<spease>
What's the best venue to ask about nix on darwin questions? Not necessarily nix-darwin questions, but using the nix package manager on darwin for darwin things (iOS etc)
<emmanuel_erc>
So I've learned with exportReferencesGraph attribute that you can use to access the closure of a derivation. I have two questions about this? A) Can you mutate the closure during the corresponding mkDerivation invocation? B) Can you still do task A while accessing the exportReferencesGraph in a overrideAttrs call?
<Orbstheorem>
Hi o/ When working on an app, I often use a nix-shell to wire in an adequate build environment. I have some integrations in my vim config (e.g. language server). Something that's been bugging me for a while is that whenever I need to add a new library, I have to close vim exit the previous shell, and re-enter the new shell and reopen vim. I wonder if there's a better way to do this (such as per-shell profiles,
cr4y1_ has joined #nixos
<Orbstheorem>
à-la nix-env).
<aasg>
I've been using direnv for development. It integrates quite nicely with nix
<Orbstheorem>
I've though about saving my vim session and add an alias `svim = nix-shell --run 'vim -S session.vim'`, which would at least ease the pain. I'd still have to close vim though.
mallox has joined #nixos
<Orbstheorem>
aasg: I don't think that solves my issue. If I understand correctly, direnv will load upon chdir'ing and place me in the right shell (exec?). Correct me if I'm wrong, but modifying my shell.nix afterwards will not magically put already-running applications in the new profile.
<aasg>
hm, I don't think direnv will change the environment of running programs indeed (aside from the shell, which it can update - might need to run `direnv reload` so it picks up the changes however)
<aasg>
I'm not familiar enough with vim sessions but looks like there's a vim plugin available that might do the trick
ericsagnes has joined #nixos
<joesventek>
Orbstheorem, maybe lorri can solve your issues?
sangoma has quit [Read error: Connection reset by peer]
sangoma has joined #nixos
<hsngrmpf[m]>
Is it possible to inject control sequences for colored output into `builtins.trace` statements?
vandenoever has joined #nixos
vandenoever has joined #nixos
vandenoever has quit [Changing host]
sangoma has quit [Ping timeout: 258 seconds]
antonyoneill has joined #nixos
cantaloupe has quit [Ping timeout: 258 seconds]
sangoma has joined #nixos
<antonyoneill>
Hey, good morning. I started my nix journey over the weekend and have installed the nix package manager on my work ubuntu machine. I'm not yet ready to take the nixos plunge. I've got a question about nix-shell GUI apps.. In my scenario, I have a dotnet project that is a one off, I'd like to install the dotnet requirements, and the JetBrains Rider
<antonyoneill>
IDE. With my current setup I have to head to the project folder and run rider from there. Is it possible to use makeDesktopItem inside of a mkShell? If yes, does it end up in my general ~/.nix-profile? Would each desktop item need to be unique?
<{^_^}>
[nixpkgs] @Ma27 pushed to master « graylog: expose NixOS test via `passthru` »: https://git.io/JTtaS
<joesventek>
So I think I want to get rid of nix-channels on my NixOS installations. Are flakes usable yet? Or should I better go with something like niv or even adding nixpkgs as git submodule for the time being?
is_null has joined #nixos
tabor has quit [Remote host closed the connection]
Izorkin has quit [Ping timeout: 264 seconds]
zupo has joined #nixos
<ehmry>
joshmeredith: I think flakes is ready for nixos installations if you want to try, but you would probably still use channels for your user profile
<ehmry>
I'm using a single flake repository to manage multiple machines
karolus has quit [Ping timeout: 256 seconds]
Ikositetrachor has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @zowoq pushed 2 commits to release-20.09: https://git.io/JTt63
sangoma has quit [Read error: Connection reset by peer]
noonien has joined #nixos
<noonien>
hello folks!
supercoven has joined #nixos
<noonien>
i'm trying to make my root ephemeral, however, i've bumped into some issues stage-1 seems to not be able to mount /var/log (a bind mount), here's the journal for stage-1-init: http://ix.io/2Azi and here's my hardware-configuration.nix: http://ix.io/2Azh
<noonien>
the /etc/nixos bindmount seems to work just fine
<{^_^}>
[nixpkgs] @domenkozar pushed 2 commits to release-20.09: https://git.io/JTt1Q
ericsagnes has joined #nixos
<simonpe^^>
I'm having some problems when downloading cached builds from my build server, I'm getting hash mismatches for a path. Who calculates this hash? Is it nix-serve or is it stored somewhere on disk when the derivation is built?
<simonpe^^>
I don't even know where to start digging
<simonpe^^>
the same problem appears on several clients so it's probably something on the server end of things
<NobbZ[m]>
Probably inputs differ?
<simonpe^^>
nix-instantiate gives me the same derivation on all machines
<simonpe^^>
idk if that answers your question
<{^_^}>
[nixpkgs] @edolstra merged pull request #100363 → flake.nix: allow inclusion of `nixpkgs` as `path:/.../` → https://git.io/JTLoF
<simonpe^^>
the specific error I'm getting is `hash mismatch importing path '/nix/store/s0js6iyw81zlsxvbw9fsxh57nfh90zc9-rbx-core-uc500x'`
<{^_^}>
[nixpkgs] @TredwellGit opened pull request #100402 → nixos/xserver: use modern video drivers → https://git.io/JTtMs
<noonien>
added "boot.initrd.preFailCommands = lib.mkAfter ''mount'';" to my configuration.nix, and got this output: http://ix.io/2Azu
<noonien>
it seems that indeed turbotank/safe/persist is not mounted at that point
<srhb>
simonpe^^: Since several machines are experiencing it, it sounds like the path changed on the substituter itself, eg. by being gc'ed and rebuilt or by disk corruption. See eg. https://github.com/NixOS/nix/issues/1885
<{^_^}>
nix#1885 (by domenkozar, 2 years ago, open): hash mismatch in downloaded path
<srhb>
simonpe^^: Maybe you can check whether that's it. :)
<noonien>
i'm guessing this has something to do with the order in which the filesystems are mounted. the bind mount is happening before the source filesystem is mounted. is there a way to fix this?
vidbina has quit [Ping timeout: 258 seconds]
<noonien>
/etc/fstab seems to be in the correct order: http://ix.io/2Azw
<simonpe^^>
srhb: okay, well I could try a complete rebuild by removing /nix/store
<noonien>
oh, nvm, /etc/fstab is not in the correct order either, /etc/nixos is mounted before /persist
<srhb>
simonpe^^: That shouldn't make a difference if it's that issue.
<srhb>
simonpe^^: But you can try rebuilding the specific path and compare the narhash.
<simonpe^^>
srhb: how would I get the narhash?
turlando has quit [Quit: Leaving]
<srhb>
simonpe^^: I believe nix path-info --json will emit it
<simonpe^^>
srhb: I don't have that binary-cache-v3.sqlite mentioned in the GH issue, this is not Hydra but a Jekins slave running Ubuntu and nix-serve
<simonpe^^>
thank you
<srhb>
simonpe^^: Hydra vs. Jenkins also shouldn't matter, only whether the served path changed hash (eg due to server side corruption or rebuild)
rprije has quit [Ping timeout: 265 seconds]
<simonpe^^>
srhb: its funny but the narHash seems to be none of the "wanted:/got:" from my error message
MmeQuignon has joined #nixos
<simonpe^^>
so there are 3 hashes involved all of the sudden
<srhb>
simonpe^^: Sounds like it may be binary irreproducible (which is often the case, sadly) which may increase the likelihood of this happening, but shouldn't be the cause itself
<srhb>
simonpe^^: (Again, only if the path changed on the server should this conceivably happen)
<simonpe^^>
srhb: it is definitely binary irreproducible
<simonpe^^>
but where is the client getting its "wanted:" hash from, and the "got:" hash?
<srhb>
simonpe^^: You can try killing your local caches in ~/.cache/nix/binary* and I believe /root/.cache/nix/binary* -- frankly I can never remember when which is used
<simonpe^^>
srhb: do you mean on the build server or on the client?
<srhb>
simonpe^^: If that does not get rid of the problem, then I guess either the server is truly serving a wrong narinfo (hash doesn't match the actual path) or something entirely different is going on.
<srhb>
simonpe^^: client :)
<simonpe^^>
I'm not really understanding how this hash verification works so it makes it a bit hard to draw conclusions :P
bahamas has joined #nixos
xelxebar has quit [Remote host closed the connection]
<srhb>
simonpe^^: Sorry, I think my memory of this issue is too vague to give you any very clear directions here. Short of nuking the binary caches and trying again to determine if it's a mismatch between server and client-side, I don't know :)
<simonpe^^>
srhb: I'm pretty sure there was a rebuild, so maybe it's nix-serve that's operating with an outdated cache. I did a complete rebuild on this machine a while back but I didn't touch any binary caches. Do you know if nix-serve uses the same binary cache database as you mentioned?
<simonpe^^>
In that case maybe wiping the binary cache on the server would fix the problem
<{^_^}>
[nixos-homepage] @garbas pushed 28 commits to master: https://git.io/JTtHs
<{^_^}>
[nixos-homepage] @garbas pushed 0 commits to redesign-community: https://git.io/JTtHZ
alp has joined #nixos
bahamas has joined #nixos
cfricke has joined #nixos
virus_dave_ has joined #nixos
<simonpe^^>
I don't know what other project than Nix that makes it possible to move the entire development environment of a large project to windows before afternoon coffee
<amercan>
Hey, a while ago I heard that NixOS has /bin/sh and /usr/bin/env links for compatibility purposes, did something change about it?
<clever>
amercan: nope, they are still there, but only outside the sandbox
xdefrag has joined #nixos
sangoma has quit [Read error: Connection reset by peer]
<inferencerules>
is there a means of authenticating to github (git or https) when running a nixos-rebuild using flakes? i'm currently in china and can't complete a --flakes rebuild unless i vpn tunnel out
<amercan>
By sandbox do you mean something like firejail or does nix use something else?
<inferencerules>
or, i guess a personal token would work too
<amercan>
(qubes os style etc. etc.)
bbb1 has joined #nixos
bbb1 has quit [Client Quit]
<jmercouris>
any reason why USB C charging won't work in NixOS?
<meh`>
jmercouris, it works for me
<jmercouris>
well, it works for me too... on Windows
<jmercouris>
so my hardware is clearly functional
mmohammadi981266 has quit [Quit: I quit (╯°□°）╯︵ ┻━┻]
Ikositetrachor has quit [Ping timeout: 240 seconds]
sangoma has joined #nixos
<emmanuel_erc>
I figure I should ask this question again, since I asked late at night (at least in my timezone).
<emmanuel_erc>
So I've learned with exportReferencesGraph attribute that you can use to access the closure of a derivation. I have two questions about this? A) Can you mutate the closure during the corresponding mkDerivation invocation? B) Can you still do task A while accessing the exportReferencesGraph in a overrideAttrs call?
mmohammadi981266 has joined #nixos
Ikositetrachor has joined #nixos
domogled has joined #nixos
_I3^RELATIVISM has joined #nixos
simukis_ has joined #nixos
trulsa has joined #nixos
amercan has quit [Ping timeout: 264 seconds]
Ikositetrachor has quit [Ping timeout: 272 seconds]
Ikositetrachor has joined #nixos
inferencerules has quit [Ping timeout: 246 seconds]
inferencerules has joined #nixos
orivej has joined #nixos
Ikositetrachor has quit [Ping timeout: 260 seconds]
<jmercouris>
I think a lot more would get done on this channel if this bot would stop spamming
<jmercouris>
we don't need to know every time a commit is pushed to mster
<{^_^}>
[nixpkgs] @NickHu opened pull request #100411 → libayatana-appindicator: generate typelibs for gobject-introspection → https://git.io/JTtpa
<simpson>
jmercouris: Very little gets done in-channel, period. The bulk of the work done is in the nixpkgs git repository. (This channel is only a pale reflection of what has been built.)
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/JTtpp
<evanjs>
I really just want syntax highlighting/language injection (ie elisp syntax highlighting in nix) lol
<infinisil>
Nice
<evanjs>
Wat, nix no support Unicode? TIL
<dminuoso>
evanjs: Gah, why would you want that..
<infinisil>
evanjs: It does though?
<evanjs>
dminuoso: which thing lol
<{^_^}>
[nixos-homepage] @garbas pushed 2 commits to redesign-governance: https://git.io/JTthI
<dminuoso>
fromElisp
<dminuoso>
:<
<infinisil>
Oh or do you mean unicode variable names?
<simpson>
jmercouris: I'm sorry that you haven't experienced much help, but your question is tough. My USB-C laptop is charging fine, and runs NixOS, so I can't reproduce your problem either.
<evanjs>
infinisil: no clue. Bottom of the repo says "no Unicode bc nix doesn't support it"
<jmercouris>
simpson: it is OK, thank you anyways :-)
<siraben>
clever: looks like the build still fails, http://ix.io/2AAg
<ToxicFrog>
Apropos the community, something that kind of worries me about the long-term health of nixpkgs/nixos is the sheer number of open issues and PRs -- it seems like lots of stuff gets reviewed and either sent back for updates and never gets updated (I've been guilty of this myself, but largely due to the next point...) or approved but never merged.
<ToxicFrog>
Does nixpkgs need more reviewers? More people with write bits? A bot that can do the merges so that once something has been approved by maintainers you don't have to wait weeks-months for someone to merge it in?
<jmercouris>
I have a feeling it is some problem with my thunderbolt drivers
<infinisil>
> substring 0 1 "🙃"
<infinisil>
,ping
<infinisil>
Um
<evanjs>
dminuoso: idk, but I'm open to whatever gives me <Lang> in nix highlighting instead of "that's a multi-line string" 😝
<infinisil>
,ping
<{^_^}>
[nixos-homepage] @garbas pushed 21 commits to redesign-governance: https://git.io/JTthn
<infinisil>
,ping
<immae>
the bot is upset :p
<{^_^}>
pong
<evanjs>
infinisil killed it lol
<infinisil>
TIL that you can break the bot by trying to get a substring from an emoji
<infinisil>
(had to restart it manually)
<{^_^}>
[nixos-homepage] @garbas pushed 139 commits to redesign-donate: https://git.io/JTthl
<evanjs>
Sounds like something you'd do 😝
<simpson>
ToxicFrog: The bigger a system is, the more outstanding work it will have to do. We have *the biggest* ports tree in all of Linux-land, by far; it should not be a surprise that we have so much work left to merge.
<siraben>
clever: it's now due to hls-plugin-api failing, I'll see if updating nix channels helps
<dminuoso>
(Or just annotate with comments I guess)
<ToxicFrog>
simpson: yeah, my concern is that it looks like it piles up faster than it can be dealt with, and the long approval->merge period discourages contribution.
<evanjs>
Yeah that too. Idk. Haven't looked into it too deeply
<lordcirth>
jmercouris, What NixOS / kernel version are you running?
<jmercouris>
lordcirth: I installed from channel 20.09
<jmercouris>
lordcirth: it is a new installation, I am still a very novice user
<jmercouris>
I tried plugging in the dock and regenerating the hardware-configuration.nix
<jmercouris>
but that didn't seem to do anything
<simpson>
ToxicFrog: The size of the pile is a function of the size of the main repository. Not sure how to explain this more clearly.
<simpson>
Merges could happen faster, but they already happen basically as fast as people and ofborg and hydra can go.
<immae>
simpson: are you saying that computer resources are already at their maximum with the current "contribution workflow"?
xdefrag has joined #nixos
nomeata has joined #nixos
<lordcirth>
immae, Last I heard, Hydra was bottlenecked by the central node
<immae>
ok
<simpson>
immae: I'm saying that we already run ofborg and hydra constantly, and people merge things all the time. The system is not dysfunctional; it's operating at a pretty competent level of throughput.
<immae>
I didn’t imply dysfunctional in my question, sorry :) I was just not aware that it was a bottleneck
meh` has quit [Ping timeout: 240 seconds]
<immae>
(and I have the same feeling as ToxicFrog about discouraging contributions :p )
meh` has joined #nixos
vandenoever has quit [Read error: Connection reset by peer]
<amercan>
Do you have the hash for 20.03.3111 by chance?
<hexa->
not sure where I can fetch that from
<buffet>
hey, does something like `mkIf condition {iftrue} {iffalse}` exist?
domogled has quit [Quit: domogled]
<inferencerules>
cloning nixos/nixpkgs and pointing inputs.nixpkgs.url to the local copy let me get around the github rate limit at least. wonder why it's an issue here.
<hexa->
buffet: if condition then true else false?
<buffet>
hexa-, aight, just checking if thats already define somewhere. thanks
medvid_ has quit [Ping timeout: 240 seconds]
alp has joined #nixos
eoli3n has quit [Read error: Connection reset by peer]
<amercan>
Well, is there a way I can get the latest minimal then? or should I wait until the website link catches on?
<polytux>
I've added a local CA to my security.pki.certificateFiles, but when I try to build again, nix complains it can't find the certificate I've added to the certificateFiles list. I've doublechecked for typos, tried different locations for the file, and googling doesn't give me much. any pointers?
mmohammadi981266 has quit [Quit: Ping timeout (120 seconds)]
<clever>
nicolas[m]: the datadog package fails because the tests want apikeys, callHackage will re-create a duplicate of the package that is non-broken, then dontCheck fixes it
<emmanuel_erc>
Ugghhh.... so I think I found a way to use removeReferencesTo in my build, however, the build-time dependency that I'm trying to get rid of is still in the closure
<noonien>
anyone running containers in nixos? either docker or nixos containers. any pitfalls i should watch out for? anything else i should read apart from the manual?
bahamas has quit [Ping timeout: 240 seconds]
vandenoever has quit [Read error: Connection reset by peer]
vandenoever has joined #nixos
vandenoever has joined #nixos
vandenoever has quit [Changing host]
fendor has joined #nixos
erasmas has joined #nixos
ManiacOfMadness has quit [Ping timeout: 272 seconds]
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<leonardp>
i've been using nixos containers on and off and there were no pitfalls encountered
<{^_^}>
[nixpkgs] @AndersonTorres pushed 3 commits to master: https://git.io/JTqCB
Chiliparrot has quit [Client Quit]
knupfer has quit [Quit: knupfer]
knupfer has joined #nixos
alp has quit [Ping timeout: 272 seconds]
<tpw_rules>
what is an appropriate way to have multiple nix environments open in different terminals? i want something like python's virtual environments
<dutchie>
lorri+direnv if you always want the same environment in a given directory, or you can pass arguments to nix-shell to specify an environment to get a shell in
knupfer has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @Patryk27 opened pull request #100433 → nixos/containers: allow containers with long names to use private networks → https://git.io/JTq6U
<lordcirth>
When configuring my system with a flake, I can't figure out how to get one file (./ipfs.nix) in the modules list to use nixpkgs-unstable. I tried "(./ipfs.nix { pkgs = nixpkgs-unstable; })" But it said:
<lordcirth>
attempt to call something which is not a function but a path
<lordcirth>
So I tried using import, but then it wants config, lib, etc, which I would have to pull in from out of scope - surely there's a cleaner way?
<emmanuel_erc>
Ok, here's a better question. Why isn't there an attribute for every derivation called something like "runtime-deps" which allows you to explicitly your runtime-dependencies?
simonpe^^ has quit [Remote host closed the connection]
jabster28_ has joined #nixos
civodul has joined #nixos
jabster28 has quit [Ping timeout: 260 seconds]
kreyren has quit [Remote host closed the connection]
kreyren has joined #nixos
<freezeboy>
infinisil ok found the problem: the pkgconfig generated contains a buggy reference to -R flag for g++, brotli's master branch already reverts this bug but there is no stable version yet that contains it
<infinisil>
Neat
<freezeboy>
should I 1) wait for a clean release 2) patch locally the file during nix build ?
Linux-Love has joined #nixos
<infinisil>
emmanuel_erc: Because there is no generic definition of a runtime dependency. It depends on how the program accesses things during runtime
shibboleth has quit [Quit: shibboleth]
<infinisil>
E.g. there's dlopen for dynamic libraries loaded at runtime, there's binaries that are accessed via PATH, there's paths accessed via other means
<infinisil>
Although, I guess all of this could be abstracted into a runtimeDependency attribute, which wraps all binaries with LD_LIBRARY_PATH, PATH, etc.
<Linux-Love>
iLinux OS https://ilinuxos.com - New Debian 10 (Stable) based distro with 500 Apps Pre-installed that runs from USB Stick!
<infinisil>
Linux-Love: Why are you advertising this here?
Linux-Love has left #nixos [requested by gchristensen (Linux-Love)]
<emmanuel_erc>
Yeah, I understand yourp oint infinisil. I'm just getting frustrated trying to get rid of a build-time dependency that is 1GB (it needs to go)
<infinisil>
emmanuel_erc: Oh so there's also the nix term of runtime dependencies, which are all build-time deps which are referenced in the final $out
cjpbirkbeck has joined #nixos
<infinisil>
For preventing certain derivations being runtime deps you can also use pkgs.removeReferencesTo
<emmanuel_erc>
So I'm using removeReferencesTo in the postInstall of a derivation
<bbigras>
freezeboy: in might be in a nice deletable file integration_fuse_test.go
<freezeboy>
then I suppose you will need to make apatch
<bbigras>
it*
<bbigras>
thanks
<freezeboy>
it is really amazing (and sad) that so many test suites in many project are not really isolated
proofofkeags_ has quit [Remote host closed the connection]
proofofkeags has quit [Remote host closed the connection]
<realrokka>
Is someone using Jitsi-Meet on NixOS? The module is working fine, except I just can't get it to activate authentication. I tried so much different custom prosdy/videobridge config stuff etc. I am stuck at this for hours ;___; please help ryantm you are my only hope :D
<bbigras>
yeah
<ryantm>
realrokka: I'm using it, but without authentication.
<realrokka>
ryantm: ;____; oh boi I really hope I don't have to reside to docker for this one ... anyway thanks for your packaging efforts mate! :)
<tpw_rules>
hi so i'm trying to reinstall an old derivation i made which has a fetchgit with a given url and rev and deepClone=true, but the hash has changed. any idea why that might be?
<tpw_rules>
the rev and url haven't
malook has quit [Quit: malook]
<ryantm>
realrokka: Feel free to open a GitHub issue with more details!
<{^_^}>
[nixpkgs] @jtojnar pushed to gnome-3.38 « fixup! gnome3.gnome-control-center: Fix crash after leaving user-accounts »: https://git.io/JTqjT
civodul has quit [Quit: ERC (IRC client for Emacs 27.1)]
<{^_^}>
[mobile-nixos] @J-Kirch opened pull request #226 → ignore local env and vim files → https://git.io/JTmeG
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
shibboleth has quit [Quit: shibboleth]
<emmanuel_erc>
euank and freezeboy: Here was the simple problem, I wasn't properly referencing that correct binary. I also had to removeReferences for a few more things. But my goal, which was to get the Haskell compiler, GHC, out of the resulting build has been achieved.
__monty__ has quit [Quit: leaving]
spacekookie has quit [Quit: No Ping reply in 60 seconds.]
cosimone has quit [Remote host closed the connection]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @jonringer pushed commit from @jfroche to master « zope_filerepresentation: fix test »: https://git.io/JTmk6
<srid>
Okay, lshw - and it is using the wrong driver
Ikositetrachor has quit [Quit: WeeChat 2.7.1]
shibboleth has joined #nixos
<avn>
srid: I use iwlwifi as AP, and it work in range from perfect to awful, mainly depending fo client side
<srid>
i experience periodic wifi disconnects on my x1 carbon, it is damn annoying
<srid>
i'm hoping updating the firmware might fix it
<srid>
https://wireless.wiki.kernel.org/en/users/drivers/iwlwifi lists iwlwifi-9000 to be latest, which is available per modinfo - though kernel is using an different version (QuZ-a0-jf-b0-55.u); so my question is - how do change the firmware?