<samueldr>
though, tbf, it's more about it being hard to support other shells... pretty sure if we can figure a shell-agnostic nix-shell it'd be desirable
butterthebuddha has joined #nixos
alexherbo2 has quit [Ping timeout: 260 seconds]
<samueldr>
hard, especially in the way nix currently is
<zeta_0>
iqubic: oh i see, i haven't set up anything like that yet
<m1cr0m4n>
Hey folks. I am trying to use django-auth-ldap with mailman, and I'm struggling to get it in the pythonpath for mailman-web. What would be the best way for me to do this? Currently, I am modifying the settings.py and appending the pythonpath with site.addsitedir, but I have to do so for all dependencies of the package too
<emily>
samueldr: yes, it's hard for nix-shell to support other shells. not so for nix run
<{^_^}>
[nixpkgs] @jonringer merged pull request #80832 → pythonPackages.seaborn: use v0.9.1 for python 2 → https://git.io/Jv0R4
<emily>
and yet the issue lies open and dormant :(
<{^_^}>
[nixpkgs] @jonringer pushed commit from @maxxk to master « pythonPackages.seaborn: use v0.9.1 for python 2 »: https://git.io/Jv0uB
<emily>
the irony of running into Nix bash-specificity problems while hacking on a build system partially borne out of the frustration of how much bash nix involves
<gchristensen>
but if you zip on past all the nixpkgs stuff and use fundamental derivations, you can do something else. I saw a neat alternative stdenv at a company which used python.
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
<iqubic>
If I want to run the dunst binary in my .xsessions file, I should end the line with &, right? Because otherwise it would block before getting to the "exec i3" line.
<iqubic>
the empty list there is just a list of additional things to ignore.
<iqubic>
nwspk: Yeah, that's what I just posted.
<nwspk>
iqubic: yeah, you were 30 seconds faster ;-)
smatting has quit [Ping timeout: 240 seconds]
<iqubic>
And I even did a quick google search to make sure I was thinking of the correct function.
<iqubic>
energizer`: You may have to import nixpkgs to get this to work properly.
tilcreator has joined #nixos
zeta_0 has joined #nixos
<zeta_0>
is bbenoist.Nix an vscodium extension for nix support?
orivej has joined #nixos
mbrgm has quit [Ping timeout: 240 seconds]
mbrgm_ has joined #nixos
mbrgm_ is now known as mbrgm
jgeerds has quit [Ping timeout: 258 seconds]
zeta_0 has quit [Quit: rcirc on GNU Emacs 26.3]
<{^_^}>
[nixpkgs] @BBBSnowball opened pull request #80849 → nixos/nextcloud: avoid loading imagick extension more than once → https://git.io/Jv0zk
Rusty1 has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @christianpoveda opened pull request #80850 → megasync: add -fpermissive to QMAKE_CXXFLAGS → https://git.io/Jv0zn
maddo has quit [Quit: See ya]
leungbk has joined #nixos
leungbk has quit [Remote host closed the connection]
m0rphism has quit [Quit: WeeChat 2.6]
leungbk has joined #nixos
catvector has joined #nixos
<leungbk>
I wrote a package recipe for an Emacs package not yet in Nixpkgs. How can I build an Emacs executable with this package using my branch of the Nixpkgs repo?
dansho has quit [Ping timeout: 272 seconds]
reallymemorable has quit [Quit: reallymemorable]
<leungbk>
The recipe is at http://paste.debian.net/1131626 but I'm not sure how to execute a nix-build operation that knows about this.
KeiraT has quit [Ping timeout: 240 seconds]
logzet has quit [Ping timeout: 240 seconds]
logzet has joined #nixos
judson_ has quit [Ping timeout: 268 seconds]
h0m1 has quit [Quit: WeeChat 2.7]
h0m1 has joined #nixos
orivej has quit [Ping timeout: 272 seconds]
KeiraT has joined #nixos
tA_ has quit [Ping timeout: 268 seconds]
Soo_Slow has joined #nixos
<energizer`>
I'm trying to use some R code. I can successfully get a nix-shell from `nix-shell r.nix` but it fails when running `nix-shell default.nix` https://bpaste.net/UKMA
tA_ has joined #nixos
judson has joined #nixos
<energizer`>
I guess I need to override a buildPhase somewhere?
<xacktm>
I have an intel hd graphics 505 that should support decoding h264/vp9 in hardware
orivej has quit [Ping timeout: 272 seconds]
<kini>
During a package build, what sets the timestamps of everything in $out, and is there any way I can disable it for the package I'm writing? The software is sensitive to its files' timestamps at runtime (unfortunately).
zupo has joined #nixos
mlatus_ has joined #nixos
mlatus_ has quit [Remote host closed the connection]
<iqubic>
That's such a weird thing.
<iqubic>
What software is it?
reallymemorable has joined #nixos
<kini>
ACL2, a theorem prover. When you "certify" a file by proving the theorems in it, it saves a certificate which is supposed to be a guarantee that the theorems in the file have been proved. If the timestamp on the file changes then it considers the certificate to be invalidated. This also applies to the files in its standard library which I am trying to build into the package (the current version in nixpkgs doesn't have the standard library).
<kini>
It also doesn't like the files to be moved, so I'm doing the build in-place in $out/share/ :)
<iqubic>
Wow. What a strange as all heck piece of software.
mlatus_ has joined #nixos
mlatus_ has quit [Remote host closed the connection]
<kini>
It is a bit idiosyncratic, yes. It also has a really long history, as it dates back to the early 1990s (though it continues to be actively developed today).
mlatus_ has joined #nixos
mlatus_ has quit [Remote host closed the connection]
mlatus_ has joined #nixos
mlatus_ has quit [Remote host closed the connection]
mlatus_ has joined #nixos
mlatus_ has quit [Remote host closed the connection]
<kini>
It used to have an in-house documentation markup language. I knew it was older than most modern lightweight markup languages like Markdown, but I figured they could have at least used e.g. XML or HTML. But ACL2 actually predates HTML, lol
<kini>
Yes, I just read that a minute ago because it was one of the handful of occurrences of the word "timestamp" in the Nixpkgs manual, where I was searching trying to find an answer for my question :)
mlatus_ has joined #nixos
<mlatus_>
[1051966856] <mlatus> test
iyzsong-x has joined #nixos
iyzsong has quit [Ping timeout: 272 seconds]
mlatus_ has quit [Remote host closed the connection]
<emily>
kini: I think your best bet is to wrap acl2 with something that copies the file to a temporary location and touch(1)s it to the correct time
cole-h has quit [Ping timeout: 268 seconds]
b has quit [Ping timeout: 272 seconds]
<kini>
emily: not sure what you mean by copying the file to a temporary location, but since ACL2 cares about the absolute paths of files as well, that sounds dangerous...
<kini>
there does appear to be a setting I can use to make it use file checksums instead of relying on timestamps, but the documentation says that will make various operations slower so I guess I'd like to avoid using that... but maybe it's good enough for now
<emily>
the overhead of checksums is not going to be relevant on modern hardware
<hyper_ch>
hmmm, I have a problem with latest kernel and sddm - I always have to kill sddm first before it actually starts the autologin.
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
alexherbo22 has joined #nixos
stree has quit [Quit: Caught exception]
stree has joined #nixos
<kini>
emily: Unfortunately they're not actually file checksums, they're more like hierarchical AST object hashes, so it involves some semantic parsing of the files... and I'm not really sure but it might have to be done multiple times per file because of multiple dependency chains and stuff. I found in the release notes that the checksum setting slows down the regression suite by ~7.5% which is nontrivial...
<kini>
but I guess there's no way to disable the timestamp settings that nix does. Apparently NAR archives don't even store timestamps so even if I build my package while retaining the timestamps they'd still not get broken if I shipped the package as a NAR...
<kini>
guess I'll just use the checksum method then :) thanks for your help. (and energizer` they're just source code files so anywhere from 1-100KB generally I guess)
ixxie has joined #nixos
<{^_^}>
[nixpkgs] @maxxk opened pull request #80874 → [20.03] pythonPackages.seaborn: use v0.9.1 for python 2 → https://git.io/Jv0KT
<clever>
syd: then its just a normal thing in nix-env, you can `nix-env -iA nixpkgs.nix` to install the new version
<syd>
replacing old 'nix-2.1.3'installing 'nix-2.3.3'
<syd>
wonderful
<kini>
omg lol, looks like there was a much simpler solution to my problem... when installing in place in $out, I used `cp -R` to move the source files there, so their timestamps were updated. I changed it to `cp -pR` and ACL2 stops complaining about the timestamps at runtime, haha
<kini>
apparently it doesn't mind when the certificate file itself has its timestamp changed -- it only cares that the source file shouldn't have its timestamp changed between when the certificate was created and when the certificate was read. problem solved... d'oh
chimay has joined #nixos
syd has quit [Remote host closed the connection]
<clever>
kini: nix always sets all timestamps to 1 second after epoch, when the build is finished
philr has quit [Ping timeout: 255 seconds]
<kini>
clever: right, I was asking about how to potentially disable that behavior (I realized this is impossible due to e.g. the NAR format not containing timestamp info) because the software I'm building (ACL2) is finicky about timestamps of its files
<clever>
kini: the most you can do is to either patch the software, or use a shell wrapper, to copy the files to $HOME and fix the timestamps
<kini>
I mean, I was asking that earlier (a couple hours ago), but now the problem seems to be solved
xkapastel has quit [Quit: Connection closed for inactivity]
<mlatus>
When switch from configuration A to B, why set up /etc before restart services? Stop scripts declared in A will not be executed if it is not also declared in B
v0|d has quit [Remote host closed the connection]
<mlatus_>
/etc
pnobre has quit [Ping timeout: 268 seconds]
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<Yaniel>
you can at least have the unstable channel for user-installed stuff and stable for system-level stuff
<lirzhv>
how is the distinction made?
<__monty__>
lirzhv: You can pick whatever packages you want from whatever channel using overlays.
<__monty__>
,overlay lirzhv
<{^_^}>
lirzhv: Overlays look like `self: super: { foo = ...; }`. Use the self argument to get dependencies, super for overriding things and library functions (including callPackage). More info: https://nixos.org/nixpkgs/manual/#sec-overlays-install
<lirzhv>
ok nice, thanks
<Yaniel>
IIRC you can run nix-channel without sudo to set your user channel
<woffs>
lirzhv: and you can use e.g. "-f channel:nixos-unstable" with nix-env
<woffs>
though I prefer declarative config and would rather pick some newer software's *.nix declarations and compile them with the current main channel I am on
<{^_^}>
systemd/systemd#12131 (by kouros17, 47 weeks ago, closed): RFE: migrate StateDirectory= and friends back from DynamicUser=1 subdir to regular one if needed
bhipple has joined #nixos
<tilpner>
__monty__: Yes, but that's backwards incompatible. As long as we have values looking like "19.09", people will touch it
<__monty__>
tilpner: Not necessarily, you can have stateVersion = newBetterName.
<tilpner>
Suggestion: Hex/base64/etc.-encode the value in nixos-generate-config, then decode in the generated config
<tilpner>
It's a blob now, people won't be tempted to change it
<tilpner>
And we can also make stateVersion an internal option and set it from a better-named option
vykook has quit [Ping timeout: 240 seconds]
<tilpner>
That way, everything stays compatible, but new users will stop doing this
<fendor_>
sorry :)
<GiGa>
I was never sure what stateVersion was for, was just aware that I was to leave it alone. I should look up what it actually does...
<tilpner>
This happens a lot, fendor_, there seems to be something tempting about the comment telling you to leave it alone
shyim has quit [Quit: shyim]
<manveru>
wouldn't it be better to just write it to /var/lib/nixos initially?
<fendor_>
it definitely makes it memorable. I dont know, I just thought that additionally to changing channels, you need to bump this on upgrade :/
<__monty__>
GiGa: It's about backwards-incompatible changes to the db schema aiui.
<tilpner>
manveru: No, that would make evaluation dependent on the machine you're evaluating it on
shyim has joined #nixos
<__monty__>
tilpner: Otoh, it doesn't really make sense to evaluate a config with an old StateVersion on a fresh machine...
<tilpner>
__monty__: Yes, it does. I can deploy to old servers from my new laptop just fine
noudle has quit []
<__monty__>
It's a weird thing to *set*.
<manveru>
it's kinda similar to crosscompiling i guess...
<tilpner>
And I can only do that, because stateVersion of the evaluated system does not depend on some global state on my laptop
<__monty__>
tilpner: NixOps can't deploy configurations with a newer stateVersion?
claudiii has joined #nixos
<m1cr0m4n>
Has anyone got a working deployment of mailman out there?
<tilpner>
__monty__: That's not the problem. The system needs to evaluate to different things, depending on when the to-be-deployed machine was first installed
vykook has joined #nixos
<__monty__>
tilpner: Yeah but if it's already been deployed you can technically figure out the stateVersion by looking at the db.
<tilpner>
__monty__: So I can have a server installed at 16.09, and one at 19.03, and they might need different workarounds to keep working. The exact same config might work on one, and not the other
<tilpner>
__monty__: No, evaluation happens on my laptop, it doesn't touch the server, and can't access its filesystem
<__monty__>
Ah, right. This is why we can't have nice things!
<tilpner>
__monty__: And while nixops has a db, my deployment solution doesn't. It also won't work if you use a Hydra to build your systems
aveltras has joined #nixos
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bennofs has quit [Quit: No Ping reply in 180 seconds.]
bennofs has joined #nixos
vykook has joined #nixos
lsix has quit [Ping timeout: 240 seconds]
<pnobre>
anyone using lightdm with enso greeter ? was trying to give it a try but doesn't seem to work...
tsrt^ has quit []
shah^ has joined #nixos
reallymemorable has joined #nixos
cole-h has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<ddima>
pnobre: I think I tried it a while ago, it was crashing/not coming up so I just went with gtk.
bvdw has quit [Quit: bvdw]
<pnobre>
yeah, same here...
bvdw has joined #nixos
infinisil has quit [Quit: Configuring ZNC, sorry for the joins/quits!]
GiGa has quit [Quit: Leaving]
infinisil has joined #nixos
Flambe has joined #nixos
<ddima>
I dont see an open issue for that, might be worth opening one if you'd really like to use that one. I was rather indifferent wrt greeters, so didn't bother back then.
waleee-cl has quit [Quit: Connection closed for inactivity]
<{^_^}>
[nixpkgs] @talyz opened pull request #80904 → nixos/haproxy: Revive the haproxy user and group → https://git.io/Jv0Ql
shafox has quit [Remote host closed the connection]
lsix has joined #nixos
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
NoctisLabs has quit [Ping timeout: 255 seconds]
abathur has joined #nixos
ryantrinkle has joined #nixos
aveltras has quit [Quit: Connection closed for inactivity]
<pnobre>
ddima: didn't really want to use it, was just seeing what options were available... just started playing with nixos this weekend and just trying different stuff and learning a bit....
<pnobre>
seems very cool up until now, although docs aren't great, but I guess everyone knows that, from what I've been reading around... and who cares about rtfm ? :-D
lsix has quit [Ping timeout: 240 seconds]
ryantrinkle has quit [Ping timeout: 258 seconds]
abathur has quit [Read error: Connection reset by peer]
abathur has joined #nixos
<{^_^}>
[nixpkgs] @bennofs opened pull request #80905 → ###### Motivation for this change → https://git.io/Jv07c
drakonis has joined #nixos
fendor_ has quit [Ping timeout: 255 seconds]
bhipple has joined #nixos
<ddima>
pnobre: there are people working on the docs, but resources are limited. also not everyone likes docbook/feel like its a hurdle, so there is some discussion there. PRs to the docs are quite welcome and easy to get in though. but ultimately, being able into the pkgs, modules and libs directly
<ddima>
+ is a valuable kill
gustavderdrache has joined #nixos
<pnobre>
yeah, that's basically where I've been learning from :-) besides a few usefull links from the docs (mostly options and pkg search)...
<ddima>
the wiki is often quite useful too. and getting inspiration from others configs on github.
<{^_^}>
rfcs#64 (by Infinisil, 6 weeks ago, open): [RFC 0064] New Documentation Format for nixpkgs and NixOS
<emily>
ajs124: no. seeing how long the scrollbar is I definitely don't feel like getting involved tbh
<emily>
I see Mr. "I use my private NixOS fork but I'm still going to complain about all your decisions" is arguing for, uh, org-mode? ...yeah, I'm definitely too jaded about bikesheds like these to comment, sorry >_>
<ajs124>
That's mostly how I feel about it, as well. I thought I'd warn you, before you go and implement asciidoc support (which IMHO sounds like a good idea, but I digress) only to be pointed to this whole thing afterwards.
<pnobre>
think I found what the problem is... /etc/lightdm/lightdm.conf has greeter-session set to lightdm-enso-os-greeter
zupo_ has joined #nixos
<pnobre>
but by xgreeters only has a pantheon-greeter.desktop
justanotheruser has quit [Ping timeout: 240 seconds]
drakonis has joined #nixos
<kini>
Does nix set a ulimit or something when building an expression? My build seems to have been killed by the OOM killer but I thought I had enough memory to do the build (from experience running the same build outside nix).
<{^_^}>
[nixpkgs] @artemist opened pull request #80909 → libfido2: add macOS support → https://git.io/Jv0Fa
<latiheis>
howdy, with fetchFromGithub, is there a way to forgo saying a revision and sha256 and get it to pull whichever is the latest in the master branch?
kahiru has joined #nixos
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<cole-h>
You probably want `builtins.fetchTarball` for that -- sha256 isn't required, and you can point it to the `master.tar.gz` URL
<gchristensen>
builtins.fetchgit, too, though it is usually not a good idea since you lose some of the benefits of Nix.
mexisme has joined #nixos
vykook has quit [Ping timeout: 260 seconds]
chimay has quit [Ping timeout: 258 seconds]
ixxie has quit [Ping timeout: 265 seconds]
<emily>
can you do something like __FILE__ in Nix?
dingenskirchen has quit [Quit: dingenskirchen]
LouisDK has joined #nixos
troydm has quit [Quit: What is Hope? That all of your wishes and all of your dreams come true? To turn back time because things were not supposed to happen like that (C) Rau Le Creuset]
<samueldr>
not in a clean way, I think (builtins.unsafeGetAttrPos "a" {a = 1;}).file
tollb has joined #nixos
<samueldr>
> (builtins.unsafeGetAttrPos "a" {a = 1;})
abathur has quit [Read error: Connection reset by peer]
<samueldr>
from nix 1.7
<samueldr>
>> There is a new symbol __curPos that expands to an attribute set containing its file name and line and column numbers, e.g. { file = "foo.nix"; line = 10; column = 5; }. There also is a new builtin function, unsafeGetAttrPos, that returns the position of an attribute. This is used by Nixpkgs to provide location information in error messages, e.g.
vykook has joined #nixos
dingenskirchen has joined #nixos
<{^_^}>
[nixpkgs] @jonringer merged pull request #80874 → [20.03] pythonPackages.seaborn: use v0.9.1 for python 2 → https://git.io/Jv0KT
<{^_^}>
[nixpkgs] @jonringer pushed 2 commits to release-20.03: https://git.io/Jv0bJ
<{^_^}>
[nixpkgs] @jonringer pushed to master « pythonPackages.chalice: disable pip version bounds »: https://git.io/Jv0bz
bobsterman is now known as liamdiprose
troydm has joined #nixos
_deepfire has quit [Remote host closed the connection]
<pnobre>
a little doubt here... trying to copy some wallpapers to /usr/share/pixmaps...
<tollb>
I have a commit that addresses an abort/coredump in gdb 8.3.1 due to a NixOS-specific patch. My patch applies cleanly on current staging. However, I noticed that there is a pending pull request to upgrade gdb to 9.1 that also targets staging and would conflict with my change. Ideally, I'd like to get the patch into both 20.03 and master, assuming it is acceptable. What is the best way to proceed?
<pnobre>
1. I've created a pkgs with them (it's this the correct way ? is there an easier way to do it ? )
<{^_^}>
[nixpkgs] @jonringer opened pull request #80911 → pythonPackages.ansible-lint: disable check on python2 → https://git.io/Jv0br
<pnobre>
2. When I set one of those wallpapers to lightdm background, using /usr/share/pixmaps/xxx.jpg doesn't work (/usr/share doesn't exist) how, should I reference the profile to get the correct path to the files ?
<cole-h>
You reference it with `{pkgname}/usr/share`, assuming you did everything correctly
<cole-h>
Err
<cole-h>
`${pkgname}/usr/share`
<pnobre>
well, the files are in there, I've just double checked them...
mexisme has joined #nixos
reallymemorable has quit [Quit: reallymemorable]
<pnobre>
thanks.. it's actually `${pkgname}/share...` (without the /usr) but that sent me in the right direction :-)
clkamp_ has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @samueldr pushed to release-19.09 « runInLinuxVM: Ensure tools requiring /etc/passwd work »: https://git.io/Jv0Nf
Naptra has quit [Quit: Leaving]
leungbk has joined #nixos
<{^_^}>
[nixpkgs] @samueldr pushed to release-20.03 « runInLinuxVM: Ensure tools requiring /etc/passwd work »: https://git.io/Jv0N3
v0|d has quit [Remote host closed the connection]
ddellacosta has joined #nixos
<latiheis>
so im working on a vim derivation, and following this page: https://nixos.wiki/wiki/Vim . Im using the builtin vim packaging and trying to get the optional plugins working, ive added the autocmd line it says on the page but it doesnt seem to be loading them when I start up a file with the right extension? it loads fine if i put them in the non optional category
tertl3 has joined #nixos
zupo has quit [Ping timeout: 240 seconds]
zupo has joined #nixos
<latiheis>
if i manually do a setfiletype it works
alexherbo2 has joined #nixos
alexherbo22 has quit [Ping timeout: 265 seconds]
<latiheis>
fixed, needed to add an autocmd to set the racket filetype, it was defaulting to scheme
<leungbk>
I'm trying to add non-melpa Emacs packages to Nixpkgs, but even after specifying the package recipes, specifying their locations in emacs-modes/manual-packages.nix, and running "nixos-rebuild switch -I nixpkgs=/home/brian/mygit/nixpkgs", trying to build an Emacs instance using one of my new packages gives the error "error: undefined variable 'isearch-prop' at /home/brian/mygit/nixpkgs/test-emacs.nix:14:5".
<leungbk>
Is there some other place I need to specify the package recipes or their locations to make the packages known to my Nix installation?
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « cantata: 2.3.3 -> 2.4.1 »: https://git.io/Jv0Au
abathur has joined #nixos
abathur has quit [Read error: Connection reset by peer]
myskran has joined #nixos
<leungbk>
jackdk: I tried that just now, but the error still appears.
<jackdk>
leungbk: at this point I suggest preparing a more detailed report. I suggest preparing a pastebin with things you have tried, detailed error messages, redacted config snippets, how you wired your new packages in, etc. I'm about to head to work and won't be able to check it, but there's not enough information for me to know.
myskran has quit [Read error: Connection reset by peer]
<latiheis>
lots of questions today sorry, i have this file im trying to work on: https://0x0.st/iNfv.nix but one of the plugins requires git to build, how can i mark git as part of the environment for this? i tried putting { stdenv, git }: but it wouldnt work
<evanjs>
and the relevant messages are: terminate called after throwing an instance of 'nix::EvalError'
<evanjs>
and what(): cannot import '/nix/store/lld6kvp1adzwfgrpfx9a5zdlp6wvjpql-source/common/pc/laptop', since path '/nix/store/ymx8ldh528rpqx57ijkw1mxhlqqv8a0v-source.drv' is not valid, at /etc/nixos/config/hardware/profiles/laptop.nix:6:6
<evanjs>
I garbage-collect a lot, and I also have autoOptimisation enabled
karetsu has joined #nixos
karetsu has quit [Client Quit]
<evanjs>
It's like the store path is removed when stuff still references it, or something?
<aaa_>
Hi, after updating my system, I can't see my wifi card, can somebody help: http://dpaste.com/15W9SSZ (certain info), let me know what other info you need?
<evanjs>
Sure enough, rebuilding just now fixed it (don't even need a full rebuild AFAIK) and e.g. running `nixos-option` after `nix-store --gc --max-freed 1G` reproduced the issue.
<{^_^}>
[nixpkgs] @velovix opened pull request #80918 → go_1_12, go_1_13: remove velovix from maintainers → https://git.io/Jv0h1
<{^_^}>
[nixpkgs] @roberth opened pull request #80919 → lib/modules.nix: Add file context to unmerged values in mergeDefinitions → https://git.io/Jv0hS
leungbk has joined #nixos
<immae>
evanjs: I’m not sure I understand your issue, are you asking why the dependencies disappeared after a garbage collection?
<immae>
if so, you need keep-outputs = true and/or keep-derivations = true in your /etc/nix/nix.conf (but that won’t solve everything)
troydm has quit [Ping timeout: 258 seconds]
turona has quit [Ping timeout: 272 seconds]
<evanjs>
immae: right that's what I was wondering. It's like it thinks it doesn't need it anymore, in spite of being part of my current system, as far as I understand
<immae>
Ah I just understood: it is actually a *runtime* dependency, not just *building* dependency?
drakonis has quit [Quit: WeeChat 2.7]
<evanjs>
well that's what I'm wondering. Trying to query totally separate options (e.g. programs.autojump.enable) still produce the error, so it seems more like a runtime dependency to me, yes
sshow has joined #nixos
tertl3 has quit [Quit: Connection closed for inactivity]
cr4y1 has quit [Ping timeout: 240 seconds]
<evanjs>
FWIW, why-depends seems to think /run/current-system doesn't depend on that store path, either
<immae>
ok
<immae>
I have this kind of double import and indeed they get garbage collected in spite of the dependency (but in my case it’s a build dependency so I don’t mind too much)
xkapastel has joined #nixos
<evanjs>
interesting. the mozilla-overlay rust channels don't seem to be garbage collected, though perhaps that's due to their use elsewhere in my configuration
jgeerds has quit [Ping timeout: 268 seconds]
<aanderse>
infinisil: ping
bvdw has quit [Read error: Connection reset by peer]
<immae>
evanjs: There are ways to write the derivations to avoid that, but I’m a bit lost for your specific case (an import)
bvdw has joined #nixos
<evanjs>
hrm alright. I do wonder if I actually need the double import, anyway
<evanjs>
at least for nixos-hardware
turona has joined #nixos
<infinisil>
aanderse: pong
<immae>
could you do something like : let your_src = fetchFromGitHub { ... } in { imports = [ "${your_src}/common/pc/laptop" ] } ?
<immae>
yes, I meant that one in the "..." in my example
<immae>
just to start somewhere
<evanjs>
ah yeah I believe so
<evanjs>
it's basically what I have for mine :D
<immae>
it’s should be equivalent but it should work with garbage collection now
<immae>
Then if you confirm we could improve from there
mexisme has quit [Ping timeout: 240 seconds]
<evanjs>
I'll check, but I also realized that I add nixpkgs-mozilla (also double imports) to my overlays. Wonder if doing the same for this would do anything. Anyhow, I'll try moving the source into the let block
<immae>
(just temporary, if it works then we can find a way to put it back in a separate file like currently)
<immae>
evanjs: you did exactly what I had in mind, and I didn’t expect an infinite recursion there :/
<noonien>
hello!
<noonien>
at the end of a successful build, i get `error: path '/nix/store/9cwxpw6391xfy334b4aivril1qrxgs1l-spanky-deps' is not valid`, how can i get more info?
<noonien>
i'm building using `buildCommand = ''do something here to create $out''`
zeta_0 has joined #nixos
<cole-h>
Is it possible to reference a local (relative) directory and not add it into the store?
<immae>
cole-h: it depends where your reference it, `./foo/bar` per se won’t go into the store by itself, it’s the thing that uses this which will make it go into it
ixxie has quit [Ping timeout: 252 seconds]
smatting has joined #nixos
<{^_^}>
[nixpkgs] @danielfullmer opened pull request #80922 → programs/chromium: allow extensions outside chrome web store → https://git.io/Jv0ji