<lukegb>
throwaway740928[: yeah, the nixpkgs learning curve is... pretty steep
proofofkeags has quit [Ping timeout: 252 seconds]
<Cadey>
oh wow i think it's actually building a kernel now
<Cadey>
:D
<samueldr>
if it evaluated, it likely will work, and give you "integration" with the out of tree modules :)
<lukegb>
writeScript is pretty simple, you can imagine it as a function that takes a filename and some content, and then returns you a path to a chmod +x file with that content
<lukegb>
it's... not quite that, but it's a decent mental model
<samueldr>
* integration limited to building them against the kernel tree linuxPackagesFor was ran against... no guarantees etc... etc...
<throwaway740928[>
lukegb: where is it documented?
<Cadey>
yeah i'm aware
<Cadey>
i just want it to work lol
muppeth has quit [Read error: Connection reset by peer]
<samueldr>
oh, btw I overexplain some things when I help, like here, it's to help future readers going through the logs with their own issues :) hi you!
<throwaway740928[>
so now there is a 'nftables-rules' file which will include the contents of the "${cfg.rulesetFile}" file
<throwaway740928[>
except that "${cfg.rulesetFile}" is 'nftables-rules' by default
<throwaway740928[>
so it will include itself?
<lukegb>
err, no
<lukegb>
for one thing: they have different filenames
<throwaway740928[>
they do?
<lukegb>
(they'll get filenames like /nix/store/hmx1c6bnibsq2185d0pm5nwjz5km0l9j-nftables-rules, and the hash at the beginning will be different for each of them)
<lukegb>
the "nftables-rules" bit is just to help make it more obvious what different things are, but it's not the *whole* filename
<lukegb>
(that's why it's important that e.g. pkgs.writeScript *returns* the path to the file)
<throwaway740928[>
ok
<lukegb>
and secondly, cfg.rulesetFile is, by default, a separate file which contains the text from cfg.ruleset
<lukegb>
Part of the point of Nix is that you can have multiple versions of software installed at once, and they won't collide because they live in separate places
<lukegb>
and this extends up to system config files - even if only one of them is active at once you can have a bunch of different versions in your nix store and roll forwards/back
<samueldr>
sheijk: please do, since it's the builtins, it would be the Nix repository
<qyliss>
so it spends ages building kernel modules that'll never be used
lambdaman has joined #nixos
<qyliss>
throwaway740928[: I recommend not sending multi-line messages, because they come through to IRC as links, so people have to click on them to read them and many won't bother
<throwaway740928[>
qyliss: thanks. good to know
<Cadey>
qyliss: yeah i'd love to make it just not build those modules but i think there isn't an easy way to do that lol
<qyliss>
Cadey: we don't even support MODULES=n :(
iceypoi has quit [Ping timeout: 260 seconds]
<qyliss>
your blog is cool btw :)
<throwaway740928[>
why does "selector 'apacheHttpd' matches no derivations"?
deltab has quit [Ping timeout: 246 seconds]
iceypoi has joined #nixos
<Cadey>
qyliss: thanks! you're actually witnessing the hacking process that leads up to one of these posts, in this episode it will be an implementation of /dev/printerfact!
lambdaman has quit [Remote host closed the connection]
ddellacosta has quit [Ping timeout: 240 seconds]
haritz has quit [Ping timeout: 260 seconds]
ddellacosta has joined #nixos
sjlnk has quit [Remote host closed the connection]
<rmcgibbo[m]>
thanks for the tip, eacameron. (https://github.com/rmcgibbo/pyfst/pull/1). someone should make a blog post documenting this or something. i use this configuration + cachix to get binaries to run on aws's aarch64 servers, which is quite effective.
<AmandaC>
I don't suppose I can get some eyes on this? (Tbh I totally forgot I even made it until I was looking through some backlogs...) https://github.com/NixOS/nixpkgs/pull/114393
<{^_^}>
#114393 (by AmandaCameron, 6 weeks ago, open): nixos/samba: Add a wants=network-online.target to the target.
lambdaman has quit [Remote host closed the connection]
<AmandaC>
I've been running something functionally equivalent on my homelab since, no unexpected side effects
<apache8080>
does nix have any way of setting the journalctl log location? it looks like journalctl only logs to /var/log/journal
veehaitch has quit [Ping timeout: 268 seconds]
<apache8080>
Right now my idea is to create a symlink with /var/log/journal to some other directory but that means I miss some of the kernel logs in journalctl
patagonicus1 has quit [Quit: Ping timeout (120 seconds)]
patagonicus1 has joined #nixos
carbolymer has quit [Quit: o/]
carbolymer has joined #nixos
mlen has quit [Ping timeout: 246 seconds]
superbaloo has quit [Ping timeout: 246 seconds]
cjay has quit [Ping timeout: 246 seconds]
bukkitgerman8608 has quit [Ping timeout: 246 seconds]
<apache8080>
I am trying to build nixos without an initrd so I set boot.initrd.enable to false but now I run into an error where system.build.initialRamdisk is missing
lambdaman has quit [Remote host closed the connection]
lambdaman has joined #nixos
ddellacosta has joined #nixos
lambdaman has quit [Ping timeout: 260 seconds]
lambdaman has joined #nixos
ddellacosta has quit [Ping timeout: 252 seconds]
jophish has quit [Ping timeout: 240 seconds]
jophish has joined #nixos
lambdaman has quit [Ping timeout: 246 seconds]
jonringer has quit [Ping timeout: 258 seconds]
beertoagunfight has joined #nixos
respawn_ has quit [Quit: Leaving]
zupo has joined #nixos
zupo_ has joined #nixos
lambdaman has joined #nixos
<afontain_>
I wonder if someone has a WIP libadwaita derivation
orivej has quit [Ping timeout: 252 seconds]
aaaaaaaaaaaareis has joined #nixos
zupo has quit [Ping timeout: 240 seconds]
lambdaman has quit [Ping timeout: 252 seconds]
<aaaaaaaaaaaareis>
I'm trying to add a git plugin to the plugin list. But the `update.py` script crashes after a couple of minutes of spamming `HTTP Error 429: too many requests, Retrying in 3 seconds...` Is this a known issue, or am I doing something wrong?
lambdaman has joined #nixos
supercoven has quit [Read error: Connection reset by peer]
lambdaman has quit [Remote host closed the connection]
lambdaman has joined #nixos
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
supercoven has quit [Max SendQ exceeded]
supercoven has joined #nixos
ddellacosta has joined #nixos
ahmed_elgabri has quit [Ping timeout: 248 seconds]
beertoagunfight has quit [Remote host closed the connection]
beertoagunfight has joined #nixos
beertoagunfight has quit [Remote host closed the connection]
<ploupiboulba[m]>
does anyone know where to ask systemd questions ? haskell-language-server crashes but the coredump gets truncated. I suspect it's due to the MaxSize config, would like to see the default config for systemd-coredump (I have none specified) but can't find where. something like `systemctl cat systemd-coredump@.service` doesn't show a [CoreDump] section.
aaaaaaaaaaaareis has quit [Quit: Connection closed]
<wrench>
Does anyone have experience with nftables under nixos? libvirtd always tries to load the ip_tables module which leads issues with the nftables service
sangoma has quit [Quit: WeeChat 3.1]
ahmed_elgabri has quit [Ping timeout: 258 seconds]
<pie_>
infinisil: ok just one more time to make sure; so if i have containers.a and containers.b, i would do that to add stuff in one expression to all arbitrary containers.*?
sss2 has joined #nixos
Izorkin has joined #nixos
ddellacosta has joined #nixos
<pie_>
thats a bit involved. the module system is obscure :P
<pie_>
or nonintuitive at least
<pie_>
ok so the keywords are "extending submodules"
<pie_>
infinisil: any advice for me having to nag you less?
fuiltilt has quit [Quit: WeeChat 2.9]
<pie_>
not sure how i would have figured that out without being a module expert, and not sure how id do that other than reading all the source and doing modules a lot and asking you stuff i dont understand anyway? :P
<infinisil>
pie_: Maybe you could write documentation on the things you learn from IRC :D
<pie_>
i keep telling myself i should keep a journal but i keep not :V
<pie_>
too much jumping around betwen stuff
<pie_>
but its an idea
<infinisil>
"Not docs! Anything but docs!" - programmers
<pie_>
the threads i linked are in a way docs x)
<pie_>
couldnt find them though
<pie_>
infinisil: you should publish a book at packt :P
<Ke>
the sad thing, when you write docs, people assume they do not exist
<Ke>
and when people read your docs, they document the wrong thing
<pie_>
mm
<pie_>
software is big and complicated and its hard to find things
<Ke>
good documentation is hard, unless you are writing haskell, in which case documentation is not really needed
<infinisil>
Ke: It sure is!
<pie_>
so you have to hope you can document core principles that are central nodes with low inference distance to other stuff
abathur has joined #nixos
<pie_>
(i.e. compression)
<infinisil>
Docs might not be needed in Haskell for showing what functions accept and return, but they are needed to explain concepts, give architectural overview, give examples, etc.
<pie_>
plenty of things with the same signature that do different stuff :p
andreas303 has quit [Remote host closed the connection]
<rgsada723>
How can I login to vscode when using liveshare that's not installed on the system. Generally what I do is add a vscode instances with the correct plugins to a nix-shell. But then when I try to login I cannot login since my browser opens a new vscode window that is installed system wide. I also cannot use the token since there is no option to put into the token. Has anyone gotten this to work with nixos?
andreas303 has joined #nixos
graf_blutwurst has joined #nixos
nicoo has quit [Remote host closed the connection]
<graf_blutwurst>
Hi I was wondering if someone can give me a hand with flakes. So I'm using flakes mainly to maintain a development shell for our project. We also want to use dhall on the CI pipeline based on a nix image (so that the versions are shared between dev shell and CI) previously with niv that was no problem as I could just have two shell.nix files that point to the same niv dep pinning. How could I do that with flakes? I can't see a w
<graf_blutwurst>
choose the particular shell.nix i want to run with nix develop in regards to a flake
<whip>
Hey i'm deploying a python app with nix and it creates a result/bin/myapp script that runs my app, how can I make it so that this script is in root and not in result/bin
<sterni>
that's not really fixable your app will always be in /nix/store somewhere
<sterni>
and nix-build creates a symlink
waleee-cl has joined #nixos
<whip>
creates a symlink to nix/store with result/bin/<name>
<whip>
but can we change that symlink in the config
<sterni>
what you can do is have something like let drv = buildPythonApplication { … }; in runCommandLocal "my-app" {} "ln -s ${drv}/bin/appname $out"
<whip>
to be <name> -> /nix/store
<pie_>
they want the former <sterni> whip: do you want ./result to be your application or ./result/<name>
<sterni>
and then nix-build --out-link target-name
<pie_>
(i think(
<pie_>
or maybe that *and* --out-link
<sterni>
buildPythonApplication creates a directory under /nix/store which has a bin subdir
<pie_>
i was thinking of figuring out how buildPyhtonApplication puts it in bin but making that symlink is a lot simpler >_>
<sterni>
nix-build doesn't know if any drv creates a dir or file in the nix store it always makes a symlink from result (or --out-link) to the store path
<pie_>
why are you using runCommandLocal instead of runCommand?
<sterni>
so it's not a link from result/bin/<name> -> /nix/store/<foo>/bin/<name>, but a symlink result -> /nix/store/<foo>
<sterni>
pie_: if you have a remote builder it will build the symlink remotely and download it again which is much slower than invoking ln
<sterni>
or habit
<pie_>
aha.
<whip>
I might have not explained well but rn I have ./result/bin/name -> nix/store/... and i want ./name -> /nix/store/...
<whip>
just move the symlink to ./ of the app
<pie_>
you can do "./result" -> /nix/store/... with the ln thing
<pie_>
you need to do --out-link name to do "./result" => ./name
<sterni>
whip: result is always the symlink
gustavderdrache has joined #nixos
<pie_>
anyone know offhand how nix gets added to PATH?
Jd007 has joined #nixos
<whip>
okay thanks
<sterni>
pie_: what do you mean
<pie_>
sterni: i want to run nix-build from a systemd service and i want to add it to PATH the same way the system does
<sterni>
pie_: well it's in systemPackages and thus in /run/current-system/sw/bin
<pie_>
sterni: yeah but its not just added to the default systempackages list (requiredpackages / default packages) - but i did end up finding the source, see above
<sterni>
why are you invoking nix-build in a systemd service
<sterni>
sonuds curse
<sterni>
you can just build it when you are building the system config …
<pie_>
i guess what i could do is limit myself to two imports of nixpkgs by adding one to _args that doesnt depend on the rest of the module system... (maybe?)
veehaitch has quit [Quit: leaving]
<pie_>
sterni: i am also the kind of making my evals way too damn long...
<pie_>
not good for dev flow..
<sterni>
well that is entirely possible without rube goldberging
<eacameron>
Does cachix have an on-premise solution?
Wizek has joined #nixos
acarrico has joined #nixos
<simpson>
No, but in that situation, you wouldn't need one. Just have a local trusted binary cache on the intranet.
<drozdziak1>
eacameron Where does your usecase intersect with cachix but not just hosting your own Nix cache?
<simpson>
drozdziak1++
<{^_^}>
drozdziak1's karma got increased to 1
<drozdziak1>
:)
Nickli has quit [Ping timeout: 276 seconds]
tv has joined #nixos
<pie_>
uh
<pie_>
_HOW_
<pie_>
Apr 15 16:54:12 nest pxe_container-start[762]: file 'nix/fetchurl.nix' was not found in the Nix search path (add it using $NIX_PATH or -I), at /nix/store/dydwjmi4z5kr2ik8k8j4bcy65mk8sn5g-nixpkgs-src/pkgs/stdenv/linux/bootstrap-files/i686...
ericsagnes has joined #nixos
<pie_>
<nix> is always present??
<pie_>
ok maybe i did something weird with scopedimport but im still pretty sure that should be in there, hasnt broken anywhere else yet
turlando has quit [Quit: Leaving]
<pie_>
ook yeah that doesnt look right x)
<pie_>
{ path = "/nix"; prefix = "nix"; }
sangoma has joined #nixos
rgsada723 has quit [Quit: Leaving]
<pie_>
still not sure how i managed that though
<pie_>
arcnmx: bonk?
Raito_Bezarius has quit [Ping timeout: 260 seconds]
beertoagunfight has quit [Ping timeout: 246 seconds]
<pie_>
hm i think i see, nixPathList only gets called if its a list, if its not a list it doesnt get called and <nix> isnt added
<pie_>
or somethin
<eacameron>
drozdziak1: I'm not sure yet :D
<eacameron>
drozdziak1: Right now the best justification I have is "laziness"
<drozdziak1>
eacameron: Yeah so cachix is just a hosted solution for a nix binary cache. The only possible headache is probably making sure everyone has your private cache keys among the trusted ones.
novafacing has quit [Quit: novafacing]
<pie_>
actually no thats still not right because it should have broken sooner then...
<eacameron>
drozdziak1: Cachix also has tooling on GitHub actions etc to make it easy to integrate. But I suspect I'll just have to roll my own.
<drozdziak1>
eacameron: So in case of your own cache the advantage is that all you have to do is successfully build the selected derivations. That's roughly what Hydra does. You just have to set up your derivations so that checks are run as well as part of your build
<drozdziak1>
I name this as an advantage because for a relatively small deployment, especially an internal one all of that can happen on one machine
JJ has joined #nixos
<drozdziak1>
i.e. building the derivations is equivalent to hosting them
<simpson>
eacameron: Is your actual situation an on-premises or an in-the-cloud situation? Or are you exploring possibilities for something still being designed? I feel that we could answer your actual question directly.
<neonfuz2>
Hello, just wondering if gnome 40 is planned to be in 21.05 or not? anyone know?
<hexa->
it is
<hexa->
it is why 21.05 is 21.05 and not 21.03
<neonfuz2>
cool, is it working in unstable yet?
<neonfuz2>
because the last release was 2 months late I think?
<hrnz>
it's not merged yet
<hexa->
no, there is an open pull request though, see #117081
<BlackBeans>
If I want a later version of a package that is in the repository, how can I do that?
<BlackBeans>
If I have a tarball containing the version of the package I want
<{^_^}>
[nixpkgs] @sternenseemann pushed to haskell-updates « haskellPackages.essence-of-live-coding-warp: fix eval and build »: https://git.io/JOca1
<neonfuz2>
BlackBeans: you can define package overrides
<BlackBeans>
I am a noob with overrides, and the actual code from the package script seems quite complicated
<cole-h>
chances are you only need to override the `src` attribute
<BlackBeans>
because the package I want to install – Anki – relies both on Rust and Python
<neonfuz2>
its not so hard
<neonfuz2>
oh yeah anki, I tried to bump anki it's difficult
<BlackBeans>
x)
<neonfuz2>
there's already an issue on nixpkgs tracking this, people have tried
<neonfuz2>
I think someone packaged the binary though
sangoma has joined #nixos
<BlackBeans>
there is a package, but it's for Anki 2.1.15
<neonfuz2>
yeah there's anki-bin in nixos-unstable
<neonfuz2>
anki-bin is 2.1.43
<neonfuz2>
it's in nixpkgs-unstable
<BlackBeans>
that would require switching to a new channel?
<neonfuz2>
you can install a single package from another channel, it's a bit more involved though
<{^_^}>
[rfc39-record] @grahamc pushed commit from rfc39 to main « Automated team sync results. »: https://git.io/JOcVg
<neonfuz2>
right now do you just use nix-env to install stuff?
<neonfuz2>
or do you edit nix config and rebuild?
<BlackBeans>
no, I try to keep everything declaratively
<eacameron>
simpson: gchristensen I'm bringing nix into our company but the CTO doesn't feel comfy with the cache being public and/or being off-premise. We use Azure currently so I'm trying to see what our options are to bringing this into our cloud.
<BlackBeans>
and, if possible, since Anki isn't a system application I would like to install it with home-manager
<neonfuz2>
ok, so I have a declarative way of doing this that i use
<gchristensen>
eacameron: gotcha, you can use a private s3 bucket very easily, or min.io
lambdaman has quit [Remote host closed the connection]
ahmed_elgabri has quit [Ping timeout: 260 seconds]
oxalica1 has joined #nixos
<BlackBeans>
now I get an error
<BlackBeans>
`error: value is null while a set was expected, at /nix/store/hash-nixpkgs-unstable-version/nixpkgs-unstable/pkgs/top-level/default.nix:74:18`
bitmapper has quit [Quit: Connection closed for inactivity]
<neonfuz2>
I usually keep my user channels empty because it defaults to root's channels, otherwise you have to upgrade both
<BlackBeans>
ah, if I try removing the `config = ...` attribute from the set passed in argument to the import, it seems to be working
<neonfuz2>
oh, hmm
<BlackBeans>
at least, it hasn't reported errors yet
<BlackBeans>
it's downloading stuff
<neonfuz2>
do you have this in a file separate from config.nix and importing it?
<BlackBeans>
it's not in config.nix
<BlackBeans>
it's in home.nix
<BlackBeans>
(actually, it's in package.nix, which is imported by home.nix)
<BlackBeans>
but I have no config.nix
<neonfuz2>
oh hmm, is this home manager or something? I've never used that
<BlackBeans>
yes, it's home manager
<BlackBeans>
although the difference is not completely clear to me...
<neonfuz2>
oh, well maybe configuration is different in that case. nixpkgs takes a configuration, and this is just passing your system config to the imported nixpkgs. If you don't pass anything it'll take defaults
conkker has joined #nixos
<neonfuz2>
I'm not sure if it's different
<neonfuz2>
I just don't have home-manager experience so idk
<simpson>
eacameron: Aha. Makes complete sense now. FWIW there's at least one more option: No caches, and make each developer rebuild stuff. This reduces the problem to a hopefully-already-solved source-control problem.
<BlackBeans>
ok...
<BlackBeans>
but what config are we talking about?
<BlackBeans>
the /etc/nixos/configuration.nix?
Wizek has quit [Quit: Connection closed for inactivity]
<neonfuz2>
mostly options for configuring nixos systems, why it may be irrelevant to home-manager
<BlackBeans>
ok
<BlackBeans>
well, it worked
<BlackBeans>
thanks!
<neonfuz2>
okay good
<edrex>
Is anyone aware of incompatibility issues between Arch Linux OpenSSH client and NixOS OpenSSH server? I am observing something weird: over double NAT with port forward, Arch can't connect (but other clients on the same network [OSX, edgeos] can). Sorry to ask here, just wondering if anyone else has seen this
Mateon2 has joined #nixos
<gchristensen>
edrex: try running with -vvvv ?
novafacing has quit [Quit: novafacing]
<lordcirth>
edrex, Does Arch have a newer ssh version than the other clients?
novafacing has joined #nixos
Mateon1 has quit [Ping timeout: 240 seconds]
Mateon2 is now known as Mateon1
<edrex>
oh, only went to -vv :) that did indeed add a new output line before the hang
yjftsjthsd has joined #nixos
<edrex>
lordcirth: yeah, newer than the other clients and one minor newer than the nixos server: `pacman -Qi openssh -> 8.5p1-1, ls -l $(which sshd) -> /nix/store/51n2b5v971a3z66izqdyflh5m64gmy38-openssh-8.4p1/bin/sshd`
bgamari has quit [Ping timeout: 260 seconds]
<edrex>
i can connect when i'm on local network, just not over double nat with the port forward. Looking through release notes and arch threads...
mkaito has quit [Quit: WeeChat 3.1]
bgamari has joined #nixos
BlackBeans has quit [Quit: Connection closed]
<pie_>
any easy way to get a bash with a cleared env?
EddyEdgy has joined #nixos
<gchristensen>
pie_: env -i bash -c $(which env)
<pie_>
cool thanks, guess i should read the env man page closer
<pie_>
oh. yeah. literally the first flag :I
<pie_>
huh. signals stuff. hm.
oxalica2 has joined #nixos
<etu>
sofus[m]: There's now a workaround for the broken screen sharing :)
<eacameron>
Is there a way to disable signature checks on a substituter? We're going for a private cache so having to hand out access to the cache *and* a key to verify signatures seems redundant. If someone has access to upload to our cache, they could have signed with our key too.
<gchristensen>
add `?trusted=1` to the end of its URL
<eacameron>
REALLY...
JJ has joined #nixos
<eacameron>
gchristensen: That almost seems too easy lol
<gchristensen>
:)
<srhb>
what
<srhb>
why is that possible o_o
<MichaelRaskin>
I think there are some cache/store «URL parameters» that are magic for Nix
<MichaelRaskin>
It's not a bug, it is a feature hopefully no easier to trigger than arbitrary reconfiguration of Nix substituter trust
<gchristensen>
yeah, you need to be rootish to set it
<MichaelRaskin>
«root-ish» rates pretty high on the list of words to duck first and investigate later
<MichaelRaskin>
(this combination of vagueness and destructive capability)
<srhb>
Love it. But OK, thankfully.
<gchristensen>
lol
<srhb>
gchristensen: btw, didn't get back to you, but I've tried to put some time for those issues into the next sprint. Hasn't been planned yet, so we'll see. :)
<edrex>
research on ssh issue didn't come up with much. Probably should just get wireguard going and then I won't care.
<jbetz_>
what is the correct way to start postgres server on NixOS with a file system backup of a database?
<jbetz_>
I've tried setting to services.postgresql.dataDir to the backup directory, but the pre-start script requires that the data directory is empty. I've also tried using pg_ctl directly, but it errors with `pg_ctl: directory "my_backup" is not a database cluster directory`.
alexherbo2 has quit [Read error: Connection reset by peer]
jarkad has quit [Remote host closed the connection]
<noonien>
it's not mentioned in the PR/commit, but this change helps improve security for nixos systems.
<noonien>
since accessing the libvirt socket can basically give users root access, (like the docker socket does),
<noonien>
and, because due to the way the module is written, the only way of accessing the socket as a non-root user, is to add it to the libvirt group, which basically makes `sudo` authentication useless.
<noonien>
the change i proposed, fixes this by allowing the necessary settings to be made so that libvirt can do SASL authentication, using PAM, which means adding a user to the libvirt group doesn't automatically give the user non-authenticated root access
justanotheruser has quit [Ping timeout: 258 seconds]
ahmed_elgabri has joined #nixos
wnklmnn has joined #nixos
justanotheruser has joined #nixos
VideoGameEnjoyer has quit [Ping timeout: 240 seconds]
VideoGameEnjoyer has joined #nixos
ahmed_elgabri has quit [Ping timeout: 250 seconds]
novafacing has quit [Remote host closed the connection]
<sterni>
orion: that's not possible
<sterni>
the hash is a hash of the compiled dependencies
<sterni>
so unless you can figure out what the compiled result is without compiling, nope :p
<orion>
I see. It's embarrassing to submit a PR to nixpkgs and have the build fail because of a wrong cargo hash. :/
<orion>
sterni: That is the overarching issue. Do you know of a way to prevent that in particular?
lambdaman has quit [Remote host closed the connection]
zupo has joined #nixos
<orion>
For example, could it be related to the nix-channel I am using?
<sterni>
orion: maybe check the Cargo.lock if it contains anything suspicious like fetching HEAD from a git repo or something
<sterni>
orion: sometimes also an outdated version of a fixed output derivation can be cached in the store, in which case you need to invalidate the hash once locally to get it to notice it
<sterni>
orion: you can do this by replacing the hash you have with lib.fakeSha256
<tpw_rules>
just to make sure, did the hash pass on your machine and unexpectedly fail in the PR?
lambdaman has joined #nixos
<tpw_rules>
if so, then sterni's suggestions are probably correct
<orion>
tpw_rules: Yes.
<tpw_rules>
did you change the cargo versions and stuff without changing the hash?
xyz81 has quit [Ping timeout: 240 seconds]
JJ has joined #nixos
malorie has joined #nixos
<orion>
tpw_rules: It's possible I did so without knowing it.
<orion>
Does the build bot use the unstable channel?
<tpw_rules>
the channel shouldn't matter. assuming there isn't another place in nixpkgs with the wrong hash
<tpw_rules>
although hm, it looks like it was a transient issue? the guy who tried was able to get it built in that last comment
<orion>
I built it on my company's build server.
<malorie>
hi all, nixos noob here. I've recently upgraded my system for the first time, and now xwallpaper isn't working anymore, due to a bug in libseccomp, which has already been fixed though, afaict. is there anyhting I can do right now, or do I just need to wait until the dependency gets bumped in the channel?
<orion>
If it's a hash of compiled dependencies, wouldn't I need a different hash for, say, OSX, arm64, etc?
<AmandaC>
the hash is for the sources of the dependencies
<AmandaC>
their versions, etc
<AmandaC>
usually you get the new hash by setting it to lib.fakeSha256 and then changing it to whatever the resulting hash is in the mismatch message
MatrixBot22 has joined #nixos
mauli has joined #nixos
vidbina has quit [Ping timeout: 240 seconds]
<orion>
Ah, so it's platform independent.
VideoGameEnjoyer has quit [Remote host closed the connection]
<jbetz_>
`pg_ctl: directory "my_backup" is not a database cluster directory` <- turns out the backup was using some Docker volume structure, so the actual data directory was in my_backupr/_data. anyway, got it working now.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]