
<T0pH4t>
what would i even be looking for in there?
<srhb>
Any difference at all.
<T0pH4t>
like the /nix/store/<hash>, will has be the same on any machine assuming the inputs are the same
<T0pH4t>
or our hashes unique per machine
<srhb>
T0pH4t: Yes, and the inputs are the same if the drvs are the same (and vice versa)
<T0pH4t>
ok so the entire pkg name including hash should match cross machine and else wise there is a difference?
<srhb>
T0pH4t: So first check that the inputs are indeed the same
<srhb>
T0pH4t: Yes.
floscr- has quit [Remote host closed the connection]
<T0pH4t>
ok thx, is there enough info in drv to figure out why a hash changed, eg like system params?
<srhb>
T0pH4t: It depends on the derivation how clear it is, but usually yes.
<dsal>
It looks like the nix-shell PS1 is hard-coded... That's unfortunate.
<T0pH4t>
ok thx alot
<srhb>
T0pH4t: And again, nix-diff is really useful to pin-point the differences.
orivej has joined #nixos
<T0pH4t>
ok, will give that a shot
orivej_ has quit [Read error: Connection reset by peer]
tarelerulz has joined #nixos
johnnyfive has quit [Ping timeout: 256 seconds]
<tarelerulz>
this is like 5 times now , does efi work? You flow , what the manual says and does not work . building the configuration in /mnt/etc/nixos/configuration.nix...error: syntax error, unexpected ID, expecting '.' or '=', at /mnt/etc/nixos/configuration.nix:69:11
<tarelerulz>
that I think is the default setting
<srhb>
tarelerulz: You have a syntax error. Perhaps you can pastebin what you have, and we can help you find it.
<{^_^}>
[nixpkgs] @veprbl pushed commit from @ryanorendorff to master « agdaPackages.cubical: init at 0.2 (#76994) »: https://git.io/Jfzmg
<tarelerulz>
I starting to wonder if something not mess up , in the file itself for my set like , nixos does not work for me .
johnnyfive has joined #nixos
rednaZ5 has joined #nixos
cfricke has joined #nixos
<srhb>
tarelerulz: Sorry, I'm having a hard time following what you're saying. You pasted an error message. That error message indicates you made a syntax error in configuration.nix. We can help you find it if you share your file online.
<camsbury>
hi there - on a thinkpad P50, trying to boot into a fresh nixos install, but I get kicked back into the boot menu immediately
<tarelerulz>
It all the default, that is why I'm lost
<srhb>
tarelerulz: You're missing a # on the very first line for one
pingiun has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
evils has joined #nixos
<rednaZ5>
I need a single package from nixos-unstable. What is best practice to achieve that? Am I right to assume that simply adding nixos-unstable as an additional channel will cause "$ nix-env -u" to replace all my packages with unstable ones? I would try it and I am currently but accessing packages by their name is currently taking about twenty minutes. (I am not sure if that has always been the case.)
<tarelerulz>
[ # Include the results of the hardware scan.
<T0pH4t>
does the nixos channel eg 20.03 just point at the nixpkgs repo tagged 20.03?
<srhb>
T0pH4t: Not "just" -- but yes, the nixos-20.03 branch is updated to point at the tested commit when the 20.03 jobs succeed.
<srhb>
T0pH4t: (in reality the "raw" branch is called release-20.03 and it's "promoted" to nixos-20.03 when the tests for that release branch succeeds)
<T0pH4t>
srhb: ok cool, I'm assuming there is like a 20.04 if changes get made?
<srhb>
Nope.
<srhb>
20.03 just gets bumped
<T0pH4t>
oh so the commit its associated with changes?
<srhb>
Yes. Not a tag, a branch.
<T0pH4t>
ok.. so i should point at release branch if i have checked out the repo to a priv dir and use a nix path against it?
<srhb>
rednaZ5: You can have multiple channels without nix-env -u grabbing whatever is "newest" from another branch, but I forget the mechanism. And going by name is pretty much broken at this stage, always use -iA channel.thing
yimmy149 has joined #nixos
magnetophon has quit [Read error: Connection reset by peer]
<srhb>
T0pH4t: No, the release branch (release-20.03) is untested. You should generally use the one called nixos-20.03
<T0pH4t>
srhb: ok thx
<tarelerulz>
sudo rm configuration.nix sould remove the config file , on the nixos install mount right?
magnetophon has joined #nixos
<srhb>
tarelerulz: You mean rm /mnt/etc/nixos/configuration.nix ? Yes.
gxt has joined #nixos
<numkem>
I've got nix running on ubuntu and I'm trying to use it as a build machine for hydra. When hydra runs a build on it I get `error: you are not privileged to build derivations` as an error. I've looked at nixpkgs for issues related to this error message and I'm finding nothing
<tarelerulz>
I started the file over , can it run like that for an efi system?
<symphorien>
numkem: you must be a trusted user probably
<symphorien>
See the manual page of nix.conf
<numkem>
symphorien: that's what i think but it's not clear in the man page, i'll try that
<srhb>
tarelerulz: If you generated the file with nixos-generate-config _and_ you're booted in efi mode, I believe it's generated with boot.loader.systemd-boot.enable = true -- so yes
<rednaZ5>
srhb: Thanks. So what is your best practice recommendation? Would it be too crazy to clone the nixpkgs-channels repository and point the "--file" option at it whenever I want to update that single unstable package?
<tarelerulz>
nixos-install can you just type that or do you need sudo?
teto has joined #nixos
<srhb>
rednaZ5: I do that quite happily for the few packages I manage imperatively.
<srhb>
rednaZ5: For everything else I import whatever nixpkgs I need from inside the expression itself.
<srhb>
tarelerulz: I think you need sudo. The installer used to run as root, the manual might not have been updated to account for that.
<srhb>
tarelerulz: actually it has. Every command with # in front of it should be run with sudo.
<srhb>
(from the manual)
<tarelerulz>
what if I want to add KDE to this?
<srhb>
tarelerulz: There's a comment in the generated file with two lines related to KDE
<rednaZ5>
srhb: Thanks.
rednaZ5 has quit [Quit: ERC (IRC client for Emacs 26.3)]
<srhb>
tarelerulz: If you add those lines in (by removing the #) you'll get KDE
drakonis1 has quit [Ping timeout: 246 seconds]
orivej has quit [Quit: No Ping reply in 180 seconds.]
<tarelerulz>
what about x11 settings?
<elvishjerricco>
So I'm using `nix build -f "<nixpkgs/nixos>" vm`, and I want to simulate the passage of time in that VM. I'm gonna test out sanoid/sync and I need to see how things work as hours and months go by
<elvishjerricco>
How would I do that?
morgrimm has quit [Ping timeout: 256 seconds]
<elvishjerricco>
sanoid/syncoid*
magnetophon has quit [Read error: Connection reset by peer]
<srhb>
tarelerulz: If you don't find it there, ask here and explain specifically what you're looking to configure.
magnetophon has joined #nixos
<tarelerulz>
do you need x11 for KDE?
<tarelerulz>
never set up KDE before
ramses_ has quit [Quit: Connection closed]
<tarelerulz>
at all
non has quit [Remote host closed the connection]
<tarelerulz>
I get mostly stuff like Ubuntu
<adisbladis>
Technically no, on NixOS currently yes.
<srhb>
adisbladis: I think the question is a bit simpler than what you're getting at :-)
magnetophon has quit [Read error: Connection reset by peer]
<srhb>
tarelerulz: The displayManager or plasma setting will imply enough of x11 to get you running, you shouldn't need to explicitly do anything barring weird hardware setups or whatnot.
<adisbladis>
cole-h: Set deployment.targetHost to something sensible like localhost or 127.0.0.1
chloekek has quit [Ping timeout: 265 seconds]
o1lo01ol1o has joined #nixos
<cole-h>
adisbladis: I see your config has it set to `192.168.0.44` -- what I was really asking is: "as long as the address in configuration.nix is the same as the targetHost, it should Just Work™," right?
bridge[evilred] has quit [Remote host closed the connection]
proofofkeags has quit [Remote host closed the connection]
<evanjs>
lassulus: okay, dumb question -- what's the easiest way to evaluate stuff (e.g. with `nix eval`) on something generated by nixos-generate?
proofofkeags has joined #nixos
<evanjs>
I might be thinking about this backwards or something
<lassulus>
not sure what you mean
o1lo01ol1o has quit [Ping timeout: 272 seconds]
bridge[evilred] has joined #nixos
red[evilred] has joined #nixos
o1lo01ol1o has joined #nixos
mog has quit [Ping timeout: 265 seconds]
<evanjs>
lassulus: like if I want to realise the configuration for a vm and then check stuff out like I normally would with nixos-option, etc
alp has quit [Remote host closed the connection]
<lassulus>
oh, hmm, well just start it in a vm-nogui? :D
<Raito_Bezarius>
evanjs: nix-instantiate?
alp has joined #nixos
mog has joined #nixos
<Raito_Bezarius>
nixops dos some stuff where it instantiates conf then parse it in Python (?) or does the inverse thing, python stuff to nix code
<Raito_Bezarius>
does*
<evanjs>
lassulus: holy crap I am tired. Was passing vm to -f and wondering if I could do headless
<evanjs>
omg I am behind on sleep :D
<evanjs>
Raito_Bezarius: yeah that too. I tried to instantiate it but it just output the paths for the vm stuffs
bqv has quit [Ping timeout: 240 seconds]
turion has quit [Ping timeout: 272 seconds]
<Raito_Bezarius>
nix-instantiate --eval --json ?
<Raito_Bezarius>
> -eval
<{^_^}>
value is a function while an integer was expected, at (string):310:1
<Raito_Bezarius>
Just parse and evaluate the input files, and print the resulting values on standard output. No instantiation of store derivations takes place.
<Raito_Bezarius>
oops sorry {^_^}
bqv has joined #nixos
jkarni_ has joined #nixos
orivej_ has quit [Ping timeout: 258 seconds]
orivej has joined #nixos
<T0pH4t>
curious if anyone knows, but why does nix need to access github if a package has a binary in a remote cache? Like what other imformation does it need thats not already listed in the source info?
erasmas has quit [Read error: Connection reset by peer]
<srhb>
T0pH4t: It needs to evaluate the nix expression to create the drv which in turn knows the hash of the package
<T0pH4t>
srhb, it can't just look at the attributes and avoid fetch?
<srhb>
T0pH4t: Uh, sure. Which package are you actually talking about?
teto has quit [Ping timeout: 252 seconds]
<srhb>
T0pH4t: If the expression is all local no access should be required to create the drv.
<T0pH4t>
well like if you have buildtins.fetchGit { url =..., rev = "..." }
<T0pH4t>
that information doesn't give you a unique identifier for src code?
<srhb>
T0pH4t: builtins.fetchGet is actually potentially impure.
<srhb>
Well, I guess not with "rev"
<T0pH4t>
wel lif its executes the fetch I would expect to it do. But i would think it would only fetch if it has to build the pkg
<srhb>
T0pH4t: It doesn't know beforehand though. But eval time fetchers can be confusing in that regard.
<T0pH4t>
doesn't know what? are you saying that it execute the fetch regardless and then afterwards only builds it if can't find an alrdy built binary?
<srhb>
T0pH4t: Contrast with eg fetchurl and friends where you specify the output hash
<T0pH4t>
ah so its going to generate a hash ?
<T0pH4t>
i mean in theory git hash should be unique to the code, but i guess u can't guarantee that...
seku has joined #nixos
<srhb>
T0pH4t: I may have to look up how this exact function behaves to be sure I'm not telling you lies, but your expectation would be right with _normal_ fixed output derivations
<srhb>
T0pH4t: That is, things like fetchurl and fetchTarball
<srhb>
T0pH4t: Because the hash is known beforehand, we can simply look 1) in the store 2) in the cache, and only 3) fallback to fetching online
<T0pH4t>
right since thats have file hash.
<T0pH4t>
hmmm so reve can't replace the hash i guess
<srhb>
I mean we probably have bigger problems if it doesn't, but then again, it's a moral decision on the part of Nix. And I forget how exactly it's handled
agsdheidjd has quit [Ping timeout: 265 seconds]
<hpfr[m]>
with builtins.readDir you can filter out directories but is there an easier way to filter on a file extension than builtins.match?
Jackneill has quit [Ping timeout: 258 seconds]
<T0pH4t>
if i give it the sha256 tho it should defer the fetch tho i think?
<T0pH4t>
guess I hsould try
<srhb>
T0pH4t: I don't think builtins.fetchGit supports that. If it did, yes, that would work
<evanjs>
Raito_Bezarius: mrm it basically just makes a system output (like what you'd see linked to /run/current-system). idk if I can actually get this out after the fact
alp has quit [Ping timeout: 256 seconds]
<T0pH4t>
nope :/ damn
<srhb>
T0pH4t: If you're not fetching from a private repo, just don't use that function.
<{^_^}>
[nixpkgs] @vbgl pushed to master « ocamlPackages.textutils: remove at 112.17.00 »: https://git.io/JfzOj
<numkem>
pikajude: I'm doing buitins.fetchTarball <url> and that returns me the path in the nix store. Now I want to add the path to the file I want to import (a .nix file)
<pikajude>
so you want to import a file that is inside the tarball?
<numkem>
yup
<pikajude>
you could just use import ((builtins.fetchTarbal {...}) + "/my-nix-file.nix)
<pikajude>
coercing a derivation to a string results in a string representing its out path
<pikajude>
(i think that's right, i haven't tested it)
<numkem>
the darn (), I keep forgeting about those
shafox has quit [Remote host closed the connection]
noudle has joined #nixos
<numkem>
pikajude: it's telling me it can't call something that isn't a fuction but a path (I'm in repl so I took the (builtins.fetchTarball ...) and put it in a variable already)
<pikajude>
oh, what does it print when you just type the variable name?
<numkem>
pikajude: the path in the nix store
<pikajude>
ok
<pikajude>
try
<numkem>
doing an ls on it has the content of the tarball
<bbigras>
Anyone has an idea why this build fails? `error: linker `aarch64-linux-gnu-gcc` not found` https://logs.nix.ci/?key=nixos/nixpkgs.88185&attempt_id=06cccd5a-9ad3-4b62-8c54-47f442e5c892
tarelerulz1 has quit [Remote host closed the connection]
<numkem>
pikajude: I spoke too soon, while the code works in repl, it doesn't in hydra. It says it can't reach the URL in restricted mode. that fetch is done in the `let` portion... Not sure how else I can do it
<T0pH4t>
ok, i'm thinking i can hack this a bit, but having a flag that determines if the package should fetch a tarball with the sha256 specified or use github, only the url for the tarball will be bogus, the idea being the sha256 will enable the binary cache to work
<srhb>
T0pH4t: Scary :3
tyang has joined #nixos
<T0pH4t>
i did say hack lol
<numkem>
srhb: I totally agree, it's just a bad habit to break
<T0pH4t>
i think having to maintain two separate pkg repos is worse, and I'm not going to give prod machines access to source
<srhb>
T0pH4t: I've been in your shoes, I ended up mirroring in the repo for local access through Other Means.
<T0pH4t>
srhb: i see, well thx for help
<tyang>
Hi! Recently I have upgraded from 19.09 to 20.03. Everything works fine except I can no longer switch to different virtual console using ctrl + alt + n, when I'm in X or wayland session. Could someone give me a hint what has changed?
<srhb>
tyang: Hrm, haven't seen anything related to that (nor do I have that issue). Does sudo chvt 2 work?
<cjpbirkbeck>
tyang: by n, you mean F{1-7}, right? i also upgraded, and i haven't that problem. what de/wm/wayland compositors are you running?
<tyang>
srhb: yes chvt works
<srhb>
tyang: Then I guess we should suspect something intercepting your keypresses
<tyang>
cjpbirkbeck: you are right, sorry for the typo. I meant F[1-7]
<srhb>
tyang: So what desktop environment are you using?
<dsal>
Is there a way to install a newer version of a package than my distro has? I see that nixos-20.03 has an old version of a thing, but nixpkgs-unstable has a newer version.
magnetophon has quit [Read error: Connection reset by peer]
<dsal>
Actually, not even necessarily install, just nix-shell at a newer version.
opthomasprime has joined #nixos
<srhb>
tyang: (And just to be certain, does your system have a Fn-lock key? It might interfere)
<tyang>
srhb: I'm using sway for wayland and xmonad under x (started with startx)
<dsal>
This is a lot nicer than what the FAQ says. :)
<srhb>
tyang: I know nothing about wayland, but I'm guessing the fact that it happens in both(?) embiggens my suspicion to your keyboard and lessens it to your DE.
<srhb>
dsal: Time for an update? :)
<dsal>
from 20.03?
user_0x58 has quit [Quit: Leaving]
<srhb>
dsal: Sorry, I meant the FAQ.
<tyang>
srhb: Yes it happens in both. I'm using a lenovo T470s. I have flipped ctrl and fn in BIOS. I have noticed that in virtual console, I need to press fn (should be actually ctrl in BIOS setting) + alt + F[1-7] in order to switch between console.
<dsal>
Oh, the FAQ... Yeah. I think it's probably valid, but your suggestion is exactly what I want for this case.
<srhb>
tyang: try fn+ctrl+alt+f1 :P
<tyang>
srhb: In x / wayland session I have tried both fn and ctrl, none of that works :/
<dsal>
nix-collect-garbage on nixos isn't so bad. On my mac, it'd be a bit devastating as it seems to want to build ghc for some of my projects.
<cole-h>
By setting `users.users.root.hashedPassword = null;`, password logins will be disabled for root, correct? Even for ssh?
vidbina has joined #nixos
<srhb>
cole-h: Only if you have mutableUsers false, iirc
<srhb>
dsal: Root everything you keep around
<infinisil>
cole-h: Probably also needs services.openssh.permitRootLogin
<dsal>
Yeah, I guess I need to figure that out. There are a few levels of indirection. Can I just make a stack project a root of some sort?
o1lo01ol1o has quit [Remote host closed the connection]
<cole-h>
And mutableUsers set to false just means I can change my password/groups/etc by modifying configuration.nix, right?
<srhb>
cole-h: With mutableUsers on (default), passwd will still work.
m4ts has joined #nixos
<srhb>
dsal: I don't know how stack + nix works, but you can root the ghc you're using by its path.
<cole-h>
infinisil: Well, I want root logins via SSH, but only with a key, so the current `prohibit-password` is fine, I think.
* infinisil
nods
<tyang>
srhb: I think you are right, it's fn-locked, but there's no LED indicator for fn lock. Sorry for the trouble and thanks for the help!
<srhb>
tyang: No problem. :)
<dsal>
srhb: how do you typically do something like that? I guess I could shove the package in home-manager or something.
<dsal>
Im' not even entirely sure what it *is*. stack just implies some ghc package that comes about.
<srhb>
dsal: That would be the probably nicer way. The primitive way is to point nix-store --add-root directly at the store path you want to keep
<srhb>
dsal: Do a long build and find its location in `top` :-P Engage the bodge!
gxt has quit [Ping timeout: 240 seconds]
magnetophon has quit [Read error: Connection reset by peer]
<dsal>
I only really want to keep it as long as I have projects that need it. But maybe that's reasonable. My computer can spend some time correcting my mistake if I'm wrong. :)
ssk has joined #nixos
<srhb>
dsal: I totally agree that a proper expression is the way to go. ;)
o1lo01ol1o has joined #nixos
<srhb>
dsal: I bet you can have stack run ghc from a given nix expr that you could root.
<srhb>
dsal: The manual probably has more information.
<dsal>
The ghc build thing is kind of a pain. I don't have a good mental model of what a root should look like, though.
<infinisil>
ssk: They aren't aimed at nixos users no. All you need is a Nix installation
<ssk>
oh neat!
<ssk>
I *think* this command is what they're asking to run `nix-build -K -E '(import <nixpkgs> {}).gfortran' --argstr system aarch64-linux` but it seems to just pull from my cache
tsrt^ has quit [Ping timeout: 240 seconds]
<srhb>
dsal: At its most basic level a gc root is a link from /nix/var/nix/gcroots to a store path
<ssk>
I'm not sure what the `-I nixos-config=/path/to/target/machine/nixos/config/copy` part should be for me
<srhb>
dsal: Its existence prevents the store path from being deleted.
<srhb>
(by garbage collection) that is
magnetophon has quit [Read error: Connection reset by peer]
<srhb>
dsal: When you eg. nix-build some expression, you've probably seen a result symlink
<dsal>
Oh yeah. That's what that is?
o1lo01ol1o has quit [Ping timeout: 272 seconds]
<srhb>
dsal: Usually, a so-called indirect gc root is created for result symlinks. That is, there's a link in gcroots to your result symlink (and in turn to the store path)
<srhb>
dsal: Once the chain is broken (eg. you delete result) there's no longer a root from gcroots to the store path, and the thing can freely be gc'ed
<infinisil>
ssk: I'm not sure why it tells you to build packages like that. I think `nix-build '<nixpkgs>' -A theRequiredPackage --argstr system aarch64-linux` should work too
<emily>
ssk: if the package was built and is in the cache then it'll be downloaded from there rather than recompiled locally
<infinisil>
I admit that I haven't fully read through the page, but all nix commands except `nixos-rebuild` work on non-NixOS as well
<dsal>
srhb: OK, thanks. I think I see what's happening here.
<{^_^}>
[nixpkgs] @veprbl pushed commit from @gnidorah to master « mame: 0.220 -> 0.221 (#88146) »: https://git.io/JfzZO
<ssk>
thanks emily and infinisil, I ran `nix-build '<nixpkgs>' -A pkgs.gfortran --argstr system aarch64-linux` and it seemed to build something. I wonder if `aarch64-linux` is actaully the system I want though. I'm trying to build for a raspberryPi, should I go into `systems.examples.raspberryPi` and pull that system string instead?
<infinisil>
ssk: Ah try `pkgsCross.raspberryPi.gfortran` instead then
<infinisil>
for the attribute
<infinisil>
Probably without the --argstr system
vidbina has quit [Ping timeout: 260 seconds]
<ssk>
ah, but I don't want to cross compile I want to use emulation (because I've tried cross compiling and run into some errors)
<dsal>
srhb: woo. Worked. I did a gc, removed my .stack-work, did a 'stack test' and a few downloads later, got results.
<srhb>
woo.
<infinisil>
ssk: Hm I see, not sure if the raspberryPi thing is needed then
<infinisil>
But really, I don't know much about different platform building
<ssk>
right, I just want to make sure I;m targeting the right architecture. I'll try to run the package I just build on the pi and see
<ssk>
thanks!!
<infinisil>
I'd just see where the pkgs.gfortran build goes
zupo has quit [Ping timeout: 246 seconds]
<{^_^}>
[nixpkgs] @ruuda opened pull request #88195 → libressl: add 3.1.1, default to it, remove 2.9 → https://git.io/JfzZV
<meh`>
could anyone point me to a recent "push luks keys over ssh in initrd" setup?
<bigvalen>
How does nixos do ZFS snapshots ? I was surprised to notice:
<bigvalen>
# zfs list|grep -c root/nixos
<bigvalen>
189
<bigvalen>
That seems..like a lot
<bigvalen>
I had zfs.autoSnapshot set to save .. 9 snapshots.
jgeerds has joined #nixos
magnetophon has quit [Ping timeout: 258 seconds]
mallox has joined #nixos
<numkem>
how do you make hydra create a channel? it used to work now it says it's empty (had to re-create the project)
turion has joined #nixos
agsdheidjd has quit [Ping timeout: 265 seconds]
<T0pH4t>
how would one use an ssh-config file with nix-build without disabling sandbox ?
magnetophon has joined #nixos
<T0pH4t>
kind of feel like ssh-config-file is pointless with nix-build otherwise...
mallox has quit [Quit: WeeChat 2.8]
<cole-h>
bbigras: Did you have any luck in tracking down that playerctl warning? If you're working on it, I'll hold off on approving it, but if not, I'll just go ahead and approve it.
<bbigras>
cole-h: No I didn't have time yet. ok thanks!
user_0x58 has joined #nixos
<gleber>
Has anyone seen `error 9 while decompressing xz file` and `HTTP error 200 (curl error: Error in the HHTP2 framing layer): retrying from offset ...` recently? I am seeing it on the most recent nixos-minimal ISO directly from Hydra using the the most recent `pkgs.nixUnstable` from nixpkgs-unstable. I tried applying all workarounds from the Aug'19 torrent of bug reports
<gleber>
This is happening on fresh VM in VirtualBox on Windows 10. I am trying to set it up as a quick and dirty remote builder
<T0pH4t>
srhb: great news, seems there are two functions fetchgit and builtins.fetchGit with the latter not supporting sha256. The former does but you have then get the ssh-config-file working, which when done allows a consumer of the packages to by pass hitting git when it finds bin in binary cache, thus avoiding any git access on a prod machine
<adisbladis>
T0pH4t: The latter is also a fixed-output of sorts, despite the lack of sha256
<Jonathan76>
Is there an easy way to start a nix-shell with specified channel, without modifying shell.nix?
<T0pH4t>
fixed-output?
<Jonathan76>
For one of the common shells which starts with `{ pkgs ? import <nixpkgs> {} }:`?
orivej has quit [Ping timeout: 265 seconds]
agam has joined #nixos
orivej has joined #nixos
rardiol has joined #nixos
<adisbladis>
T0pH4t: Depsite the fact that you don't give it a sha256 explicitly a sha256 is still in play. builtins.fetchGit clones outside of the sandbox into ~/.cache/nix and calculates the sha256 after the fact
<adisbladis>
Jonathan76: Set NIX_PATH, or in the case where you have `{ pkgs ? import <nixpkgs> {} }` you can pass an argument `nix-shell --arg pkgs 'import <nixpkgs> {}'`
<Jonathan76>
Thanks adisbladis: I was going to say, I have been wanting to do this for weeks, and just now I remembered that nix-shell has flags like `--arg`, but I still cannot get it to work..
rogue_koder has joined #nixos
<Jonathan76>
I was missing the `{}` which calls the import.
turion has quit [Ping timeout: 272 seconds]
seku has quit [Quit: Connection closed]
cjpbirkbeck has quit [Remote host closed the connection]
<agam>
Dumb question: I ran `nix search emacs`, got `nixpkgs.emacs26-nox` as one of the results, but when I try `nix-env -qa ...` with either "emacs26-nox", or "nixpkgs.emacs26-nox", I get the "... matches no derivations" error; but with "emacs-nox" it works fine, why is that?
<agam>
(I got what I wanted, the install works, it's all good, so ... this is really just more out of curiosity :-) )
<yimmy149>
agam: there are no derivations with emacs26 in the attribute name, so I don't think the -f <nixpkgs> is going to change the result. If you use -a, you need to match an attribute name. without the -a, you would use the package name
<{^_^}>
[nixpkgs] @sorokin opened pull request #88202 → expose gcov, gcc-ar, gcc-ranlib and others gcc tools in gcc-wrapper (fixes #86272) → https://git.io/JfzWR
<yimmy149>
I think generations are managed via symlink, so you can probably revert manually
<freeman42x[m]1>
yimmy149: how do I revert manually?
<{^_^}>
[nix] @edolstra pushed to auto-uid-allocation « Add "uid-range" and "systemd-cgroup" system features »: https://git.io/JfzWN
<freeman42x[m]1>
aight, I'd rather give up and create a new VM
<yimmy149>
I'm not on a nix system to check, but have a look in /nix/var/nix/profiles
<sshow>
I'm having issues with `fetchgit` when running nix-build, as I'm targeting a private git server, I'm getting "Please make sure [....] access rights. Host key verification failed." What ssh config does it use?
pjt_014 has joined #nixos
<{^_^}>
[nixpkgs] @kthielen opened pull request #88205 → build latest hobbes with gcc 9 → https://git.io/JfzWA
<tokudan>
hmm... what package would a build tool require?
<infinisil>
tokudan: Um, what build tool?
<tokudan>
infinisil, yeah, that's what I'm wondering about. I only have "/nix/store/x7fr0bvnwvqvr3zg60di9jxvfwimcw7m-bash-4.4-p23/bin/bash: no: command not found
<MichaelRaskin>
Wait, it's clear you need to read configure's output
<infinisil>
Huh, there sure are a lot of no's
<MichaelRaskin>
It has something like «checking for asdasd... no»
<MichaelRaskin>
(meaning asdasd is not found)
fresheyeball has quit [Quit: WeeChat 2.7.1]
<{^_^}>
[nixpkgs] @orivej-nixos pushed commit from @orivej to master « solfege: 3.22.2 -> 3.23.4 »: https://git.io/JfzlS
<infinisil>
Wait why does ,locate return so many results
<infinisil>
Locally I don't get any for nix-locate bin/no -w --top-level
page has quit [Quit: leaving]
<tokudan>
MichaelRaskin, yeah, configure output for vice is very verbose and it checks for a *lot* of software that's not required
<tokudan>
infinisil, what package is nix-locate in?
<infinisil>
,locate nix-locate
<{^_^}>
Found in packages: nix-index
<infinisil>
:)
<infinisil>
,locate
<{^_^}>
Use ,locate <filename> to find packages containing such a file. Powered by nix-index (local installation recommended) https://github.com/bennofs/nix-index
<tokudan>
ok, that output is more useful than getting "... and 89 more"
<infinisil>
The fact that there's 89 results means that something went wrong in the search and probably none of them are relevnat
<niso>
newbie question: i'd like to configure multiple dhcpd instances via the nixos module - what would be the best way to approach this?
<T0pH4t>
ugg so nix copy keeps giving me an error saying it can't copy a corrupted path. The package in question is from git and was prefetched with nix-prefetch-git to acquire the sha256. When i issue a verify path on the store path given it says the hash is different. Any ideas?
<T0pH4t>
i've blown away these directories twice...
<T0pH4t>
i have sandbox in relaxed mode on nix-build not sure if that makes any difference
MmeQuignon has quit [Ping timeout: 260 seconds]
<niso>
i think my options are: A) running dhcpd in a container B) forking the module and rewriting it to support multiple daemons
<azazel>
niso: goal? I don't remember the actual nixos conf for dhcpd, but usually one daemon can listen on multiple interfaces and serve different pools of IPs
fusion809 has joined #nixos
aveltras has quit [Quit: Connection closed for inactivity]
<niso>
azazel: my goal was to use use differnt subnets for .e.g. a container infra. The idea behind running multiple dhcpd instances was to make it as fault tolerant as possible
AmandaC has joined #nixos
<niso>
azazel: thus the idea was to run per subnet a dhcpd/nat/dns server
Desetude has quit [Quit: WeeChat 2.7.1]
<niso>
if i'd run it in one instance other subnets would be effected by changes (e.g. restarts of dhcpd)
cosimone has joined #nixos
<T0pH4t>
is there a way to have nix-build keep the .git directory. somehow a package on build is corrupting itself and it be good to know if files are being changed on build
AmandaC_ has joined #nixos
philr has joined #nixos
AmandaC has quit [Ping timeout: 256 seconds]
<infinisil>
T0pH4t: You can pass `leaveDotGit = true` to pkgs.fetchgit
<T0pH4t>
ah ok, thx will give that a shot
kvda has joined #nixos
growpotkin has quit [Ping timeout: 256 seconds]
kim0 has quit [Quit: Connection closed for inactivity]
ixxie has quit [Ping timeout: 265 seconds]
hmpffff has joined #nixos
Acou_Bass has quit [Ping timeout: 265 seconds]
ixxie has joined #nixos
morgrimm has joined #nixos
<T0pH4t>
ok so now its detecting the hash mimatch at build. I guess I would like to know how the sha256 generated by nix-prefetch-git is different from nix-build
<T0pH4t>
like how is that even possible?
<yimmy149>
Not an expert, but sometimes you need to pass --unpack to the prefetch command
emilsp has quit [Read error: Connection reset by peer]
lemsip has joined #nixos
Acou_Bass has joined #nixos
<T0pH4t>
yimmy149: whats that do?
<yimmy149>
I think it unpacks the src files instead of leaving that as a tarball. But it definitely changes the sha256 output, I had similar issue the other day.
<T0pH4t>
that came back with nothing
<T0pH4t>
hmm :/
hmpffff has quit [Quit: nchrrrr…]
<yimmy149>
you still need the --type sha256 parameter
morgrimm has quit [Ping timeout: 265 seconds]
rardiol has quit [Ping timeout: 240 seconds]
<T0pH4t>
i had: nix-prefetch-git --unpack --hash sha256 <repo> <rev> <hash>
maddo has quit [Quit: See ya]
<T0pH4t>
outputs an empty line
rardiol has joined #nixos
<infinisil>
--unpack is only needed for fetchFromGitHub or fetchzip I'm pretty sure
<infinisil>
And that with nix-prefetch-url
<infinisil>
T0pH4t: Pass --leave-dotGit
<infinisil>
Then you get a hash for a pkgs.fetchgit that has leaveDotGit = true
<T0pH4t>
infinisil: i did that, but it doesn't use the .git from the original repo
<infinisil>
Yeah it's not exactly the same, that's because the builds needs to stay pure
<infinisil>
So it needs to remove anything from the .git that's impure
<T0pH4t>
so leaving the sha256 changed the hash
<T0pH4t>
interesting, it matches the hash it expected
<T0pH4t>
so whats that mean, its accidently leaving behindthe .git folder?
<infinisil>
No idea what you're saying
<T0pH4t>
so i checked the sha of the path nix-store complained about
<T0pH4t>
and it dind't match teh shaw i was expecting until i added the --leave-dotGit to prefech
<T0pH4t>
so does that mean nix is accidently leaving behind the .git folder?
reallymemorable has joined #nixos
<infinisil>
Show all of the things you did with commands and their outputs
<infinisil>
It's hard to know what's going on otherwise
<infinisil>
T0pH4t: Nah not difficult, just put the fetchgit { .. } part into its own file and add `with import <nixpkgs> {};` before that
<infinisil>
Then `nix-build` that file
civodul has quit [Ping timeout: 272 seconds]
<T0pH4t>
ok
teto has quit [Ping timeout: 260 seconds]
<T0pH4t>
ugg nix just keeps yankin my chain, i update my other build with the new hash and now its changed the hash again :( ...... will work on that private build file now