<dansho>
i'm trying to use fetchgit in combination with callCabal2nix, but i think it's failing because the repo doesn't have a .cabal file in the root (its in ./massiv/massiv.cabal). maybe? https://paste.ee/p/mZRsN
<KarlJoad>
dansho: You should be able to change $sourceRoot?
Rusty1 has joined #nixos
<cole-h>
(e.g. sourceRoot = "./massive"; I think)
<rmcgibbo[m]>
nix build question: if you invoke `nix build` and it builds multiple drvs, is there any way from the logs to figure out how much time each drv took to build?
<rmcgibbo[m]>
* nix build question: if you invoke `nix build` and it builds multiple drvs, is there any way from the output or logs to figure out how much time each drv took to build?
<colemickens>
is the system architecture available in an ouptut attribute? config.system.build.system? or something?
rb2k has quit [Read error: Connection reset by peer]
_d0t has quit [Ping timeout: 260 seconds]
<colemickens>
aha, it's in config.nixpkgs.system (from reading eval-config.nix)
<energizer>
rmcgibbo[m]: (i'm a novice) i've been wondering that and afaict the only way to do it currently is either running each build interactively and timing it or looking at the birth/modify times on the log file
<energizer>
rmcgibbo[m]: there is an open issue for better logging in nix but no progress in a long time iirc
<energizer>
rmcgibbo[m]: nix-output-monitor parses the build log interactively
<colemickens>
(I'm filtering over nixosConfigurations in my flake and want to filter on the arch it was built for)
<rmcgibbo[m]>
energizer: I suppose birth/modify times on the log file are okay. So far, I've been getting log files only from `nix log {drvpath}`, but I suppose they must be store somewhere on the system...
<energizer>
rmcgibbo[m]: /nix/var/log
lordcirth has quit [Remote host closed the connection]
<rmcgibbo[m]>
That'll have to do (and seems like it might be correct-ish). Thanks for the suggestion.
<rmcgibbo[m]>
energizer:
lordcirth__ has joined #nixos
<KarlJoad>
Since I can't seem to find quite what I would like to test with, does someone have an arduino who is willing to test something for me?
<V>
rmcgibbo[m]: regarding the long message thing earlier, each time you edited that, it sent a new link, because it was all logically one message. even though the code part didn't really change
meck has quit [Ping timeout: 260 seconds]
<rmcgibbo[m]>
Oops, I apologize. Thanks for letting me know.
mbrgm_ has joined #nixos
<V>
I had to click on three links, just to find out what you wrote (when two of them were adding a single character here or there)
<rmcgibbo[m]>
Understood.
<V>
so with IRC it's better to split code out from commentary, and to make "edits" by either writing a sed that would fix your message, or putting it on a new line w/ an asterisk
epta has quit [Ping timeout: 240 seconds]
<V>
since we're all living in the 90's ^^'
lordcirth has quit [Ping timeout: 265 seconds]
mbrgm has quit [Ping timeout: 260 seconds]
mbrgm_ is now known as mbrgm
<V>
don't get me wrong, I'd genuinely love it to have the ability to edit messages; indeed, I think that IRC has a lot of faults
nwm__ has joined #nixos
rb2k has quit [Read error: Connection reset by peer]
myme has quit [Ping timeout: 260 seconds]
ericnoan has quit [Ping timeout: 260 seconds]
lordcirth__ has quit [Read error: Connection reset by peer]
lordcirth__ has quit [Remote host closed the connection]
derpadmin has quit [Remote host closed the connection]
ktosiek has quit [Ping timeout: 264 seconds]
<energizer>
rmcgibbo[m]: the other column i've been wondering how to fill is which host it built on, and i dont know how to do that without parsing directly from nix-build
<Fare>
Ouch. Sometime last month or so I updated NixOS and with it CUPS... and now the PPD I had crafted for my Brother color printer doesn't work anymore, resulting in PostScript source being printed rather than correctly interpreted.
<c3hou>
Fare: ouch... which printer is it?
<Fare>
If I try a different PPD I can get the printer to print B&W, but I can't any way to have it print in color from Linux anymore.
<Fare>
Brother MFC-9970CDW
<Fare>
I tried a PPD from a Mac, but the PPD has lots of references to files in /Library/Printers, that, if commented out, result in the same mess.
<superloach>
looks like the live image is getting stuck on "waiting for device /dev/root"
<Fare>
I also tried different Brother printers; some work in B&W. One worked in color... with colors inverted(!) when printing a pdf.
<superloach>
colemickens: any idea how i could remedy this? I'm sure the device is somewhere in /dev, but it's not autodetecting
<superloach>
appears that internal keyboard isn't working, so I'll have to get a usb to drop into a shell lol
<superloach>
hmm, quick assessment reveals that the drive is *not* in /dev
<superloach>
I'd expect it at /dev/mmcblk1* and it's not there
h0m2 has joined #nixos
<superloach>
as well as the internal on /dev/mmcblk0*, which isn't there
detran_ has quit [Ping timeout: 264 seconds]
h0m1 has quit [Ping timeout: 265 seconds]
<bqv>
infinisil: you decided against the backwards-compat?
<colemickens>
superloach: what kind of computer is this?
<superloach>
it's a chromebook, model CYAN, running UEFI firmware
<superloach>
I don't expect this to be easy, though I haven't had as much trouble with other distros
<superloach>
this would probably be appropriate for #nixos-exotic, but they don't seem very active
<superloach>
I'd have a feeling it'll just be a kernel param from the grub menu or something
<samueldr>
superloach: how did write the live image to USB?
<superloach>
I wrote it to SD, not USB, and I just used `dd`
<samueldr>
there is a kernel param (I forget) if you try to add it to "multi boot" kind of linus usb
<samueldr>
ah, if it's only sd to block storage it should be fine...
<samueldr>
... except that maybe the driver isn't present by default in the initrd
<superloach>
there's no mmc drives showing up
<superloach>
CYAN also has mmc for internal
<superloach>
which is typically mmcblk0, and sd is mmcblk1
<samueldr>
do you know on other distros which driver is used for that?
<superloach>
arch works fine
<samueldr>
we could easily confirm that it's missing from the live image
<superloach>
I also recall that ubuntu works
<samueldr>
well on the initrd for the live image
<superloach>
ah right
<superloach>
I'll see if I can find what's needed
<samueldr>
if you lsmod from any of those two, and share, it might be obvious enough
<samueldr>
meanwhile, do you have an usb drive to try?
<samueldr>
I would assume using the docker image can work too (if that's the kind of thing that you prefer)
<samueldr>
Fare: (I don't know) wouldn't that be similar to redirecting a keyboard to only when it's expected to communicate?
<superloach>
samueldr: swag, I'll use that
<Fare>
maybe. When working at $ELGOOG, we had yubikeys that never output anything to console, but worked beautifully with the entire corporate single-sign-on thingie.
<Fare>
I have no idea what they were using. Maybe I can locate someone inside who can tell me if that part was opensource.
<samueldr>
actual yubikey protocol or another?
<samueldr>
maybe that was the secret sauce
<Fare>
I have no idea, I was just a user of that thing, and when I bought one, I half-expected to have it work essentially the same way.
<samueldr>
superloach: if you wanted to use the stable release, you'll want to checkout the nixos-20.09 branch, if you wanted to use unstable nixos-unstable
<samueldr>
superloach: otherwise you *may* have more builds than the minimum to do
<superloach>
I'm aiming for stable
<samueldr>
superloach: the nixos-* branches are updated when the channels update
<samueldr>
superloach: additionally, using nixpkgs-* branches, or master directly, is "risky" (probably not that much) as it's not been validated to run the tested jobset, which includes the automated testing
<samueldr>
so a good `git checkout nixos-20.09` before building should get you everything up to the iso from the cache :)
<superloach>
understood
<superloach>
doing a shallow clone to save time lol
<samueldr>
yeah, Nixpkgs has... history
<superloach>
I can tell from the slow clone
nwm__ has joined #nixos
<c3hou>
Fare: ah, interesting. Haven't encountered that so far :O
<superloach>
samueldr: thanks so much for your help btw, hopefully this works
Fare has quit [Ping timeout: 264 seconds]
nwm has quit [Ping timeout: 264 seconds]
<samueldr>
either it works or it's really something else
gustavderdrache has joined #nixos
<superloach>
the way I forgot to patch the module in before I started the build.. huge brain time
spacefrogg has quit [Quit: Gone.]
aw has quit [Quit: Quitting.]
ManiacOfMadness has quit [Ping timeout: 265 seconds]
aw has joined #nixos
spacefrogg has joined #nixos
<superloach>
starting the build again with the patch now
fresheyeball has joined #nixos
nwm has joined #nixos
<superloach>
samueldr: is it normal for `efi-image_eltorito` to take a long time?
<superloach>
oh nvm it just finished
<superloach>
speak of the devil
<samueldr>
yes it is
<superloach>
aaand build is done, wow
<samueldr>
some of the steps to build the iso are long, but the good thing is it's a single command to do them all
<superloach>
yeah definitely, super cool
<samueldr>
and relatively easy to customize, when you know where to look
<superloach>
time to write the iso and test it, fingers crossed
<superloach>
samueldr: moment of truth, rebooting now
<samueldr>
you're on another device, right?
<superloach>
yeah I'm on irc on my vps, over ssh, from my windows laptop :P
<samueldr>
right
<superloach>
worked right away!!
<samueldr>
alright
<samueldr>
superloach: want to make the PR?
<samueldr>
(against master, which we can backport afterwards)
<superloach>
yeah sure! I'll credit you for the help, of course
<samueldr>
ping me (same username on github, @samueldr) and I'll review/approve if it fits contributions guidelines
<superloach>
awesome.
srk has quit [Remote host closed the connection]
srk has joined #nixos
<colemickens>
samueldr++ superloach++
<{^_^}>
samueldr's karma got increased to 312
<{^_^}>
superloach's karma got increased to 1
<colemickens>
future_helped_users++
<superloach>
:D
<samueldr>
superloach: btw, CYAN really was what made me 90% confident of the answer
<samueldr>
I don't have that chromebook, but I know some about the idiosyncracies
<samueldr>
if you were able to get to the initrd from SD, then the hard part was done already
<samueldr>
many windows-based laptops can't, at all, boot from SD
<superloach>
ahh nice
<superloach>
I feel like I've seen your name around in chromebook circles
<samueldr>
unlikely unless you've been in galliumos' IRC in that time I had an akali360 getting some help from mrcrhomebox
<superloach>
ah, I don't think I have
<superloach>
btw, does nix have any support for using the gallium kernel stuff?
<superloach>
CYAN can't get sound without their patches
K0kada has quit [Ping timeout: 265 seconds]
<samueldr>
not really, but you can apply patches to the kernel, or have your own kernel build
<samueldr>
not as trivial as that previous change though
<superloach>
right, I'm sure
kapil_ has joined #nixos
<samueldr>
(and there's many ways to go about solving that)
<superloach>
I'd assume it's possible to make a personal nix repo? if so, I may look into packaging their kernel packages
<superloach>
*patches
<{^_^}>
[nixpkgs] @superloach opened pull request #110435 → nixos/modules: add "sdhci_pci" to availableKernelModules → https://git.io/JtOM4
<superloach>
hopefully that pr is alright, I'm not really familiar enough with the nix ecosystem yet
<c3hou>
n
<c3hou>
Sorry, typed in the wrong window!
<superloach>
haha no worries
<samueldr>
looks about right, letting the automatic stuff do its work
<superloach>
awesome
<superloach>
I'll carry on with the installation process then
<samueldr>
superloach: one thing to do better would have been to prefix the commit message with all-hardware.nix, but looking at that file's history really you're not an outlier
zebrag has quit [Read error: Connection reset by peer]
rickard has joined #nixos
zebrag has joined #nixos
<mjlbach>
Is there an equivalent to installing an output path with nix-env with nix profile? `nix profile install` seems to only work with flakes/derivations
<mjlbach>
ex: `nix profile install /nix/store/ylbmsxkayniklphvm7ba463ynh9r1589-home-manager-path` won
<mjlbach>
*won't work
<bqv>
adisbladis: something's wrong
<bqv>
> pkgs.emacsPgtkGcc.configureFlags
<{^_^}>
attribute 'emacsPgtkGcc' missing, at (string):471:1
<colemickens>
I'm surprised we don't have gallium-patched kernels in nixpkgs.
kapil_ has quit [Ping timeout: 240 seconds]
<superloach>
yeah tbh
<bqv>
i've been after that a while too
<superloach>
ah heck, is that something I'd need the gallium kernel for
<superloach>
I recall there being an issue with gallium 2 on the uefi, so it probably is
<bqv>
currently my chromebook just runs void-linux, with a gallium kernel, and nix for most packages
<bqv>
it's a horrible mixture
<superloach>
arch kernel seems to have what it needs tho
<bqv>
but then you have to be an arch user
<superloach>
I was previously on arch with the gallium kernel, very janky
<superloach>
I'm just saying as far as deducing what I need lol
<bqv>
i'd have loved to have used guixsd on the laptop, but there's even less chance their kernel works
<superloach>
never heard of that
kapil_ has joined #nixos
<superloach>
very smart of me to switch to nixos before I even know nix
<superloach>
I would like to package the gallium kernel, once I know how
<DigitalKiwi>
i can't figure out how to get my sd card reader to work in my hp pavillion 14 chromebook :(
<DigitalKiwi>
it used to...
<superloach>
perhaps it's that sdhci_pci module :P
<superloach>
that was missing from the live initrd, and prevented me from booting on CYAN
kapilp has joined #nixos
kapil_ has quit [Read error: Connection reset by peer]
gustavderdrache has quit [Quit: Leaving.]
<colemickens>
I can't think of why your KB wouldn't work if it works with stock Arch
<superloach>
yeah, not sure why
<colemickens>
maybe worth diffing the kernel config from arch/nixos?
<superloach>
yeah probably
<superloach>
honestly would rather just apply all the gallium patches on nix
<superloach>
currently trying to determine how that works :P
<colemickens>
galliumos/linux has a v5.2 as most recent tag, but has a v5.9.8-galliumos branch ?
<superloach>
oh does it?
<superloach>
that's great news
<superloach>
last I heard, they were still on 4.x
<superloach>
I'd probably want to stick with the tagged version
<{^_^}>
[nixpkgs] @starcraft66 opened pull request #110439 → Updated MultiMC to the latest development build and fix library error → https://git.io/JtOH5
<holymonson>
supersandro2000 hi, how can I get #109368 and #108590 reviewed? darwin members seem busy, but I can't push further on updating darwin until both get merged.
<bqv>
but i keep vivaldi around for stuff that needs DRM or other dumb stuff
<bqv>
so yeah, if brave's better than vivaldi, i'll take it
<Nickli>
been using Waterfox for many years and have it setup for what i want
<bqv>
i was happy with *firefox* let alone waterfox, but mozilla's quite ametaphorically shat the bed in the last few months, so that's not gonna happen
<{^_^}>
atlas-engineer/nyxt#32 (by schvabodka-man, 3 years ago, open): Adblock/Umatrix like functionality?
<bqv>
oh right, you want a full on umatrix-like extension
<bqv>
this is not the browser for you, then
<bqv>
for everyone else, blocker-mode does fine
<mjlbach>
I mean, I've used nyxt
<mjlbach>
I like it, the keyboard driven nature is great
<mjlbach>
(I use adblock ~= extension not ad block as in the concept of blocking ads )
zupo has joined #nixos
<bqv>
i usually take it to mean the concept, since i don't think there even is an extension called "adblock" in any browser anymore at this point
<Nickli>
"adblock" the plugin sold out anyway
<mjlbach>
I mean, yeah the two extensions named in the issue are Adblock and later Ublock (the latter of which being more popular nowadays). Seems like they are close to webextension support which will be great
<mjlbach>
bqv: Not to harp on this, but I didn't like your tone. I actually *did* just perform a search on google, and the open issue I linked to is, in fact, the first result.
<mjlbach>
And the rest of the links are not pertinent.
regnat[m]1 has quit [*.net *.split]
SethDoty[m] has quit [*.net *.split]
Tek[m] has quit [*.net *.split]
anzbev[m] has quit [*.net *.split]
jvytee[m] has quit [*.net *.split]
enblanco[m] has quit [*.net *.split]
scentofiron[m] has quit [*.net *.split]
BjrnForsman[m] has quit [*.net *.split]
coolneng[m] has quit [*.net *.split]
Sylvain[m] has quit [*.net *.split]
felschr[m] has quit [*.net *.split]
OJ[m] has quit [*.net *.split]
matrix61[m] has quit [*.net *.split]
dumuzid[m] has quit [*.net *.split]
themalik[m] has quit [*.net *.split]
agentofuser has quit [*.net *.split]
anderscs[m] has quit [*.net *.split]
Hasnep[m] has quit [*.net *.split]
evhan[m] has quit [*.net *.split]
linus0[m] has quit [*.net *.split]
xd-inc[m] has quit [*.net *.split]
mullein[m] has quit [*.net *.split]
critbase[m] has quit [*.net *.split]
hexa- has quit [*.net *.split]
littlebenlittle[ has quit [*.net *.split]
viq[m]1 has quit [*.net *.split]
ptival[m] has quit [*.net *.split]
Schicko has quit [*.net *.split]
danmatrix[m] has quit [*.net *.split]
hazel[m] has quit [*.net *.split]
danielrf[m] has quit [*.net *.split]
jlle[m] has quit [*.net *.split]
nicolas[m] has quit [*.net *.split]
yangm has quit [*.net *.split]
crazazy[m] has quit [*.net *.split]
srid has quit [*.net *.split]
leibniz[m] has quit [*.net *.split]
M86ul[m] has quit [*.net *.split]
lambdaclan has quit [*.net *.split]
Minijackson has quit [*.net *.split]
Ericson2314 has quit [*.net *.split]
eadwu has quit [*.net *.split]
domenkozar[m] has quit [*.net *.split]
nasirhm has quit [*.net *.split]
cYmen has quit [*.net *.split]
eddyb[legacy] has quit [*.net *.split]
benschza has quit [*.net *.split]
drvirgilio has quit [*.net *.split]
zhaofeng has quit [*.net *.split]
numkem has quit [*.net *.split]
joepie91 has quit [*.net *.split]
astronavt has quit [*.net *.split]
zgrep has quit [*.net *.split]
fryguy has quit [*.net *.split]
Khetzal has quit [*.net *.split]
<bqv>
mjlbach: that's the problem i was trying to highlight. people searching, seeing "nyxt has no adblock extension" and completely ignoring the fact that blocking ads was actually the first and is the most developed feature
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 3 commits to master: https://git.io/Jt3fA
<supersandro2000>
holymonson: can't tell you. I fixed nixpkgs-reviews for darwin stdenv yesterday so I maybe can test them but I am not sure if they are correct and go into the right direction.
nvmr has joined #nixos
sigmundv_ has joined #nixos
<raboof>
does `nix-env --query --available --out-path` work for you on nixos-unstable? it reports about missing a name on `autotrace-0.31.1` for me, which I don't quite get
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed commit from @Scriptkiddi to master « nixos/bitlbee: Add types to options (#110446) »: https://git.io/Jt3ks
jybs has quit [Ping timeout: 240 seconds]
jybs- has quit [Ping timeout: 240 seconds]
thc202 has joined #nixos
addcninblue has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @SuperSandro2000 merged pull request #110448 → nixos/crashdump: Add type to option → https://git.io/Jt3ea
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed commit from @Scriptkiddi to master « nixos/crashdump: Add type to option (#110448) »: https://git.io/Jt3kg
<{^_^}>
[nixpkgs] @SuperSandro2000 merged pull request #110450 → nixos/activemq: add type to configurationDir → https://git.io/Jt3vt
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed commit from @Scriptkiddi to master « activemq: add type to configurationDir (#110450) »: https://git.io/Jt3k2
ahmedelgabri has quit [Ping timeout: 264 seconds]
<tad-lispy[m]>
Hi! Back to web development with Nix. I'm trying to debug my nix build process. Is there a way to run an interactive shell that is exactly like the build environment? I know about `nix-shell --pure`, but it doesn't prevent network access and leaves some crucial variables (like `HOME`).
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed commit from @Scriptkiddi to master « nixos/module/jboss: Add types to options (#110451) »: https://git.io/Jt3kw
__monty__ has joined #nixos
sangoma has joined #nixos
titouan7 has quit [Quit: titouan7]
lsix1 has joined #nixos
titouan7 has joined #nixos
addcninblue has joined #nixos
winny has quit [Ping timeout: 268 seconds]
logand has joined #nixos
attila_lendvai has joined #nixos
attila_lendvai has joined #nixos
<attila_lendvai>
is there a protocol for calling attention to a PR that is (hopefully) ready for merging? it's the Ethereum Swarm Bee client: https://github.com/NixOS/nixpkgs/pull/109393
<hyperfekt>
anyone have an idea why a service created with systemd.services works fine on a normal nixos installation but has an almost empty unit file when added to an installation medium?
ShaRose has quit [Quit: I appear to have left for some reason.]
<supersandro2000>
attila_lendvai: at least if you do not have such big PRs. discourse has also a thread for such things
domogled has quit [Quit: domogled]
domogled has joined #nixos
ShaRose has joined #nixos
<attila_lendvai>
supersandro2000, also, shall i squash most of the commits in PR's? i heard that adding contributors should be a separate commit. but what about the nitty details of finding the right setup? it would make sense to me, but the contribution doc doesn't mentiont this.
ahmedelgabri has quit [Ping timeout: 240 seconds]
<supersandro2000>
attila_lendvai: yes please. One commit for the change in the maintainers.nix, each package and the module
winny has joined #nixos
addcninblue has quit [Ping timeout: 260 seconds]
addcninblue has joined #nixos
<{^_^}>
[nixpkgs] @andyrichardson opened pull request #110458 → vscode-extensions.jpoissonnier.vscode-styled-components: Init at 1.4.1 → https://git.io/Jt3Ib
<{^_^}>
[nixpkgs] @B4dM4n opened pull request #110459 → gst_all_1: use newScope to create package set → https://git.io/Jt3Le
<{^_^}>
[nixpkgs] @andyrichardson opened pull request #110460 → vscode-extensions.eamodio.gitlens: Init at 11.1.3 → https://git.io/Jt3LU
<{^_^}>
[nixpkgs] @andyrichardson opened pull request #110461 → vscode-extensions.ms-vsliveshare.vsliveshare-pack: Init at 0.4.0 → https://git.io/Jt3Lw
<tmeusburger>
hello, I'm trying into a bizarre problem when using poetry2nix. when I try to build the environment I get this error message "Could not find a version that satisfies the requirement importlib-metadata<4,>=0.23; python_version == "3.7" (from argcomplete==1.12.2) (from versions: none)". what I don't understand is why it tries to do dependency resolution at all, all the requirements are already specified
<tmeusburger>
in the poetry.lock file. does anybody know how to deal with this?
<tmeusburger>
*trying -> running
<adisbladis>
tmeusburger: I have a hunch. Is this code public somewhere?
<adisbladis>
The root cause is packages using setuptools-scm, and there is no .git or such in the Nix sandbox so the versions can't be inferred in the python build process
Izorkin has joined #nixos
ahmedelgabri has joined #nixos
<adisbladis>
I want to inject some hook that makes setuptools-scm aware of versions from Nix, but haven't gotten around to it yet
<energizer>
idk if you have this already but often (on an older version of poetry2nix) i had to add foo-bar = self.foo_bar or vice versa
<adisbladis>
energizer: That's been fixed a long time ago :)
<energizer>
nice adisbladis++
<{^_^}>
adisbladis's karma got increased to 128
<tmeusburger>
adisbladis: sadly the code is internal to the company that I'm working at. I'll take a look at the overrides, thanks for your quick response!
<{^_^}>
[nixpkgs] @primeos pushed to master « lprof: Mark as broken »: https://git.io/Jt3tI
<tmeusburger>
adisbladis: I have tried to create an override but it didn't seem to do anything. https://pastebin.com/9SH86HqX contains the relevant code, is this correct or did I make a mistake?
<adisbladis>
tmeusburger: The general structure looks about right
<adisbladis>
tmeusburger: I suggest looking at the argcomplete sources to see if the format is exactly as you've specified or not
<{^_^}>
[nixpkgs] @SuperSandro2000 pushed 2 commits to master: https://git.io/Jt3Ov
<DigitalKiwi>
people always complain that haskell gives bad error messages but like have they never read python error messages or had 30 pages of c++ template errors ;_;
berberman has quit [Ping timeout: 265 seconds]
berberman_ has joined #nixos
pakettiale has joined #nixos
<supersandro2000>
DigitalKiwi: did you try putting a tab in 200 lines yaml file?
m4tsa_ has joined #nixos
<DigitalKiwi>
all programming languages error messages suck but at least some of them won't make it off of your machine :(
<c3hou>
If my NixOS config is done with flakes (and so its nixpkgs follows flake.lock rather than a channel), is there an idiomatic way to get channel-using commands like nix-shell and nix-env to use the same commit as the flake?
<superloach>
mjlbach: you can do "${pkgs.xyz}" for the store path i think
<NobbZ[m]>
c3hou set `NIX_PATH` through your config
<NobbZ[m]>
It is not perfect as old shells might lag behind.
<c3hou>
Nobbz[m]: thanks, I'll try that out
tmeusburger has quit [Quit: leaving]
<{^_^}>
[nixpkgs] @AndersonTorres merged pull request #108578 → nixos/chrony: add support for Network Time Security (NTS) authentication → https://git.io/JLNpm
<{^_^}>
[nixpkgs] @AndersonTorres pushed 5 commits to master: https://git.io/Jt3sf
<{^_^}>
[nixos-homepage] @github-actions[bot] pushed commit from GitHub Actions to master « Update flake.lock and blogs.xml [ci skip] »: https://git.io/Jt3sm
<djanatyn>
i use `git pull --rebase -Xours` to sync my nixpkgs fork with upstream (while retaining commits i've placed on top). i use the `ours` merge strategy so that when my pull requests are merged, my fork's commits become empty: https://paste.sr.ht/~djanatyn/e37d087c461f82e128e91029a03b8a4c4037eb1f
<djanatyn>
i still get stopped by git in the middle of a rebase, though, shown in my paste. is there a way to tell git to drop those commits when using this merge strategy, instead of creating empty ones?
pakettiale has quit [Quit: Ping timeout (120 seconds)]
<superloach>
CMCDragonkai: naiive way would just be to do `bash build_binaries.sh` in the buildPhase - but im not too familiar with nix yet, so dont take my word on that
zebrag has joined #nixos
<superloach>
oh you can do `builder = ./build_binaries.sh` i think
<superloach>
search for buildPhase on the nixpkgs wiki CMCDragonkai
<jcleng>
hi ,i will install php-amqp,need to study Nixpkgs Manual?not found phpize file
<cw>
I am trying to install NixOS in Proxmox. I can boot the live system, ping google.com, and curl http://google.com. But any attempt to reach https fails.
<CMCDragonkai>
Well I managed to disable the cmake configure with `dontUseCmakeConfigure`
<CMCDragonkai>
but I think it now fails to compile a simple executable
<CMCDragonkai>
So I need to recover some of the configure phase back
simonpe^^ has quit [Remote host closed the connection]
<tobiasBora>
Hum, I've a very strange bug sometimes in NixOs (it's the second time this bug occurs): suddently, I can't open any graphical program (okular, gimp, firefox... See errors here http://paste.debian.net/1182229) in all Konsole that were openned before. Similarly, I can't press Ctrl-Alt-T to open a new terminal (while before it worked fine). But I can still open the application from the KDE launcher.
<tobiasBora>
And from the newly openned Konsole, I can run again okular, firefox... Any idea what's wrong? The only thing I did was starting a nix-shell -p djvulibre, doing a convertion, and the bug started. I stopped the shell, and the bug is still present. The stranger thing is that it also impact other Konsoles that are not related at all with my shell.
<lddless>
Can I ask questions about packaging here? Can’t seem to find another suitable channel, but figure I’d ask anyways.
<bbigras>
lddless: yes
<lddless>
A binary I am packaging expects `ls` to be in `/bin` to perform a check if musl is being used by calling ldd on it. I’ve asked upstream maintainers to switch it to looking in PATH for ls, but until that makes it into a stable release, is there any way I can make this work? (I’m packaging a binary.)
<lddless>
(Oh I said I’m packaging a binary twice.. my bad)
<sterni>
lddless: You could either patch the binary which is probably non trivial in this case bcause the path to ls in the nix store is longer than "/nix/store" so you'd have to reverse engineer it quite a bit I guess, a simpler version is probably to add a wrapper script which starts the binary in some kind of filesystem namespace where /bin/ls points to the right ls version similar to what systemd does
<pinpox>
Mic92: oh wait.. what, it succeeded after trying again?
zupo has joined #nixos
<pinpox>
I think it .. fixed itself?
<lddless>
Is there any example of the latter, sterni?
<pinpox>
Definitely ran than more than 2 times
<lddless>
(Preferably simpler than systemd.. if I understand you correctly.)
<sterni>
lddless: I don't think it's easy to do with systemd, but it has a similar feature for services
<{^_^}>
[nixpkgs] @dhess opened pull request #110493 → nix: add Security to build inputs. → https://git.io/Jt344
<sterni>
lddless: I don't know anything of the top of my head, sorry, but I think some sandboxing tool for linux should be able to do it
<sterni>
maybe someone here knows
<lddless>
now I wish I had my bouncer up heh
waleee-cl has joined #nixos
<lddless>
I’ll idle for as long as I can I suppose
Synthetica has joined #nixos
K0kada has joined #nixos
ebopp has quit [Ping timeout: 256 seconds]
jonringer has quit [Remote host closed the connection]
<V>
like both buildInputs and nativeBuildInputs need to be built before a package is built, it's just that the nativeBuildInputs will be built for the host doing the build, and buildInputs will be built for the target
<{^_^}>
[nixpkgs] @jonringer pushed commit from @r-ryantm to master « python37Packages.mautrix: 0.8.6 -> 0.8.9 »: https://git.io/Jt30Q
<V>
and then the runtime dependencies are calculated from what in buildInputs is still referenced in the final package
<pinpox>
Mic92: etu Sorry, still no luck. After adding a bit of configuration the error returned. I've posted the error and the minimal configuration I used here: https://github.com/NixOS/nixpkgs/pull/108602
<rb2k>
anyone got a second for a quick review on https://github.com/NixOS/nixpkgs/pull/110502 ? :) Should be pretty straight forward, just adding an previously undeclared dependency.
<leo60228>
pypi2nix is marked as abandoned, has tons of open issues, and hasn't had a commit since May
<leo60228>
however, it's still very useful, and i'm not aware of a successor to it
<leo60228>
does anyone know of one?
<cole-h>
Unfamiliar with pypi2nix, but I've seen lots of python-Nix using poetry2nix (or poetry-nix or whatever it's called)
madmaurice has joined #nixos
<leo60228>
oh poetry2nix looks very nice
<leo60228>
from what i can tell if i want to package an existing project with it i should create a pyproject.toml, generate a poetry.lock, and then i can use mkPoetryApplication with projectDir as upstream source, pyproject as my translated pyproject.toml, and poetrylock as my generated poetry.lock?
sangoma has quit [Ping timeout: 246 seconds]
<leo60228>
hm poetry reminded me of naersk and more generally incrementality in *2nix tools
<leo60228>
tools like carnix and it appears poetry2nix have a 1:1 mapping of packages to derivations, which has perfect incrementality but adds a lot of overhead to large builds
addcninblue has quit [Ping timeout: 260 seconds]
<leo60228>
things like nixpkgs' rust infrastructure have an n:1 mapping of packages to derivations, which has no overhead but has no incrementality
<leo60228>
as a middle ground, naersk uses an n:2 mapping, where the root package being translated has a derivation and all other packages share a derivation
<leo60228>
this is definitely better than what nixpkgs' rust infrastructure does, but it's still not perfect
<leo60228>
i don't know what the right solution here is, though
<leo60228>
completely spitballing here, but one solution could be to have one derivation for all packages that appear multiple times in the dependency graph, a derivation each for all other direct dependencies of the root package, and then a final derivation for the root package?
<leo60228>
that adds quite a bit of complexity but it seems like it might work as a good middle ground between incrementality and overhead...
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 83 commits to staging-next: https://git.io/Jt36z
Siyo has joined #nixos
<{^_^}>
[nixpkgs] @github-actions[bot] pushed 84 commits to staging: https://git.io/Jt36g
strobelight has joined #nixos
ahmed_elgabri has quit [Ping timeout: 272 seconds]
<mahogany>
Anyone familiar with resolving issues using FODs and self references?
CodeBitCookie[m] has joined #nixos
ahmed_elgabri has joined #nixos
<CodeBitCookie[m]>
Hello
<CodeBitCookie[m]>
I am new to nixos i do know how it stores stuff and the general idea but almost all of my packages are building from source. Is this normal?
<rmcgibbo[m]>
Huh. https://github.com/NixOS/nixpkgs/pull/110517 (ammonite-2.3.8) is taking over an hour to build on my machine, but the previous evaluations on hydra only report taking ~1 second and the change to the package in the PR looks trivial and unlikely to substantially effect the build/test times.
<lordcirth__>
CodeBitCookie[m], that's not normal on x86_64, but I haven't tried i686
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « Merge pull request #109474 from NixOS/haskell-updates »: https://git.io/Jt3Xe
lsix1 has quit [Ping timeout: 260 seconds]
<mica[m]>
CodeBitCookie: look at your config and make sure you haven't set to build from source.
<cole-h>
Inspect your overlays (if any).
<CodeBitCookie[m]>
mica how do i see that
<cole-h>
If you overlay something that loads of things depend on, you'll have to rebuild everything (that you use) that depends on it.
<cole-h>
If you don't know what that means, chances are you aren't using overlays. It'd make it easier if you post a repo with your config or something.
emmanuel_erc has joined #nixos
<emmanuel_erc>
So what is the directory "/homeless-shelter" about?
<CodeBitCookie[m]>
cole-h is pastebin okay
<emmanuel_erc>
I'm trying to create a directory in a docker image that I'm trying to make.
<cole-h>
If it's a single file, yes. Otherwise it'll become a pain.
<cole-h>
emmanuel_erc: That's what HOME is set to in Nix builds.
<emmanuel_erc>
I see
<cole-h>
*in the Nix sandbox, which Nix builds use
<emmanuel_erc>
Ah, that makes sense.
<emmanuel_erc>
Which means that in my case, I don't want to think of the $HOME variable as that in my actual build
<V>
it's to catch builds that try to mess with the home dir
<emmanuel_erc>
In the sense, that $HOME won't be that while the application is running.
<V>
it really shouldn't be persisting that into the actual output
<lordcirth__>
emmanuel_erc, builds should not rely on anything outside their .nix file(s). Anything that resolves differently in different places/times, like $HOME, is out.
<V>
the application should just check $HOME when it's running and not save it when it's being built
<lordcirth__>
The running app can look for $HOME, though, yeah
qbit has joined #nixos
<qbit>
woo
<emmanuel_erc>
Yes, but in this application, in one of the dependencies, there are runtime dependencies whose locations are resolved by accessing environment variables (I don't understand why they have to do this)
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « all-cabal-hashes: update to Hackage at 2021-01-22T18:56:49Z »: https://git.io/Jt3XZ
<V>
overengineering :)
<emmanuel_erc>
I have to prepackage the expected files during the nix build.
<{^_^}>
[nixpkgs] @peti pushed to master « all-cabal-hashes: update to Hackage at 2021-01-22T18:56:49Z »: https://git.io/Jt3XZ
<lordcirth__>
emmanuel_erc, they look up the runtime dependency paths at build time?
<qbit>
hitting an issue with the chroot and nginx not being able to make the cache directory
<emmanuel_erc>
No, they look up the runtime dependencies during runtime (as expected), but I have to make sure they are able to find them by placing them in the correct location during the build.
<lordcirth__>
qbit, you probably want to use systemd tmpfiles to make the dir
ahmed_elgabri has joined #nixos
<qbit>
ooo
<lordcirth__>
Making directories exist isn't nginx's job
asbachb has joined #nixos
<lordcirth__>
Contrary to the name, tmpfiles don't have to be temporary.
<emmanuel_erc>
Ultimately, I'm trying to resolve what $HOME is the docker image created by "pkgs.dockerTools.buildImage"
<emmanuel_erc>
s/is the docker/is in the docker/
<qbit>
sure, i am very much a systemd noob - is there a doc for it? I see the options stuff, but I don't know how to make that entry relative to the nginx dir
<lordcirth__>
emmanuel_erc, as what user?
<qbit>
(options being systemd.tmpfiles.rules )
<lordcirth__>
qbit, yeah one sec. What is the dir nginx needs to exist?
<asbachb>
sys-fs-fuse-connections.mount, sys-kernel-debug.mount anybody is knowing these services? I'm wondering wyh that kernel-debug mount is activated.
<qbit>
trying to make "cache_dir" relative to the nginx chroot
<lordcirth__>
qbit, sorry, what is doing the chroot?
<qbit>
nginx
ahmed_elgabri has quit [Ping timeout: 264 seconds]
<emmanuel_erc>
I think I see a way to resolve my problem.... One of the arguments to pkgs.dockerTools.buildImage is "runAsroot"
<lordcirth__>
qbit, can you not create it as an absolute path from outside the chroot?
<qbit>
mmm, no
<qbit>
mkdir: cannot create directory ‘/nix/store/8c49yv31w4536cdxx8n55vr5k7pqa8xi-nginx-1.18.0/http_cache’: Read-only file system
<emmanuel_erc>
Whatever $HOME resolves to within that script probably isn't going to be /homeless-shelter
sangoma has joined #nixos
<qbit>
so that won't work..
<lordcirth__>
qbit, that's in the nix store, you don't want to make it there
<lordcirth__>
(you can't anyway)
<qbit>
right
<lordcirth__>
You probably want /var/lib/ or /var/www
abathur has quit [Quit: abathur]
<{^_^}>
[nixpkgs] @bryanasdev000 closed pull request #108285 → virtualbox: Includes the necessary patches for the build in the 5.10 kernel → https://git.io/JLdGR
<{^_^}>
[nix-mode] @leungbk opened pull request #127 → nix-drv-mode.el: make regexp more selective → https://git.io/Jt3Xj
<{^_^}>
[nixpkgs] @jonringer pushed to master « Revert "lib: Clean up how linux and gcc config is specified" »: https://git.io/Jt3Q4
tilcreator has joined #nixos
zarel has quit [Ping timeout: 240 seconds]
cyphase has joined #nixos
<tilcreator>
is there a way to install and activate a kernel module on the live ISO? I want to install NixOS on bcachefs and need that module loaded to mount the fs. All I can find requires an restart (which is not possible on the ISO).
<gchristensen>
edit /etc/nixos/configuration.nix add boot.supportedFilesystems = [ "bcachefs" ]; and run sudo nixos-rebuild switch ... that might get you there
<bqv>
I love the idea of it, but I think I gotta wait until it's upstreamed so I don't get caught with my pants down again
<jumper149>
viric: luks1 actually worked smoothly af
energizer has joined #nixos
lunc has left #nixos [#nixos]
jonringer_ has joined #nixos
<tilcreator>
<gchristensen "edit /etc/nixos/configuration.ni"> Sadly didn't work, or at least mount still doesn't recognices bcachefs as a vaild fs
bdju has quit [Read error: Connection reset by peer]
pushqrdx has joined #nixos
<tilcreator>
<bqv "I love the idea of it, but I thi"> Do you know of any ETA?
<colemickens>
lovesegfault: sorry, moving it here since it's more on-topic
<colemickens>
hosts: files mymachines resolve [!UNAVAIL=return] dns myhostname
g-w1 has joined #nixos
<colemickens>
makes me think nss would try to use systemd-resolved, but maybe it does in a way that doesn't use search domains by default, despite them being configured?
bdju has joined #nixos
<colemickens>
idk, I'm annoyed that this doesn't just work.
oi has joined #nixos
<g-w1>
How would the best way to use ethernet out on my laptop that has wifi to connect to a desktop with ethernet in? I know it is possible, but cant seem to find any good guides online?
<qbit>
oh, well - i have the dir created and nginx is configured to use it, but I get:
<virus_dave>
OK, so i haven't gotten too far using `buildBazelPackage`. The bazel build step attempts to launch a bazel server, which is immediately crashing for me because it cannot bind a local socket. This is pretty fundamental to how bazel works. I've tried adding `__darwinAllowLocalNetworking = true;` to the buildBazelPackage's `buildAttrs` argument, based on some quick googling, but that doesn't seem to help. Any ideas / pointers?