vika_nezrimaya has quit [Quit: ERC (IRC client for Emacs 26.3)]
<kini>
am I doing something wrong, or does `ffmpeg` in nixpkgs build with x264 while `ffmpeg-full` doesn't?
orivej has quit [Ping timeout: 256 seconds]
orivej_ has joined #nixos
<kini>
er, not that it's broken, just that it doesn't include it by default when you add `ffmpeg-full` to `environment.systemPackages`, whereas you do get it if you put normal `ffmpeg` into `environment.systemPackages`.
Fare has joined #nixos
ndfh^ has quit []
therealsarcastic has joined #nixos
simukis__ has joined #nixos
therealsarcastic has quit [Ping timeout: 256 seconds]
hlisp has joined #nixos
simukis_ has quit [Ping timeout: 272 seconds]
tmaekawa has joined #nixos
reallymemorable has quit [Quit: reallymemorable]
codygman has quit [Read error: Connection reset by peer]
codygman has joined #nixos
slack1256 has joined #nixos
h0m1 has quit [Ping timeout: 260 seconds]
slack1256 has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @hyperfekt opened pull request #86699 → add maintainer scripts for haskell package generation → https://git.io/Jfsi0
<cole-h>
hyperfekt++
<{^_^}>
hyperfekt's karma got increased to 5.00000000000000004
<danderson>
so, I'm thinking of sending a patch to the chromium nixos module, to allow optionally installing the missing proprietary bits from Chrome. Specifically, the Widevine video DRM plugin and the audio codecs that only ship in Chrome.
h0m1 has joined #nixos
<cole-h>
Doesn't that already exist?
<danderson>
Is there precedent for how to handle options that bundle non-libre things into base NixOS?
<danderson>
okay, so I guess I just need to figure out how to customize the package in my nixos install. Thanks!
<cole-h>
:)
<cole-h>
danderson: btw did you get a chance to ask Graham your question?
tmaekawa has quit [Quit: tmaekawa]
ixxie has quit [Ping timeout: 246 seconds]
<quinn>
danderson: pro-tip: clone the repository, and cast ack/ripgrep on it. there are a lot of "options" that only show up as a .override or some other little thing you wouldn't think of
ixxie has joined #nixos
<danderson>
quinn: good tip, thanks! I have a nixpkgs clone handy already, but didn't occur to me that packages would have knobs not exposed in modules.
<danderson>
that's a tiny bit annoying for customization, but oh well I guess
<cole-h>
Usually, the module knobs are for the module itself. For customizing the package, you are expected to create an overlay or use some similar means.
<danderson>
Sure, that bit makes sense. The bit that sucks is there's no clean way to browse "what tweakables are available" compared to module options
<cole-h>
Ah, yeah. Depending on my level of energy, I either `rg 'widevine'` in my checkout or use GitHub's terrible web search on upstream
<cole-h>
Would be nice if there was a collection of anything that is `enableSomething ? false` and similar
<danderson>
yup. The nixos options reference has its own issues, but it's about a thousand times better than either of those :)
<danderson>
but oh well, now that I know that's a thing...
<cole-h>
Maybe using one of the various nix parsers, you could just look for `?` in the top-level attrset of each package and collate them somehow
hlisp has quit [Ping timeout: 260 seconds]
pjt_014 has joined #nixos
tsrt^ has joined #nixos
<{^_^}>
[nixpkgs] @cole-h opened pull request #86701 → improve robustness of various Rust packages (part 2) → https://git.io/Jfsi1
<quinn>
danderson: true. it is unfortunate. there is also the fact that some people have their own custom stuff, and there are external modules (musnix, home-manager). i check the configuration collection too https://search.tx0.co/?q=home-manager&i=nope&files=&repos=
dingenskirchen has quit [Remote host closed the connection]
<colemickens>
I honestly just can't deal with it. I don't have time to baby sit it.
<cole-h>
Just delete i3status-rust >:)
<colemickens>
:S
hmpffff has quit [Ping timeout: 260 seconds]
hlisp has joined #nixos
<colemickens>
I've thought about it. Or trying to let packages be skipped automatically, etc.
<colemickens>
idk, I kind of hope the i3status-rs folks are just having a productive week and it calms back down :P
akamaus has quit [Ping timeout: 260 seconds]
<colemickens>
that's maybe not a nice wish, I hope they all the productivity.
turona has quit [Ping timeout: 272 seconds]
<colemickens>
I guess I should wish for a cargoSha256 CLI fetcher util.
<energizer>
why are there things in nixpkgs that cause the build to fail?
turona has joined #nixos
VPS has quit [Quit: Departure requested.]
<energizer>
answer: because there's nothing preventing them from being added
<cole-h>
What do you mean?
<jlv>
I'm pretty sure `fonts.fontconfig.defaultFonts.emoji` is broken. It sets the second to top font, but Noto Color Emoji is glued as the top emoji font. Even a `binding="strong"` alias doesn't work. With `fonts.fontconfig.penultimate.enable = true;`, Noto Color Emoji isn't stuck at the top, but `fonts.fontconfig.defaultFonts.emoji` has no effect.
<energizer>
cole-h: nixpkgs allows deps to be added that break dependents?
<energizer>
or, more precisely, nixpkgs allows deps to be changed in a way that breaks dependents?
ninegrid has quit [Ping timeout: 246 seconds]
<cole-h>
energizer: Sorry, I'm still not parsing. Are you asking if nixpkgs allows breaking changes in dependencies?
VPS has joined #nixos
<energizer>
cole-h: suppose a i want to upgrade a dependency in nixpkgs. how do i find out whether that breaks some other package that depends on it
<colemickens>
nixpkgs-review ?
<cole-h>
`nixpkgs-review`, usually
<energizer>
does that get run automatically before merging a change?
<cole-h>
No
<cole-h>
It's usually a part of the review process.
<cole-h>
You can run it yourself (and tick one of the checkboxes in the PR template, while you're at it), or one of the reviewers will usually run it.
<energizer>
what happens if some other package is broken by my change?
<cole-h>
You can try to fix it
<energizer>
what if i dont want to fix it?
<cole-h>
Which may or may not entail going upstream and asking them to update to the latest version of library X
<cole-h>
If you don't want to fix it, you mark the package as broken.
<energizer>
does this happen often?
<cole-h>
I dunno. Why?
<colemickens>
break into the other devs computer, upgrade their dependency too, then they'll have to update their app/library.
<colemickens>
works every time
hlisp has quit [Ping timeout: 256 seconds]
reallymemorable has joined #nixos
<cole-h>
I mean really, if you don't want to even attempt to fix newly-broken packages, I would say don't submit the version bump and leave it to someone more invested...
<energizer>
i'm sorta surprised at that description (i wouldve expected this all to be automated) so i'm wondering if i misunderstand something about nixpkgs maintainence policies or if just nobody's gotten around to automating this
ddellacosta has quit [Ping timeout: 272 seconds]
<energizer>
like normally in my (non-nix) projects if somebody comes by and bumps a dep and it breaks something, the CI runs automatically before merge, fails and blocks the merge. but that's not happening here
<cole-h>
Well, usually a reviewer would note that it breaks things
<cole-h>
Which typically blocks merging
<{^_^}>
[nixpkgs] @rardiol opened pull request #86712 → nixos/hostapd: country selection, CRDA, logging → https://git.io/JfsPx
corpix has quit [Ping timeout: 240 seconds]
<colemickens>
energizer, image three PRs are proposed in a row, updating libva, mesa, and gcc.
<colemickens>
each of those require rebuilding nearly the entire world. now what?
LnL has quit [Ping timeout: 260 seconds]
<colemickens>
s/image/imagine
<cole-h>
^
<colemickens>
I thinkof the channel system as more of the QA gating.
LnL has joined #nixos
LnL has joined #nixos
LnL has quit [Changing host]
<energizer>
colemickens: what's the question?
<cole-h>
energizer: If nixpkgs-review was automated, imagine what would happen if there were 3 PRs in a row that would rebuild the world
hlisp has joined #nixos
<colemickens>
energizer: I guess I should let someone with authority speak (or someone link to prometheus) but computing resources are limited or cost constrained.
<energizer>
sounds like the world would be rebuilt 3 times
<cole-h>
Exactly
cgw has quit [Ping timeout: 256 seconds]
<cole-h>
If you want to sponsor build machines for that, by all means.
<Ashy>
in a package definition i'm trying to copy in some files from other packages to the $out directory, is there a good example somewhere of how to do that?
Darkmatter66_ has joined #nixos
akamaus has quit [Ping timeout: 246 seconds]
hlisp has quit [Ping timeout: 265 seconds]
<cole-h>
Why do you need files from other packages?
Darkmatter66 has quit [Ping timeout: 256 seconds]
<Ashy>
it's to include packages in an interpreter that come from different github repos
<energizer>
Ashy: maybe you want something like pythonWithPackages, but for j?
ixxie has quit [Ping timeout: 260 seconds]
ixxie has joined #nixos
<Ashy>
possibly yeah
waleee-cl has quit [Quit: Connection closed for inactivity]
<Ashy>
all of the j addon repos checked out only add up to less than 300MB so i think it's fairly common for j users to install them all on win/mac/other distros
<cole-h>
Whether or not it's common, it's always better to have choice :)
<cole-h>
It feels a bit sad to just give up, rather than give it that extra little push. But I can sympathise with a long waiting period making you reluctant to keep working on it.
<multun>
I actually was very frustrated with my first PRs too
<multun>
but it turns out it they all got taken care of in the end
<multun>
most without pushing at all
<cole-h>
Maybe I just got very lucky, but most of my PRs have seen relatively quick turnaround.
<multun>
same
<cole-h>
Only a handful that I needed to post on the "PRs already reviewed" thread
inkbottle has quit [Quit: Konversation terminated!]
<bqv>
the problem i have for the most part is that an "extra little push" can actually be quite a lot for me, thanks to various issues, and i'm coming to terms with the fact that i'm not the greatest at energy accounting so if i pour everything i have into them and get nothing out, then i'm expended for days for no reason, which is frustrating for everyone involved...
<multun>
just don't do the extra push!
<cole-h>
Here, I'll give them a little push for you, then, bqv :)
<bqv>
thanks :p
<cole-h>
Feel free to ask for help. If it's not me, somebody else will (hopefully) extend good will.
<bqv>
not sure those would necessarily be the same though
<multun>
zdump /etc/zoneinfo/America/Los_Angeles
<danderson>
/etc/zoneinfo/America/Los_Angeles Sun May 3 22:37:03 2020 PDT
<danderson>
so... hm.
<multun>
well the zone file looks alright
<multun>
I think that's just the utility pretty printing the tz offset that failed
<cole-h>
bqv: Just perused the diff again. I think the only thing I would change is `platforms.all` -> `platforms.unix`. `all` contains `x86_64-windows`, believe it or not, in case we ever start supporting it...
<multun>
danderson: where did you get that +0000?
<danderson>
multun: that was from `timedatectl`
<danderson>
but gnome and `date` also print the wrong time
reallymemorable has joined #nixos
<danderson>
» date ~/tail/corp
<danderson>
Mon 04 May 2020 05:39:27 AM America
<bqv>
cole-h: ah ok, i'll update that soon
<danderson>
(it's currently 22:39)
<multun>
danderson: wow that's strange
<multun>
danderson: can you run `strace date` and paste it somewhere and give the link?
<cole-h>
bqv: Does guix explicitly support darwin, though?
<danderson>
sure
<cole-h>
bqv: Might make even more sense to do `platforms.linux` if they don't make any guarantees about macOS
hlisp has joined #nixos
palo1 has joined #nixos
<bqv>
i can't recall, and my matrix server's down for a bit so i can't ask. better safer than sorry though, yeah
<energizer>
cole-h: i'm curious what makes you want to build/use(?) guix?
<cole-h>
I'll take a look, then
morgrimm has joined #nixos
<cole-h>
energizer: To make sure bqv's effort doesn't go to waste :)
<energizer>
ah :)
<cole-h>
bqv: I'd go for `platforms.linux` -- an (admittedly year-old) Reddit post says Mac is not supported
<bqv>
makes sense, ok
<cole-h>
But I don't see anything mentioning darwin/macOS on the download page, so I think it's a safe bet it's unsupported
<multun>
openat(AT_FDCWD, "/nix/store/6m2k8kx8h216jlx9dg3lp4m90bz05yck-glibc-2.30/share/zoneinfo/America/Los_Angeles", O_RDONLY|O_CLOEXEC) = -1 ENOENT (No such file or directory)
<multun>
what the fuck
<cole-h>
danderson: Huh, it's not even looking in the expected place
palo1 is now known as palo
<danderson>
I mean, it's not wrong...
<danderson>
that file indeed does not exist
<cole-h>
The fact it doesn't look at tzdata or even /etc/localtime is shocking...
hlisp has quit [Ping timeout: 240 seconds]
<multun>
it looks at /etc/localtime on my ubuntu machine
<multun>
something went horribly wrong there
<cole-h>
Same here on Arch
<cole-h>
...
<cole-h>
danderson: Kill it with fire :D
<danderson>
it's just a nixos 20.03 :(
<energizer>
it looks at /etc/localtime here on nixos 20.03
<cole-h>
🤔
fabianhjr has quit [Quit: Leaving.]
<energizer>
er, nixos-unstable
<multun>
it indeed also looks at /etc/localtime on my 20.03 nixos
<bqv>
yeah, looks at it for my unstable
alp has joined #nixos
<danderson>
okay so... what the hell?
oborot has quit [Ping timeout: 246 seconds]
<bqv>
your machine has a distinct lack of temporal focus
<bqv>
run "fortune" for high gains!
shabius has quit [Ping timeout: 265 seconds]
hlisp has joined #nixos
<danderson>
okay well, nixos seems to have assembled a completely broken system for me, which is nice...
linarcx has joined #nixos
morgrimm has quit [Ping timeout: 258 seconds]
<cole-h>
* assembling a working system left as an exercise to the reader
<danderson>
multun: is your 20.03 up to date?
<bqv>
:D
<multun>
I'll update it
<danderson>
cool, I'll have a broken system buddy :D
<cole-h>
bqv: Looks like the GH comment filter is busted again, so my build command won't do anything until that gets fixed in a few hours. I'm gonna let guix finish building and then go to sleep, and I'll (try to) look at it again tomorrow
<bqv>
ahh
<bqv>
okay
<cole-h>
bqv: Don't be afraid to ping me if I haven't reviewed it by the next time you see me pop up in IRC ;^)
<bqv>
sure :p
<danderson>
rebooting to see if giving nixos a clean slate helps, brb
<cole-h>
inb4broken system
<danderson>
okay, even weirder...
<danderson>
At the display manager login prompt, the time is correct
<danderson>
but then I log in, and it goes wrong again.
<cole-h>
And it still goes looking in glibc?
<energizer>
danderson: what's your locale?
<danderson>
and skipping Wayland, logging in at a tty, the time is correct
<danderson>
so... is gnome messing up my env or something...
Izorkin has quit [Ping timeout: 264 seconds]
reallymemorable has quit [Quit: reallymemorable]
<danderson>
energizer: en_US.UTF-8, per `locale`
<cole-h>
danderson: Is it still busted if you use another DE, or even a WM like i3 or sway?
<multun>
danderson: can you prodive a dump of your environment?
<rnhmjoj>
bqv: actually i want the opposite: i need to format the string "float in range [0.1, 1]" but get "float in range [0.10000, 1]"
<multun>
floscr: create a wrapper script
<bqv>
floscr: i think systemd.services.*.path exists for that reason, iirc it accepts packages directly...
<rnhmjoj>
builtins.toJSON might work, let's try
zeenk has joined #nixos
<bqv>
rnhmjoj: ah, ok
<multun>
floscr: people often name the bare program foo-unwrapped, and create a shell script called foo that sets the environment then calls foo-unwrapped
tertl3 has quit [Quit: Connection closed for inactivity]
<floscr>
So I would then wrap every script in nix with something like `grep=${pkgs.gnugrep}/bin/grep` ?
<rnhmjoj>
bqv: works great, thank you
<floscr>
That's what I thought about doing, just adding these headers to every shell script
therealsarcastic has joined #nixos
Acou_Bass has joined #nixos
<bqv>
:)
<floscr>
Or maybe just source the user bash path (from an rc file) in every script?
<bqv>
floscr: nah, you should be fine to just use .path
<floscr>
which .path?
<bqv>
e.g. you can use systemd.service.myService.path = with pkgs; [ polybar pulseaudio gawk ]
<floscr>
ah
<bqv>
(gnugrep is in there by default)
<bqv>
that's what i meant above
<bqv>
sorry if unclear
<floscr>
Ah got it, that looks good, I'll give it a try, thx
turlando has joined #nixos
opthomasprime has left #nixos [#nixos]
therealsarcastic has quit [Ping timeout: 264 seconds]
rauno has joined #nixos
cybrian has quit [Quit: cya]
<floscr>
bqv: Works! Thanks
<bqv>
np!
sshow has quit [Ping timeout: 244 seconds]
knupfer has joined #nixos
<energizer>
is there a nix code generator library written in go?
lsix has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 merged pull request #86701 → improve robustness of various Rust packages (part 2) → https://git.io/Jfsi1
<cjpbirkbeck>
is possible to change where /bin/sh symlinks to, i.e. change it from /nix/store/...-bash-.../bin/bash to something like /nix/store/...-zsh../bin/zsh or /nix/store/...-dash-.../bin/dash? i remember seeing an option for that once (with warnings), but i can't find it now
Darkmatter66 has joined #nixos
Darkmatter66_ has quit [Ping timeout: 272 seconds]
<bqv>
you can just disable it entirely
<bqv>
but bear in mind that's quite liable to break a few things
<bqv>
environment.binsh
doublej472 has quit [Read error: Connection reset by peer]
<kayg04>
can anyone explain to me how exactly channels should be managed on a server with multiple users?
<kayg04>
Let's say channels are added on a system-level, the root user. and user X wants to update the channels, would them running nix-channel --update have any effect if they themseves do not have any channels?
<edxsa>
How do you all use nixos to replace the server system, I am going to write the nixos installation image to the server disk, load it completely into the ram, and install it, but I am not sure if 2G RAM can put it down.
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « hyperrogue: 11.3f -> 11.3l »: https://git.io/JfsQY
<sephii>
Since upgrading to 20.03 I have a strange issue with my wireless mouse (Logitech M570): while having the left button pressed, I get "ButtonRelease" events even though the button is still pressed. This makes a lot of operations very difficult or impossible (drag n dropping, text selection, moving windows, etc). I tried changing the batteries but it didn't change anything, and using a different mouse (a wired
<sephii>
one) works. Could it be a problem with the new kernel or a driver?
<pjt_014>
I spent quite a bit of time to prevent that
<pjt_014>
you can try it yourself
<MichaelRaskin>
sephii: I think 4_19 is still in master
<jluttine>
weird, my acme renewals aren't running. it's more than 1 month since the previous update although i haven't modified the renew interval.. has there been some changes in nixpkgs that could cause this?
floscr has quit [Remote host closed the connection]
<pjt_014>
sephii: possibly
<armin>
pjt_014: i get a line " 16 linux-2.33.2" and i'm pretty sure i never used kernel 2.33 on this system.
<niso>
jluttine: yes, with 20.03 acme was completeley refactored
<keithy[m]>
Hi guys I am trying to get syncthing working in 20.03 its erroring with [start] 00:49:40 WARNING: Failed to initialize config: config file version (30) is newer than supported version (29). If this is expected, use -allow-newer-config to override.
<niso>
jluttine: the script was replaced with lego
<pjt_014>
armin: that's a weird historical thing (might actually be parts of the 2.x kernel, not sure). replace the 1st [0-9] with [45] to work around it
<niso>
jluttine: s/completely refactored/fixed by moving to another script/
freeman42x has quit [Client Quit]
<jluttine>
niso: ok, yeah, i think i've added email and acceptTerms after that because it complained about those some time ago. i wonder if there's some other config i should add or change to make it work. i wonder what could be wrong..
munksgaard has left #nixos ["WeeChat 1.0.1"]
<pjt_014>
sephii: It'd be a matter of simply commenting out the pkgs_latest line in your config
fendor has joined #nixos
<pjt_014>
it'd possibly be a lot of downloading/rebuilding but you'd know for sure if it was the kernel that was at fault
<pjt_014>
even if that didn't fix it you'd have more info for the bug report
<armin>
pjt_014: that will work of course, assuming that you only ever used kernel versions 4 and 5.
<niso>
jluttine: sadly i can't help you with that since i wasn't able to use the acme script before due to it's limitations
<jluttine>
the latest rows are after reboot about 1 month ago, and the last line is just "Started Renew ACME Certificate for somedomain.com". Note that there's no stopped nor succeeded lines after that. So maybe lego has crashed or something else failed so that it hasn't been able to finish
stree has quit [Read error: Connection reset by peer]
therealsarcastic has quit [Ping timeout: 240 seconds]
<niso>
jluttine: maybe trigger it manually and see what happens?
<jluttine>
niso: yeah, restarting the systemd service manually seems to fix it. weird.
<pjt_014>
armin: I made this script for my own use and just happen to know what major versions I've used--it's more designed for listing the minor versions anyway
<sephii>
pjt_014: what's `pkgs_latest`? I don't have such a line in my config. Can I just set the `boot.kernelPackages` to `pkgs.linuxPackages_4_19`?
<armin>
pjt_014: was just trying to point out a bug, no offense. :)
<pjt_014>
on a terminal line type nix build nixpkgs.linuxPackages_[tab]
<pjt_014>
the autocomplete should let you see all the options
<pjt_014>
maybe not since I only see _latest and not any specific _5 ones but maybe so
<pjt_014>
sephii: if that doesn't work then nix build nixpkgs.hello will download the autocomplete suggestions
hlisp has quit [Remote host closed the connection]
epic_guy[m] has joined #nixos
<sephii>
pjt_014: ah that's nice, didn't know there was autocompletion on that (too bad it doesn't work on fish though). I see there's 5_5 and I'm using 5_4, maybe I'll try upgrading first and see if it helps.
hlisp has joined #nixos
<pjt_014>
sure
<pjt_014>
you didn't happen to use nix-collect garbage -d did you?
<sephii>
Yes I did :(
<pjt_014>
in the future I *heavily* recommend nix-env --delete-generations +2; nix-collect-garbage
<pjt_014>
lets you keep the latest two and trash errything else
<pjt_014>
It's a good compromise on the great primeval urge to delete large amounts of stuff while not borking too much up
<immae>
Hello there, I have many web applications, and I have a cache issue for static files due to the fact that the elements of the nix store are at timestamp 0, how do you deal with those usually? (I’m using apache fwiw)
therealsarcastic has quit [Ping timeout: 260 seconds]
alp has quit [Remote host closed the connection]
alp has joined #nixos
<yuuki_>
I'm trying to learn Nix by equipping some projects with `shell.nix`. In one of them this will make `entr` (file watcher) available, which I'd then like to use in a shell script `dev.sh`. But I couldn't find anything about the recommended way of handling the path to bash in such scripts. Obviously `#!/usr/bin/env bash` won't really work in an environment where everything is handled by nix. Is it the
<yuuki_>
common "best practice" to just use `nix-shell` in the shebang to make sure that "$SOME_SHELL" is installed whenever that particular `.sh` files is sourced/called/used?
<infinisil>
yuuki_: /usr/bin/env bash will work fine, /usr/bin/env is one of the things even available on NixOS
<infinisil>
You can just assume you have all the binaries available in it, and running `nix-shell` in the project will make that true
<jluttine>
is there something wrong in systemd timer if `systemctl status sometimer.timer` shows `Trigger: n/a`?
<jluttine>
that's what my acme timers are showing
k65 has joined #nixos
<yuuki_>
Got it, thanks!
<infinisil>
yuuki_: But if you prefer, you could also package up the script into its own Nix derivation (with all dependencies), and then add that to buildInputs in the shell.nix
<jluttine>
that is, `systemctl status acme-mydomain.com.timer` shows `Trigger: n/a`. sounds suspicious
<infinisil>
Though for just development dependencies I probably wouldn't do that
<yuuki_>
Yeah in this case it's a dev only script akin to watch files and on change rebuild etc.
redkahuna has quit [Ping timeout: 258 seconds]
<infinisil>
jluttine: There was a bug with that recently
<reirob>
and other libraries are missing too: -lc, -lm, -lrt, -lutil, -ldl, -lpthread -lgmp
cosimone has quit [Quit: Quit.]
chagra_ has joined #nixos
<reirob>
The strange thing is, that I have the same packages installed as on the 19.09 VM
ixxie has quit [Ping timeout: 256 seconds]
splintah has quit [Ping timeout: 246 seconds]
<reirob>
I mean in the /etc/nixos/configuration.nix
<srk>
that won't propagate to package build
<{^_^}>
[nixpkgs] @Mic92 merged pull request #86759 → [20.03] kubeval: don't build against schema by default → https://git.io/JfsN7
<{^_^}>
[nixpkgs] @Mic92 pushed 3 commits to release-20.03: https://git.io/JfspU
ixxie has joined #nixos
<srk>
if you build with nix-build the build is isolated and needs all the deps provided explicitely
<srk>
does the project sets CFLAGS, uses pkg-config? I've seen something like this recently but it was due to how the projects buildsystem modified the environment
<reirob>
So, why is ist working on 19.09 and not anymore on 20.03?
hmpffff has quit [Ping timeout: 260 seconds]
<reirob>
Any idea?
<srk>
something change :) you can even grab the repo and bisect between these two
<srk>
*changed
<srk>
(grab the nixpkgs repo)
therealsarcastic has joined #nixos
<reirob>
Changed where? In which package?
alp has quit [Ping timeout: 272 seconds]
<srk>
are you running sandboxed?
<srk>
that's what git bisect can tell you
cfricke has quit [Quit: WeeChat 2.8]
<reirob>
Don't know what sandboxed means
<srk>
build isolation is done via sandboxing, it's enabled by default so that should be good
<reirob>
I have installed 20.03 on my laptop and trying to make work what worked on 19.09 VM
<reirob>
The same .nix packages
<srk>
yes but different nixpkgs
cfricke has joined #nixos
cfricke has quit [Client Quit]
<reirob>
I am just running: nix-build release0.nix
<reirob>
But it seems as if the linker cannot find system libraries
<reirob>
Like libpthread, libc, etc.
gxt has joined #nixos
<srk>
(you can also try with specific compiler via haskell.packages.ghc883 or haskell.packages.ghc865 instead of the default, probably ghc883 on 20.03)
hlisp has joined #nixos
klntsky has quit [Remote host closed the connection]
klntsky has joined #nixos
<reirob>
thank you srk. I still do not understand why this doesn't work in 20.03 when it worked on 19.09
<reirob>
does it mean, that every time when I will make an upgrade things can break?
<reirob>
And how do I try to build with another compiler?
<reirob>
I mean, what should I do with haskell.packages.ghc883?
<reirob>
Do I have to give it to nix-build somehow as an argument?
<reirob>
I think I found!
<srk>
you can switch it in your expression instead of pkgs.haskellPackages you can use pkgs.haskell.packages.<GHCVER>
griff__ has joined #nixos
<reirob>
thank you
<reirob>
trying out immediately
hlisp has quit [Ping timeout: 246 seconds]
<srk>
things shouldn't break and often it's something in your env/setup that could cause breakage. I've never had the issues with linking haskell programs though
<edcragg>
SHAs for git source downloads seem to change whenever you base a branch for a package at a different point in history, i don't really get why
splintah has joined #nixos
ixxie has quit [Ping timeout: 265 seconds]
<reirob>
finished compiling, but same error :(
<LnL>
are you using leaveDotGit/deepClone?
philr has quit [Ping timeout: 256 seconds]
<reirob>
collect2: error: ld returned 1 exit status
<reirob>
For me it looks like the linker called by GHC (seems to be ld.gold) does not find these libraries.
<reirob>
Are they installed? How can I know? If no, how can I install them. But shouldn't they be dependencies of the GHC packages that are using them?
<clever>
vandenoever: how much ram is nix using?
<reirob>
to give more info, few lines deeper there are lines like:
hlisp has quit [Ping timeout: 265 seconds]
<edcragg>
LnL: i'm using deepClone, yes. so it hashes all the git objects downloaded, and if a submodule upstream changes, they get hashed too, even if the submodule ref is the same? i'm guessing...
<vandenoever>
clever: i did not measure, i switch from fetchgit to fetchGit now and that does work
<LnL>
edcragg: yeah that keeps all the git data which will change when the repo updates
<vandenoever>
clever: i think fetchgit was loading the whole dir (with builds) in memory
<edcragg>
LnL: thanks, is there a risk that it might change on any given build? seems like that could be a problem...
cosimone has joined #nixos
vidbina has quit [Ping timeout: 256 seconds]
<LnL>
yeah, I wouldn't recommend using deepClone if possible
<edcragg>
hmm, interesting. i thought i needed it to get a recursive pull for submodules
knupfer has quit [Read error: Connection reset by peer]
<LnL>
I mean fetching each repository separately
<reirob>
@srk, any idea what else I should try?
<edcragg>
LnL: and then it's easy enough to combine the paths into the source for the build? i'll have to look into that
<srk>
reirob: updating the channel?
<srk>
reirob: do you possibly have any overlays installed?
<srk>
reirob: also you forgot to paste more info :)
<srk>
reirob: this is the proper shell command nix-shell -p "pkgs.haskell.packages.ghc883.ghcWithPackages (p: [ p.ghci ])"
<hyper_ch>
clever: may I pick your brain for nginx stuff?
<reirob>
Now for your question regarding overalys. I will make grep -i overlay /etc/nixos/* : there is nothing (well actually a commented section for when I was trying to make the nvidia card working, but it's commented)
<{^_^}>
[nixpkgs] @flokli pushed 2 commits to release-20.03: https://git.io/Jfsh5
<yorick>
why is programs.ssh.knownHosts a loaof?!
syd has quit [Quit: Connection closed]
lsix has joined #nixos
<reirob>
I have another suspicion why the linker does not work on my 20.03 machine. On this machine I put the Haskell nix packages (ghc, stack, cabal2nix) in /etc/nixos/configuration.nix, whereas on the 19.09 vm I was installing them with nix-env. Can this be the reason?
<reirob>
And now trying to build with nix-build release0.nix
<srk>
,imperative reirob
<{^_^}>
reirob: nix-env has multiple drawbacks as an imperative package manager. nix-env -u will sometimes upgrade to the wrong thing; the outputs to install are very finicky to override; and packages that have been removed or are otherwise unavailable will remain in your profile without any warnings. Consider using a ,declarative setup instead.
<reirob>
just to confirm, the nix-env --install stuff didn't help :(
lsix has quit [Ping timeout: 272 seconds]
<reirob>
Ok, now I will read about how to build a nix-shell and try to build the stuff inside the nix-shell
<adisbladis>
reirob: I was just gonna say that. Don't do global installs of development dependencies.
<reirob>
@adisbladis : but this was the reason for me to switch to nixos. That I have a central file with all the packages and configurations that I need for my work
lsix has joined #nixos
<reirob>
So when I need to install on a new machine I just have to copy my data and the configuration.nix and reinstall nix
<reirob>
this was one of the main reasons
hlisp has joined #nixos
<reirob>
and the fact that all the haskell people are saying that nix is good for haskell development
cmacrae` has quit [Ping timeout: 272 seconds]
<srk>
you can improve that a bit later, e.g by installing one ghcWithPackages with the package set you need (even via home-manager) but for projects nix-shell is enough
<reirob>
anyway, thank you so much for the pointers so far. Will look now how to make a nix-shell and compile inside the shell.
<jD91mZM2>
Hi, so I'm trying to port Oomox to Nix, so I can generate a GTK theme that follows my global colorscheme. It's not going great... Mainly, python complains about missing Gtk. Could anyone care to take a look? https://gist.github.com/jD91mZM2/828e244ff327133a97b71ba5661d50cd
<adisbladis>
reirob: NixOS is for making a reproducible operating system, nix-shell is for reproducible development environments
<davidtwco>
reirob: If I can't put a `shell.nix` in-tree, then I just keep them in my Git repo alongside `configuration.nix`, so it isn't much different from the workflow you're suggesting when going to a new machine - except I need to run `nix-shell` (which isn't that much less convenient, and being able to share your `shell.nix` is nice).
<{^_^}>
[nixpkgs] @expipiplus1 opened pull request #86773 → nodePackages: init he and escape-string-regexp → https://git.io/Jfsjm
<adisbladis>
Let's say you're hacking on something linking against a libjpeg & invokes gcc
<adisbladis>
If you add packages in systemPackages (or with nix-env) you won't have the required environment to find/link against those libraries
<adisbladis>
Which you will in nix-shell
<reirob>
thank you @davidtwco, @adisbladis and @srk, I will definitely consider this. I guess it is just the time to get used to the new tools :)
<srk>
yw! exactly!
<notgne2>
jD91mZM2: I've never heard of this project and it looks pretty cool, I actually already have my Nix configuration theme everything but GTK using base16 colors already, so I'd be interested in this working too
hlisp has quit [Ping timeout: 272 seconds]
<notgne2>
jD91mZM2: if you make any progress please give me a ping, unfortunately I can't say much to help with the GTK issue
<jD91mZM2>
notgne2: I've manually patched the Yaru theme for this before, but it's a pain to keep it up to date lol. Yeah sure, I'll give you a ping :)
CcxWrk has quit [Remote host closed the connection]
CcxWrk has joined #nixos
ccppuu has joined #nixos
<{^_^}>
[nixpkgs] @sternenseemann opened pull request #86777 → perlPackages.StringSimilarity: init at 1.04 → https://git.io/JfGeE
<maralorn>
chrismatheson: I have environment.variables = genAttrs [ "CURL_CA_BUNDLE" "GIT_SSL_CAINFO" "SSL_CERT_FILE" ] (_: "${pkgs.cacert}/etc/ssl/certs/ca-bundle.crt"); in my config. That fixes cert problems like that.
<maralorn>
Allthough there probably is a proper fix which should probably be upstreamed to nixpkgs.
stree has quit [Read error: Connection reset by peer]
<maralorn>
Interesting: To test that my hackage2nix.yaml is fine, I’d like to run hackage2nix. command-not-found tells me to install cabal2nix, but a nix-shell with cabal2nix does not contain hackage2nix …
<maralorn>
hyper_ch: It‘s a nice feature rich shell?
numkem has joined #nixos
<chrismatheson>
yeah into zshrc
<chrismatheson>
hyper_ch: b/c/ its the default shell from the next version of OSX.
<chrismatheson>
as much as possible i try to jusst adjust to defaults rather than customize everything :)
<hyper_ch>
is OSX still using bash v3.x?
<chrismatheson>
shrugs i just use whats there :)
<hyper_ch>
last time I had to touch OSX it was an ancient bash v3 version
therealsarcastic has quit [Ping timeout: 260 seconds]
<emily>
they will never ship bash 4 because it's gplv3
fusion809 has joined #nixos
<emily>
hence the move to zsh
<hyper_ch>
bash 4 offers a lot of goodies
<hyper_ch>
ha, tomorrow new cell phone arrives (I think)
<emily>
I don't think Apple are going to be convinced by a few goodies when they essentially funded the development of a production-ready C/C++/Objective-C compiler to avoid GPLv3 :p
<emily>
but really, zsh has lots of fancy stuff too, no need for a shell war
<jakobrs>
malalorn: Works for me
<jakobrs>
nix run nixpkgs.cabal2nix -c which hackage2nix
knupfer has quit [Quit: knupfer]
<numkem>
is it possible to set up an overlay that we could take a pacakge we'd write that would have gone to nixpkgs but just put it in an overlay (without having to add the super/self)
<maralorn>
jakobrs: Which branch?
<jakobrs>
branch of what?
<maralorn>
And can you tell me which cabal2nix version?
<maralorn>
jakobrs: Of nixpkgs. I guess I mean channel.
<jakobrs>
nixpkgs-unstable
<{^_^}>
[nixpkgs] @sternenseemann opened pull request #86780 → perlPackages.XMLRPCLite: init at 0.717 → https://git.io/JfGvX
<jakobrs>
hackage2nix 2.15.1
orivej has joined #nixos
<jakobrs>
nixpkgs commit 10100a97c89 (the latest release of nixpkgs-unstable)
<reirob>
But I still have the same problem as before, my haskell project does not link. It compiles but in the end I get:
<reirob>
collect2: error: ld returned 1 exit status
<reirob>
`cc' failed in phase `Linker'. (Exit code: 1)
<reirob>
builder for '/nix/store/a50lvmpc1pvmfs30jaw0h5mj687zmpxf-rtetote2-0.0.0.1.drv' failed with exit code 1
<reirob>
What I was doing:
<reirob>
cabal2nix --shell . > shell.nix
<reirob>
nix-shell --command "cabal configure"
<reirob>
nix-build shell.nix
waleee-cl has joined #nixos
<reirob>
So now I am using only stuff in an environment, but it still has problems with linking!
<reirob>
I am on NixOS "20.03"
<reirob>
What am I missing?
<reirob>
@srk, are you still here by any chance?
jakobrs has quit [Quit: WeeChat 2.8]
<srhb>
reirob: Sounds like this is a package specific problem. What's your expressions and dependencies? The error line you gave does not have enough context either. :)
rauno has quit [Remote host closed the connection]
<reirob>
Can I paste here the shell.nix?
<srhb>
reirob: No, please use some pastebin.
<reirob>
ok, let me find out how to do this. just a moment
<reirob>
I guess the interesting part is lines 336 to 350
<reirob>
The same GHC package worked on another machine with NixOS 19.09
<reirob>
Though there I did not do the nix-shell way, but today @srk and other nice people were suggesting me to build a proper shell.nix file. But it is still not linking
<{^_^}>
[nixpkgs] @cdepillabout pushed 2 commits to haskell-updates: https://git.io/JfGLJ
reallymemorable has joined #nixos
sarcasticadmin has joined #nixos
<ixxie>
domenkozar[m]: I'm interested in developing a NixOps backend for Hetzner cloud features; would it better to keep it seperate from the current Hetzner backend, or contribute such features to the existing backend?
<{^_^}>
nix#2431 (by sgraf812, 1 year ago, open): fetchGit fails with a not very helpful error message when fetching a revision not in the remote's HEAD
<srk>
reirob: it works now *without changes*, git --reset HARD, git pull and ./new test
<srk>
reirob: nix-build
<reirob>
ok
<reirob>
let me try again
<srk>
reset is wrong, sorry
<srhb>
evanjs: iirc by default it tries to reach it from ref = master
<srk>
reirob: git reset --hard HEAD
<srhb>
evanjs: Which branch did you push it to?
<evanjs>
srhb: okay that makes sense. I read that in the thread but I needed to read it a different way before it clicked, thanks :D
<evanjs>
feature branch :D
<evanjs>
lol
<evanjs>
so rev isn't enough, but simply specifying ref is enough as well
<reirob>
@srk : it worked!
<srhb>
evanjs: Right, that's the gist of the issue you linked. :)
<srk>
reirob: \o/ now you can compare the two ;)
<evanjs>
srhb: yup. it's early and I haven't gotten my refill of caffeine yet. Sometimes I need a few rereads/rewordings before things click haha. Thanks!
justanotheruser has quit [Ping timeout: 272 seconds]
<evanjs>
srhb: yeah I think I was mainly misreading master as HEAD, thus my confusion :P now, to patiently wait for https://github.com/NixOS/nix/pull/3166 to make it into a release...
<aanderse>
anyone ever had this before: provisioned a user, they were the second user on a system so got id 1001, when they login and type "ls" they get "--color=tty: No such file or directory" as an error, no output :S
sarcasticadmin has quit [Quit: WeeChat 2.3]
<{^_^}>
[nixpkgs] @Mic92 merged pull request #85603 → edk2: Support build on macOS → https://git.io/JfUpe
<immae>
aanderse: maybe ls has an alias in your system profile/bashrc that looks like alias ls='ls --$SOME_VAR_THAT_HAPPENS_TO_BE_EMPTY_FOR_1001 --color=tty'
<evanjs>
interesting... my problem might also have been related to nix-prefetch-git in general... I used `--branch-name`, but while it stated it was checking out the desired branch, it didn't seem to give me the rev of such
<reirob>
This is crazy. I took your shell.nix, your default.nix, renamed test to the name of my cabal file without extension
<reirob>
it didn't work
<aanderse>
clever++
<{^_^}>
clever's karma got increased to 405
morgrimm has quit [Ping timeout: 272 seconds]
<aanderse>
thats a preemptive ++ because if clever is on the case i know the problem will be solved :)
<clever>
lol
<reirob>
Then I compared the cabal files and the only difference is that in my cabal file I have several Executable stanzas!
<aanderse>
clever: type ls returns "ls is aliased to `ls --color=tty'"
<clever>
aanderse: what about `which ls`
eyenx has joined #nixos
eyenx has quit [Changing host]
eyenx has joined #nixos
<reirob>
So my conclusion is that callCabal2nix doesn't work when the cabal file tries to build more than one executable
<reirob>
or so
tobi has joined #nixos
<srk>
oO interesting!
<clever>
reirob: what if you run cabal2nix on the cabal file manually?
<srk>
were they named the same?
tobi is now known as Guest88005
<aanderse>
clever: /home/username/.nix-profile/bin/ls ... but now i'm super confused because this user has no idea what nix or nixos is... so using nix-env seems unlikely, will verify :S
<reirob>
@clever this is what I was doing this morning, before @srk taught me how to use callCabal2nix
<reirob>
it didn't work still
<clever>
aanderse: `command ls -l /home/username/.nix-profile/bin/ls`
<aanderse>
user confirms they have no idea what nix-env is and never typed "nix-env <anything"
<reirob>
@srk, what was named the same?
<reirob>
U mean the executables?
<clever>
aanderse: a common issue, is that `comand-not-found` will tell you to install busybox, because it has everything
<reirob>
No, they were different
<clever>
aanderse: and busybox's ls likely doesnt support --color
<srk>
reirob: that's weird, I have few projects with mutliple executables and it works fine
chloekek has joined #nixos
<aanderse>
lol ok user lied, there is one nix-env in history :P
<aanderse>
nix-env -iA nixos.klibcShrunk
<Guest88005>
hello, I'm not sure if this is the right place to ask, but is there any reason why https://github.com/NixOs/nixpkgs/pull/22789 isn't getting merged? It has been open since more than three years
<{^_^}>
NixOs/nixpkgs#22789 (by e-user, 3 years ago, open): freeipa: init at 4.4.3
<clever>
aanderse: never trust the user :P
<clever>
aanderse: always confirm what they claim
<aanderse>
can i remove the ability to call nix-env on a system?
<immae>
aanderse: you can make it less automatic by removing the ~/.nix-profile/bin from th ePATH? this way the user has to know what he does to use nix-env?
proofofkeags has quit [Remote host closed the connection]
<clever>
immae: the bashrc stuff in nixos will put it back in PATH on login
proofofkeags has joined #nixos
<immae>
ah
<reirob>
@srk, no idea if lowercase does something different, but I was using these executables now for 3 years :)
<srk>
aanderse: heh, that could be a home manager tweak :D
<reirob>
started with cabal, then at some point used stack (all this was on ubuntu)
<reirob>
now I want to use nix
<reirob>
ok, I will try if this works when I make upper case executables, even though it is not very unix-style ;)
<clever>
reirob: it should be all lowercase, but what you pasted had a E in it
lunik1 has joined #nixos
<reirob>
Ah, you mean not the name of the executable programs but of the "Executable" keyword?
<mikky>
hey, I'm trying to understand the idea of lib.mkDefault, lib.mkOverride and lib.mkForce. Can't find any docs and reading the source code confused me even more (mkDefault is basically an alias for mkOverride 1000). What are their respective use cases?
<aanderse>
well, thanks as always clever :)
o1lo01ol1o has joined #nixos
<clever>
reirob: yeah
<lunik1>
I am trying to set up ssh on initrd. Does the private key file have to be somehwere that is mounted for initrd?
<reirob>
ok, changing to lower case "executable" and retrying
<reirob>
@srk lower case "executable" doesn't make a difference
<reirob>
i mean fails too
<clever>
reirob: can you pastebin the entire cabal file that is failing?
<reirob>
yes
dingenskirchen has quit [Remote host closed the connection]
ambroisie has quit [Read error: Connection reset by peer]
o1lo01ol1o has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
noudle has joined #nixos
drakonis has joined #nixos
ambroisie has joined #nixos
<mikky>
clever: let's assume the option is a primitive. iirc, those can only have a single value and cannot be merged (yields error somewhere along "defined more than once"). From what you said, it seems to me that "option = value" will win over "option = mkForce value", which doesn't make sense to me.
morgrimm has joined #nixos
<clever>
mikky: mkForce has a stronger (smaller number) then the normal setting
<mikky>
clever: huh, must have misread something. I think I've read that "normal" priority is 10
<clever>
1500 is the default= inside an option, 1000 is mkDefault, 50 is mkForce, 10 is mkVMOverride
<jumper149>
How does services.openssh add port 22 to networking.firewall.allowedTCPPorts? Also: Why don't I get something like "attribute already defined" when I set it in my configuration.nix. And most importantly how does it still append/prepend port 22 if I set networking.firewall.allowedTCPPorts in configuration.nix?
<AmandaC_>
So I've got hydra on one of my nixos machines. It seems `nixos-rebuild` is failing due to dqlite being flaky, but hydra has a successful build form an hour ago, presumably including that broken package, is there any way I cana take the nix store path given from hydra and boot it?
AmandaC_ is now known as AmandaC
tokudan has quit [Remote host closed the connection]
bgamari has quit [Remote host closed the connection]
<virus_dave>
hey folks - is pkgs.fetchzip nondeterministic (across darwin & linux)?
chagra_ has quit [Ping timeout: 265 seconds]
Rusty1 has joined #nixos
<clever>
virus_dave: it should be deterministic, but darwin is case insensitive by default, which can cause issues
rauno has quit [Remote host closed the connection]
Rusty1 has quit [Client Quit]
<virus_dave>
oh interesting - would that be in the case of files with "overlapping" names differing only by case? or some other problems as well?
bgamari has joined #nixos
<gchristensen>
yea
Neo--- has quit [Ping timeout: 240 seconds]
growpotkin has joined #nixos
<infinisil>
adisbladis: Hey, I've got something I want to implement for poetry2nix
<infinisil>
And I just wanna check if that's something you wouldn't be opposed to
<infinisil>
Namely: Adding a way to have a development environment for an application, by building a python environment which has an .egg-link package referencing the mutable src directory
<gilligan>
infinisil: i was talking to adisbladis the other day concerning poetry2nix. I think it's important to keep the scope of poetry2nix managable and to figure out what the most important usescases are. I promised to put something into writing but still have that on my TODO list..
asbachb has quit [Ping timeout: 245 seconds]
<morgrimm_>
Anyone using flakes in their config right now with home-manager?
Neo--- has joined #nixos
<infinisil>
gilligan: Is the discussion public?
<gilligan>
infinisil: most of the discussion needs to still happen in the first place ;-)
<infinisil>
Ahh
butcher has joined #nixos
<gilligan>
my main point/concern is that it's dangerous to design any application based on "i would find XYZ useful"
<infinisil>
Hm yeah
<gilligan>
which isn't to say that XYZ cannot be a totally valid and useful feature
<gilligan>
but there needs to be something to align this with
<gilligan>
infinisil: i'm happy to share with you whatever i come up with
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
griff__ has quit [Quit: griff__]
<infinisil>
gilligan: So the need for this shell feature at least is pretty clear to me: Nix projects should always also have a shell environment you can use to develop the project in, and it's what `poetry shell` does already but in a non-Nix way
shafox has quit [Remote host closed the connection]
<infinisil>
So it's pretty much exactly what the name poetry2nix would imply, making a poetry project work in a Nix way
o1lo01ol1o has joined #nixos
<ixxie>
so it would hotload the directory into the shell to give you a live environment as you are coding?
<infinisil>
ixxie: Python would know that ./src is the package you're developing, meaning any changes to the files would be reflected with any python execution
<gilligan>
infinisil: like i said, personally i'd prefer to have a minimal document outligning poetry2nix main usecase scenarios before changing/altering the API further
vandenoever has quit [Read error: Connection reset by peer]
<infinisil>
And I'm saying that I feel like this should be part of it :)
<infinisil>
Or alternatively, change the module to support multiple instances
<azazel>
mmm... I've updated my nixos-unstable and now the terminal hasn't a $TMP dir setup? do anyone knows why?
<ixxie>
hmm kinda funny there is no way to declare some kind of 'alias' of a module which functions exactly like the original but doesn't clash in the namespace
reallymemorable has quit [Quit: reallymemorable]
<azazel>
I think that there is a way to "republish" a module with another prefix
<infinisil>
Because e.g. what if a service needs /etc/service.conf
<ixxie>
there is a service called minio, which is an S3 clone. Their documentation advocate running seperate instances to support multitenancy, and indeed their server doesn't really support mutliple keys
<clever>
ixxie: then you want to create a minio@foo service in systemd
<ixxie>
arianvp[m]: any alternatives? I mean if there is something comparable that is easier to do multitenancy with.... :D
<arianvp[m]>
Ceph
<arianvp[m]>
It has block storage and is actually battle tested
mr_sm1th has joined #nixos
<arianvp[m]>
I've had many data loss issues with minio in the past. And it doesn't document anything about disaster recovery. Ceph is well-documented and wellmaintained
zupo has joined #nixos
<ixxie>
clever: ah so I make my own systemd service to run as many instances as I want
<clever>
ixxie: yeah
<clever>
ixxie: look at openvpn or wireguard for examples
lsix has quit [Quit: WeeChat 2.8]
<clever>
ixxie: nixos lets you define a services.openvpn.foo and then start openvpn@foo
<clever>
and if you need a 2nd vpn link, it just makes another instance
<keithy[m]>
Hi guys, Hi guys I am trying to get syncthing working in 20.03 its erroring with [start] 00:49:40 WARNING: Failed to initialize config: config file version (30) is newer than supported version (29). If this is expected, use -allow-newer-config to override.
<{^_^}>
[nixpkgs] @adisbladis opened pull request #86830 → podman: Don't symlink the $bin output → https://git.io/JfGce
<keithy[m]>
I could use some clues on overriding ExecStart
<lassulus>
I'm running a hybrid, using mostly stable and some packages from unstable
<keithy[m]>
how does rebuild pick up changes?
<keithy[m]>
I attempted to change the guiAddress listening ip and expected it to restart the service, but it didnt.
<{^_^}>
[nixpkgs] @eadwu opened pull request #86835 → modules/filesystems/zfs: use current kernel → https://git.io/JfGcb
knupfer has quit [Ping timeout: 260 seconds]
<clever>
keithy[m]: rebuild checks if the file in /etc/systemd/system/ changed
<{^_^}>
[nixpkgs] @vbgl opened pull request #86836 → ocamlPackages.async: remove at 112.24.00 → https://git.io/JfGcj
<maralorn>
Does anyone know what‘s up with the channel releas-process? Afaict release-20.03 fulfills all properties for the release, but apparently all release services are failing: https://status.nixos.org/prometheus/alerts
<keithy[m]>
so if the configuration file changed it might not pick it up
<clever>
keithy[m]: depends on how its told to use that file
<clever>
keithy[m]: the guiAddress is directly in the ExecStart line of the .service file, it should restart when changed
<keithy[m]>
ok
<clever>
keithy[m]: can you confirm the new address is in /etc/systemd/system/syncthing.service ?
<keithy[m]>
so Im not changing it correctly
<maralorn>
keithy: In general you should re set the stateVersion to what it was before. It is not supposed to be changed. It is a way for nixos to with which configuration you installed nixos so that it does not break your stateful data on disk.
<keithy[m]>
not changed
<clever>
keithy[m]: how did you try setting guiAddress?
<clever>
keithy[m]: and what does `nixos-rebuild switch` output?
<keithy[m]>
btw does that need a sudo normally?
<clever>
keithy[m]: switch always needs root
<{^_^}>
[nixpkgs] @rby opened pull request #86837 → quicklisp-to-nix: fix missing '&&' after test command → https://git.io/JfGCt
<keithy[m]>
got the change now!
<energizer>
clever: i missed the context but i wouldnt think "0.0.0.0:8384" is the right choice for syncthing, that seems insecure
<energizer>
keithy[m]: ^
justanotheruser has quit [Quit: WeeChat 2.7.1]
<keithy[m]>
on an internal network, I wasnt panning to ssh tunnel in to the server.
justanotheruser has joined #nixos
<energizer>
defense in depth
<danderson>
anyone have an example of building a derivation in shell.nix and then including it as a buildInput?
<danderson>
using lorri to set up my dev env, and I need a custom build of a package.
<clever>
danderson: let foo = stdenv.mkDerivation { ... }; in stdenv.mkDerivation { name = "shell"; buildInputs = [ foo ]; }
<cole-h>
^
<keithy[m]>
now to learn how to set this thing up
seku has joined #nixos
<danderson>
Thanks! I raced with your reply and figured it out, glad to see I got it right :)
<danderson>
(first rule of the internet: the way to figure out the answer to a question is to ask a hundred people something dumb, which makes you figure it out 5s later)
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<fonnyx>
Hello. I am trying to package some awful install script that tries to write to /usr/ during the installation. What is the best way to deal with this? (Some sort of chroot thing?)
<infinisil>
fonnyx: Patching the install script
<fonnyx>
Hmm. It's an ELF actually. Or I would do that.
<clever>
sphalerite: is that `nix run --store` also chroot'ing the `hello` binary as well??, or was it just working by chance due to deps in the real /nix/store ?
chloekek has quit [Ping timeout: 260 seconds]
chloekek has joined #nixos
<sphalerite>
clever: yes, nix run does the namespacing for the stuff inside as well :)
<sphalerite>
clever: the older tools like nix-shell don't though.
<clever>
sphalerite: nice, but that will likely fail when debian disables user namespaces
<sphalerite>
yes, the blog post mentions that it requires user namespaces.
<sphalerite>
Why is debian going to dosaible user namespaces?
<clever>
sphalerite: they patched linux to make namespacing require root by default, for any operation
<cole-h>
qyliss++ Thanks for your feedback, it was extremely helpful. (As you probably already know) I unmarked the doas PR as WIP :D
<{^_^}>
qyliss's karma got increased to 58
fenedor has joined #nixos
ixxie has quit [Ping timeout: 260 seconds]
MichaelRaskin has quit [Ping timeout: 256 seconds]
user_0x58 has quit [Quit: Leaving]
ixxie has joined #nixos
selfsymmetric-mu has joined #nixos
<kl3>
hi, i've got a few simple patches for nix to build on openbsd, is there a mailing list i can send them to? last time i tried discourse for submitting patches it was plain broken and i do not use github, running out of options here.
fendor has quit [Ping timeout: 260 seconds]
maddo has quit [Quit: See ya]
fresheyeball has joined #nixos
<simpson>
That is a good question. I think that git and GH are pretty central to the contributor workflow, so while it's probably possible for somebody to take your patches on an ad-hoc basis, I'm not sure what long-term contributions would look like.
<fresheyeball>
how can I find out the specific git hash of the nixpkgs on my local nix_path?
mammothbane has joined #nixos
<kl3>
simpson: well, i'm happy to push my fork somewhere and have it pulled/merged with git, whatever works to contribute without registering with github
konobi has joined #nixos
fenedor has quit [Read error: Connection reset by peer]
morgrimm_ has quit [Ping timeout: 258 seconds]
<mammothbane>
anyone familiar with ruby packaging? i'm trying to package a local gem and put it in the load path for a bundlerEnv-derived ruby wrapper
MichaelRaskin has joined #nixos
<mammothbane>
but adding the local gem to bundlerEnv's buildInputs doesn't work, and afaict bundix doesn't support local gems yet. mostly just asking if it's even possible with the current machinery
<{^_^}>
[nixpkgs] @leenaars opened pull request #86851 → libxcrypt: init at 4.4.16 → https://git.io/JfG8A
EdLin_ has quit [Remote host closed the connection]
mammothbane has joined #nixos
<virus_dave>
another QQ: Are there standard ways of wrapping OSX .dmg or .pkg installed GUI applications? I've seen the `undmg` package, but that only seems to unwrap the disk image, not wrap the contained GUI app
tnias has joined #nixos
morgrimm has quit [Ping timeout: 272 seconds]
philr has joined #nixos
mbrgm has quit [Ping timeout: 244 seconds]
mbrgm has joined #nixos
ixxie has quit [Ping timeout: 240 seconds]
Fare has joined #nixos
ixxie has joined #nixos
zxcvjlu has quit [Remote host closed the connection]
<glittershark>
Has anyone else run into issues with (I assume) locales in dockerTools-built docker images?
<glittershark>
<glittershark>
specifically I'm seeing a jar that's decoding ü as ?
justanotheruser has quit [Ping timeout: 260 seconds]
<glittershark>
in general how does nixos handle locales?
<clever>
glittershark: for a docker container, try adding glibcLocales to the contents list
<azazel>
oh, nice to know
<glittershark>
ooh
<glittershark>
I already tried `locale` but I'll try that too, ty
<{^_^}>
[mobile-nixos] @samueldr opened pull request #138 → README: revise README, appearance and content → https://git.io/JfG0f
detran has joined #nixos
<hr[m]>
Could a maintainer please merge https://github.com/NixOS/nixpkgs/pull/86853 I'm one of the package maintainers for the PR but the other maintainer has not been active on github for > 2 months and did not merge the previous PR.