<{^_^}>
[nixpkgs] @Mic92 pushed 2 commits to release-18.09: https://git.io/fAF6N
notlar has joined #nixos
<{^_^}>
[nixpkgs] @Mic92 merged pull request #47227 → php: set mysql socket path if mysql[i] or pdo_mysql support is enabled → https://git.io/fAFZM
<{^_^}>
[nixpkgs] @Mic92 pushed 2 commits to release-18.03: https://git.io/fAF6p
abueide has quit [Ping timeout: 252 seconds]
wpcarro has joined #nixos
mayhewluke has quit [Ping timeout: 245 seconds]
mayhewluke has joined #nixos
Cale has quit [Remote host closed the connection]
hakujin has joined #nixos
wpcarro has quit [Ping timeout: 240 seconds]
Cale has joined #nixos
Cale has quit [Remote host closed the connection]
Cale has joined #nixos
abueide has joined #nixos
Kelppo has quit []
<{^_^}>
[nixpkgs] @matthewbauer pushed commit from @dtzWill to release-18.09 « NIX_PATH: don't prepend $HOME-based value in session variable, set later »: https://git.io/fAFiM
Kelppo has joined #nixos
nisstyre is now known as `nisstyre
silver has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @dtzWill closed pull request #47261 → inkscape, libreoffice-still: fixup to avoid needing to use old poppler → https://git.io/fAFMV
wpcarro has quit [Ping timeout: 252 seconds]
Rusty1 has quit [Quit: Konversation terminated!]
<kandinski>
how can I fake `nix-env --update` so I'll find out which paths would just be downloaded from cache and which would require a compilation?
<shreyansh_k>
kandinski: --dry-run?
<kandinski>
shreyansh_k: thanks, exactly that
sir_guy_carleton has joined #nixos
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
hakujin has quit [Ping timeout: 240 seconds]
wpcarro has joined #nixos
wpcarro has quit [Ping timeout: 264 seconds]
<kandinski>
shreyansh_k: how can I find out *why* some derivations are rebuilt and some are downloaded? I'm running 18.03 on the main NixOS install, but also the unstable channel on my user for fresher packages. I thought unstable had every derivation built and available in the caches as a binary.
<infinisil>
kandinski: such as? There are some things that can't be cached, e.g. Unfree packages, or ones that are more efficient to be built locally
<kandinski>
infinisil: I don't think I have many unfree packages. While I pastebin the list, which ones would be more efficient to build locally?
<infinisil>
kandinski: probably all of the ones you're seeing
<shreyansh_k>
kandinski: in addition to that, foss packages are always build and cached. unfree packages may restrict 3rd party distribution.
<infinisil>
Also, stuff that depends on your personal settings (e.g. Username) can't be cached
<kandinski>
infinisil: I see
<infinisil>
Oh and I guess general NixOS derivations which are usually rather fast to build, aren't cached too
<infinisil>
Like the user environment, which depends on which packages you use
<infinisil>
Or the user units
<kandinski>
aha, that makes sense
<kandinski>
I just want to know whether I'm doing something wrong here
<kandinski>
Do you mind taking a look at this? I think I've done something to my system in such a way that I'm building the world: https://pastebin.com/VDWPqtwV
graphene has quit [Remote host closed the connection]
<kandinski>
I'm building rust and tex and cpython (twice)
graphene has joined #nixos
<kandinski>
sorry, not cpython, but fortran and numpy and openblas, yes
<kandinski>
I'm quite confused
Fare has quit [Ping timeout: 260 seconds]
<{^_^}>
[nixpkgs] @aminb opened pull request #47262 → alloy: add alloy5 and take maintainership → https://git.io/fAFyk
<shreyansh_k>
kandinski: Have you tried updating your channels? maybe your copy of channels are old and hence no longer cached.
<shreyansh_k>
I think only the latest builds are cached.
endformationage has quit [Quit: WeeChat 1.9.1]
<clever>
shreyansh_k: the binary cache is never garbage collected
<kandinski>
I did update my channel
<kandinski>
shreyansh_k, clever: could it be that derivations "remember" whether they are installed from a channel, or from a local expression in the filesytem?
<kandinski>
I've been doing a lot of experiments, and I was trying to clean up by adding unstable as nixpkgs for my only user channel, updating the channel and then nix-env -u to get the latest binaries instead of my experiments.
<clever>
nix-env doesnt remember which channel it came from
<clever>
and nix-env -u will just search all channels, and install whatever matches
<kandinski>
hmm
<clever>
what does `nix-channel --list` report, for both your user, and root
vandenoever has quit [Ping timeout: 260 seconds]
<kandinski>
I was looking at that, and at $NIX_PATH
<clever>
nix-env ignores NIX_PATH
<kandinski>
ah
<shreyansh_k>
Someone more experienced may be able to answer. In fact, like you, I'm trying to install libreoffice-fresh and it's compiling on my computer.
<shreyansh_k>
So, maybe someone could explain why all other packages are being downloaded from cache but libreoffice-fresh is being compiled.
<kandinski>
my thoughts, as a n00b, were "run the system on something stable, and have latest packages on my user"
<clever>
if you try to install something you already have, it will instead upgrade it
<clever>
and use the exact version you pointed it to
<clever>
you need to use -iA nixpkgs.foo to select from unstable
<kandinski>
unless there is good reason, it seems to me the principle of least surprise would be "give preference to the user's channels over root's channels, wouldn't you say?
<clever>
if the channel names collide, yes
<kandinski>
clever: thanks, I think I'm getting what is happening, but not why it happens.
<kandinski>
clever: but what if they don't?
<clever>
but if the names dont collide, you can select between the 2 sets
<shreyansh_k>
"nix-env -u" install the latest updated package from all the searched channels
<clever>
-i (without A) and -u will search all channels
<kandinski>
I see
<shreyansh_k>
kandinski: If should use -iA instead.
<kandinski>
so if I had added unstable with `nix-add https:...unstable nixos`, then I would be guaranteeing that every time I install or update something as the user, I'd be doing it from unstable
<clever>
kandinski: probably, but then you also get warnings about collisions on channel names
<kandinski>
clever: thanks. I am less confused now.
trcc has joined #nixos
<clever>
i just run nixos-unstable on most of my machines
iyzsong has quit [Read error: Connection reset by peer]
<kandinski>
clever: including your main work machine? I'm a bit wary of not being able to do paid work.
<shreyansh_k>
infinisil: I think figured libreoffice-fresh. Second last build failed 5 days ago and the channel was last updated 2 days ago. The latest build was success 18h ago. So, I think I should retry installing after a channel update, right?
<clever>
kandinski: rollbacks allow you to undo any upgrade, as long as the disk is not corrupt
<kandinski>
clever, it's not so much that as the laptop being stuck in compilation for days.
<etu>
kandinski: I run nixos-unstable at all my laptops, including work laptop
<kandinski>
with stable channels, most everything is available in the caches, correct?
<clever>
kandinski: hydra pre-compiles things so it shouldnt be building as much, and i just dont update that often
<clever>
kandinski: all non-small channels refuse to update until hydra has cached everything that can build
<clever>
so unstable still has 100% cache coverage
<kandinski>
clever: we just had that conversation where I found out an update made me compile a lot of stuff (like numpy and openblas etc): https://pastebin.com/VDWPqtwV
iyzsong has joined #nixos
<clever>
kandinski: do you have any overlays or overrides?
<kandinski>
clever, etu: I might get to the point where I can run unstable when my configuration.nix and my user-local package sets are more stable than they are now. I'm still learning how everything works.
mkoenig has quit [Read error: Connection reset by peer]
<kandinski>
clever: I don't know what overlays or overrides are, so I'm guessing no.
<clever>
kandinski: oh, and some of the pythonPackages sets arent build by hydra, ever
<clever>
kandinski: but nix-env -u can sometimes pick the wrong one
<kandinski>
clever: aaaah
<clever>
i once got stuck in a 12 hour build of python junk, installing the wrong variant of youtube-dl
<kandinski>
I need numpy/blas for work, and I use youtube-dl for fun, so...
<clever>
it helps to declaratively specify what packages you want, and which attr set to get them from
<clever>
which also helps with channels
<kandinski>
clever: which part of the manual is that?
<clever>
nix-env -iA nixos.mystuff, will install all of my stuff, from the nixos channel
<kandinski>
I've been running nixos for five months now, but only recently I've freed up enough time to study how to do things right.
<shreyansh_k>
clever: It is.
<clever>
the override in the gist, will create a set called newstuff
<clever>
and if you nix-env -iA nixos.newstuff, it will install (or upgrade) everything to the versions listed there, from the nixos channel
<kandinski>
hmm
mkoenig has joined #nixos
<clever>
but it wont remove things you took out of the set
<kandinski>
and if I run nix-env -iA nixpkgs.newstuff, it will do it from the nixpkgs channel?
<clever>
`nix-env -iA nixos.newstuff -r` will atomicly install(or update) everything in the set, and remove anything not in the set
<clever>
yep
<clever>
you could also make 2 sets of packages, and choose to always install set1 from nixpkgs, and set2 from nixos
<kandinski>
how do you decide what to put there and what to put in your configuration.nix?
<kandinski>
yes, I see
<clever>
i try to use nix-env for temporary things, and move them to configuration.nix later
<kandinski>
or this is also how I would maintain my own package set if I decided to go to nixops for everything
<kandinski>
our company's build artifacts, built by nix/hydra, on a separate channel
<clever>
but there are also things that break the build often, and then i move them to nix-env, because they get in the way of changing 1 line in a config file
<kandinski>
that would include any forked library if we have them
<clever>
your custom channel could even have such sets, directly in it
<kandinski>
clever: what do you mean?
<clever>
and then anybody with the channel can just nix-env -iA company.all-tools
<clever>
just make a channel, with the default.nix returning tools, and a set called all-tools
hyper_ch2 has joined #nixos
<kandinski>
and that would bring some packages from nixos infra, and some from our own infra?
<kandinski>
or would we have to build the whole of nixpkgs?
<clever>
it would get whatever packages the "all-tools" set in the "company" channel listed
<clever>
the company channel could then refer to an approved nixpkgs rev, that hydra has pre-built your tools for
<kandinski>
(I think I'm having trouble with the nix-specific jargon, sorry for that)
<kandinski>
05:58 < clever> your custom channel could even have such sets, directly in it
<clever>
you could make your own nix channel, for the company, which returns company specific tools in it
<kandinski>
what do you mean by "such sets, directly in it"? That I wouldn't need to have the equivalent of newstuff in my confi.nix, because it would already be in the channel?
<clever>
and a common set like newstuff above, could be inside that channel
<kandinski>
aha
<clever>
and then people can just `nix-channel --update ; nix-env -iA company.all-tools` to both upgrade, and get any new tools added to the list
<kandinski>
so everybody is on, say, release18.09, and they all have also the company channel enabled either as root or as their user
<kandinski>
yep, thanks
<kandinski>
or even in the configuration.nix, right?
<clever>
that would be a bit more complex, but still possible
<kandinski>
right, probably not worth it
<kandinski>
also company is using Ubuntu, I don't think I can persuade them to turn to nixos (nor should I until I know it better, it wouldn't be responsible)
<clever>
ive seen another company where the new guy tried to switch them to nixos before learning it fully himself
<clever>
and now the company is very anti-nix and doesnt like the other guy that is trying to promote nix
<kandinski>
clever: thanks a lot. I'm planning to migrate all my personal stuff to nixos, though, and this has given me a few ideas. I do fork libraries frequently to fix issues or add features, and then refer to my own forks while upstream decides on the PR. NixOS seems a better way to manage that workflow.
<kandinski>
Yeah, I don't want to be that person.
<kandinski>
I'm moving my own stuff little by little, though. My only dislike so far is the long compilation times.
<kandinski>
I reinstalled several ubuntu machines this weekend (for daughters and such) and each install takes literally 10 minutes.
<kandinski>
Longer installs are only due to slower disks.
<clever>
if you ensure every machine is on the same nixpkgs rev (update the channels at the same time, or other stuff), you can share build products between machines
<kandinski>
clever: I guess that's the next thing to learn. Set builder on my baremetal host, and have my own caches for everything.
<clever>
i also have my local hydra pre-building my nixos configs against nixos-unstable
<clever>
so i can both see if it will break without updating
<kandinski>
I see
<clever>
and i can pick a rev it has pre-built, and update to it almost immediately
<clever>
i can see that 3 days ago, something stopped building
<clever>
and then 1 day ago, it fixed itself
<kandinski>
and do you then download from the cache that you built, to avoid long compilations?
<clever>
checking the logs, it looks more like a garbage collection ate the compiler (bugs in some auto-gc stuff), and the build failed when it shouldnt have
<clever>
yeah
monotux has joined #nixos
<kandinski>
clever: can you share your setup?
<sphalerite>
,pr jackdk
<{^_^}>
jackdk: 0. Sorry. 1. If you complain about PR being in limbo, always provide a number («I have an unmerged PR (#23924)…»), bot will print PR title and it might bring you an impulse-merge. 2. Please look up who commits similar work and ping these people (or request-review), keeping up with all the PRs is not really possible.
<clever>
kandinski: hydra is configured to build everything in release.nix
<clever>
other machines not covered by hydra also exist in that repo
civodul has joined #nixos
<kandinski>
I'm updating nixos and also user-local packages, so the browser is having trouble rendering the page
<jackdk>
Thanks sphalerite. #45969 is the PR. Does request-review mention a specific command or bot-trigger or something? I can't find anything obvious on a web search
<sphalerite>
jackdk: requesting a review on a PR from somebody notifies that person. Since you're making a new package here I'm not sure finding anyone in particular is appropriate
<sphalerite>
jackdk: but your PR LGTM, I've pinged the bot to build it and probably someone like Mic92 will merge it soon :)
<kandinski>
clever: or, otherwise, who would you recommend I crib my dotfiles from?
<sphalerite>
jackdk: also re. FON bitmap fonts, it looks like it's possible but not trivial to fiddle them into something that will work on linux
<jackdk>
What would you recommend, then? Make the name indicate that its a TTF-only package or leave them unpackaged? How many people use bitmap fonts these days?
<sphalerite>
I use them, they're great for putting a lot of text in a small space.
<sphalerite>
jackdk: oh actually http://www.seasip.info/Unix/PSF/ might be helpful for this. But yeah up to you if you want to go to the effort of also packaging that ;)
brejoc has joined #nixos
<{^_^}>
[nixpkgs] @andrew-d opened pull request #47263 → xsecurelock: add extra derivation parameter and NixOS module → https://git.io/fAF9z
periklis has joined #nixos
graphene has quit [Remote host closed the connection]
graphene has joined #nixos
reinzelmann has joined #nixos
bbarker has quit [Read error: Connection reset by peer]
<{^_^}>
[systemd] @fpletz merged pull request #23 → nspawn: chown() the legacy hierarchy when it's used in a container → https://git.io/fAFVp
<{^_^}>
[systemd] @fpletz pushed commit from @evverx to nixos-v239 « nspawn: chown() the legacy hierarchy when it's used in a container »: https://git.io/fAF7Y
johanot has joined #nixos
Thra11 has joined #nixos
palo has quit [Ping timeout: 252 seconds]
asymmetric has joined #nixos
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Thra11 has quit [Ping timeout: 245 seconds]
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
mayhewluke has quit [Ping timeout: 252 seconds]
ckauhaus[afk] is now known as ckauhaus
mayhewluke has joined #nixos
<ckauhaus>
Some channel op should remove the NixConf CfP from the topic...
<ckauhaus>
I think the program has been decided upon
<sphalerite>
ckauhaus: yeah I've poked gchristensen a couple of times now, never at the right time it seems ;)
jackdk has quit [Ping timeout: 260 seconds]
<Myrl-saki>
sphalerite: ckauhaus: Closed for Papers.
<sphalerite>
haha
sir_guy_carleton has quit [Quit: WeeChat 2.0]
georges-duperon has quit [Ping timeout: 252 seconds]
<ckauhaus>
:-)
<ixxie>
I'm having problems with my Home-Manager installation....
<ixxie>
home-manager switch returns error: file 'home-manager/home-manager/home-manager.nix' was not found in the Nix search path (add it using $NIX_PATH or -I)
<srhb>
??/
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
<srhb>
srhb has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<makefu>
hopefully he didn't got k-lined
thc202 has joined #nixos
CapsLock has joined #nixos
<CapsLock>
hi everyone
<CapsLock>
just wondering how to add some file so nix-store for given package ; as an example I would like to add an extension to wireshark ; it should be stored in lib/wireshark/extcap ; how can we do that reliably ?
<johanot>
makefu: she did. :(
<etu>
Nooo, for how long? srhb tend to know things and be helpful.
trcc has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @flokli opened pull request #47264 → systemd: bump to rev 31859ddd35fc3fa82a583744caa836d356c31d7f → https://git.io/fAFF7
trcc has joined #nixos
<makefu>
johanot: dang
<symphorien>
CapsLock: once wireshark (whithout the plugin) is built its nix store path is completely immutable. You should find a way to tell wireshark "find the plugin at this location".
<CapsLock>
symphorien: thanks :)
<CapsLock>
I think I can do that also by extending actual wireshark package
<symphorien>
the usual way is to wrap the package. Assume that wireshars looks for its plugins in the env var WIRESHARK_PLUGINS. You make a derivation which is just a shell script (see makeWrapper in the nixpkgs manual) which exports this env var pointing to the store path of your plugin, and then execs the vanilla wireshark.
<symphorien>
examples of this are firefox, python.withPackages etc.
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @flokli to staging-18.09 « systemd: update to fix nspawn containers (#47264) »: https://git.io/fAFNA
ixxie has joined #nixos
ThatDocsLady has joined #nixos
fendor has joined #nixos
johanot has quit [Ping timeout: 264 seconds]
johanot has joined #nixos
jophish has quit [Quit: WeeChat 2.0]
<betaboon>
are environment-variables (like PYTHONPATH) passed to systemd preStart ?
cryptix has joined #nixos
<sphalerite>
betaboon: those defined for the service?
<betaboon>
sphalerite: yes
<betaboon>
just as context: i have a nixos-service-module, which wraps a python-service, which has a prestart script which calls a setup.py-entrypoint -> on 17.09 it worked fine, now on 18.09 it complains that there is no such module when running the prestart script
<cryptix>
hey - what's the easiest way to deploy a shell script? i tried pks.writeScriptBin and added the output to environment.systemPackages. the output is there but it's missing the executable flag
<sphalerite>
cryptix: that's very odd, writeScriptBin should set it as executable
<{^_^}>
[nixos-org-configurations] @edolstra pushed 2 commits to master: https://git.io/fAFjh
<cryptix>
sphalerite: hrm.. i'll paste it somewhere.. maybe i shouldn't put it in a let? seemed cleaner though.. https://pastebin.com/raw/Rq7giEdm
carlosdagos has quit [Quit: Connection closed for inactivity]
<betaboon>
sphalerite: to answer my own question: yes the environment-variables are available within preStart.
<cryptix>
sphalerite: sorry, nvm. the error is actually '... is marked as an executable but could not be run by the operating system'
<cryptix>
i guess it's the scripts content - will debug further
Anton-Latukha has joined #nixos
<sphalerite>
cryptix: you probably just need a shebang
<sphalerite>
cryptix: change it to writeShellScriptBin, or put #!${pkgs.stdenv.shell} at the top of the script
<exarkun1>
Using nixops, I can define a machine as a function accepting `lib`, `resources`, `config`, `pkgs`, etc. where do I read about these and other parameters?
cryptix has quit [Remote host closed the connection]
<azazel>
exarkun1: just in nixops code
<exarkun1>
:`(
<exarkun1>
Okay, thank you.
<betaboon>
azazel exarkun1: lib, config and pkgs behave just as in nixos AFAIK. and resources are the resources that you define. what else would you need to read about them ?
<azazel>
it wuld be useful to know if there are others,
<azazel>
for example
<exarkun1>
betaboon: It would be nice to be able to read what you just said in the documentation. Then I'd know I could go look at the nixos docs for those things. Also, "resources are the resources that you define" isn't quite enough for me to get a handle on that concept. Are you saying that's a purely user-defined value? Where does it come from? Anywhere else in the nix for my deployment?
<exarkun1>
betaboon: Also, what azazel said.
<betaboon>
not saying that the nixops-documentation is perfect. It's far from that...
ixxie has joined #nixos
<exarkun1>
I appreciate the info. I don't have other specific questions right now, just trying to improve my self-education loop.
<azazel>
exarkun1, betaboon also the documentation about the vpn: when/how it's setup, and the overall knowledge of the config options to "deal" with it aren't so clear to me,exarkun: I learned most of the missing thing from the examples/ directory inside the repository
<azazel>
but be aware that some examples aren't up to date
<betaboon>
azazel: yeah i had to do alot of digging in the code etc as well
<{^_^}>
[nix] @edolstra merged pull request #2312 → search: include version for matches too! → https://git.io/fNz4k
<typetetris>
Hi there! I'm a little confused about the 'sandbox' option of nix. In the manual of nix it is claimed, that you need to run nix as root, to use 'sandbox=true'. Is it sufficient if nix-daemon runs as root or do I need to call nix-build as root?
<sphalerite>
typetetris: the former
<typetetris>
thanks.
<sphalerite>
typetetris: running nix-build as root does *also* allow sandboxing, even without a daemon. But with the daemon running as root, regular users can run sandboxed builds too.
<typetetris>
I will probably set it to relaxed, so the users can do whatever they want.
<typetetris>
How can I get the transitive closure of all the store paths needed to build a certain derivation? (I assume: Step 1) Get all the direct build dependencies of the derivation (how to do that)? Step2) Get the transitive closure of all the run-time dependencies of those via nix-store -qR ?)
<{^_^}>
b1000101: 0. Sorry. 1. If you complain about PR being in limbo, always provide a number («I have an unmerged PR (#23924)…»), bot will print PR title and it might bring you an impulse-merge. 2. Please look up who commits similar work and ping these people (or request-review), keeping up with all the PRs is not really possible.
<etu>
b1000101: I highly suggest you to rebase your commits to squash/fixup them to only be like two or three, one for adding yourself to the maintainer list and one for refactoring/updating the package and one for updating the module
<sphalerite>
etu: I think the commits there are sensible
<sphalerite>
oh wait no
<etu>
sphalerite: "Missed semicolon"
<sphalerite>
not the "missed semicolon" and stuff I guess
<etu>
Most are good, some can be squashed
<sphalerite>
b1000101: you may also want to ping the existing maintainers canndrew and jb55 in the github issue
<makefu>
gchristensen: thats too bad, there were times (like 1 month ago) when nothing would have happened ...
* etu
is sure that srhb finds her way back here:)
abueide has quit [Ping timeout: 260 seconds]
<b1000101>
sphalerite yeap, that's a totally stupid mistake, I know - is there a way to squash them remotely? Sry for the stupid question, I still have to learn git properly
abueide has joined #nixos
srhb has joined #nixos
<sphalerite>
srhb: welcome back!
<sphalerite>
b1000101: not a stupid mistake at all :) you'll need to do this locally AFAIK
<srhb>
Thanks!!
<sphalerite>
the only way I know of squashing is git rebase -i
<makefu>
sphalerite: or cherry-picking
<sphalerite>
makefu: I don't know that one ;)
patrl has joined #nixos
<sphalerite>
makefu: huh, the manpage for git-cherry-pick doesn't contain the string "squash" anywhere either
<makefu>
sphalerite: ha, yeah, sorry - git rebase -i is the easiest way to do
<sphalerite>
makefu: I'm curious though, how does it work with cherry-pick?
<srhb>
sphalerite: Probably using --no-commit :)
<sphalerite>
aaaah right
b1000101 has quit [Ping timeout: 256 seconds]
<sphalerite>
b1000101 really doesn't have a very reliable connection :p
b1000101 has joined #nixos
<b1000101>
sry I got disconnected, has anybody replied, I didn't get any messages
<sphalerite>
nix-diff is great. Just did a big refactor of one of my systems and I could make sure that only the things I wanted changed did actually change :D
<srhb>
Yep, it's lovely!
<b1000101>
my connection is alright, but I had to go to DC :)
<srhb>
b1000101: If we're talking about the trezord PR things have gotten a bit confusing because of the backport PRs fragmenting the conversation. I think it would be helpful if we could stick to the master one and open backport PRs once that's in. I'm perfectly happy merging the (squashed) version to master, but I honestly still don't understand what the changes do re. 18.03, and if it might break anything for
<srhb>
existing users. :)
Acou_Bass has quit [Ping timeout: 252 seconds]
<srhb>
Wow, OK, my multiline protection really IS broken. That explains a few things...
Acou_Bass has joined #nixos
<sphalerite>
splitting one message into two doesn't usually trigger multiline protection AFAIK
<sphalerite>
at least not weechat's stock one
<sphalerite>
what actually happened earlier? -> #nixos-chat ;)
<srhb>
I thought I configured it to do so way back, but maybe I'm mistaken...
ixxie has joined #nixos
<etu>
srhb: Welcome back:)
<srhb>
etu: Thanks :)
devx has quit [Ping timeout: 246 seconds]
<b1000101>
srhb - yeap, alright. Basically a new version of our "bridge" (trezord) came out that needs udev rules to be a modified in order to work with the hardware. That's all there's to it. The previous bridge (2.0.12 and 2.0.14) is not really working with one type of the hardware anymore. So it should not break anything, rather the opposite as one hardware device is currently non-functional in NixOS.
<makefu>
srhb: everyone was worried that you got k-lined for the day
<srhb>
Hehe, kind of you all. :)
<srhb>
b1000101: OK, that is a very helpful explanation. :)
<srhb>
b1000101: I'll take a look tonight, if you squash it all down before then that would be helpful. :)
graphene has quit [Remote host closed the connection]
<betaboon>
can i somehow enforce rebuild of all dependencies already in nix-store when using nix-shell etc ?
hotfuzz has joined #nixos
hotfuzz_ has quit [Ping timeout: 240 seconds]
<gchristensen>
srhb: welcome back!
ericsagnes has quit [Ping timeout: 240 seconds]
<Taneb>
haskell.packages.ghc843.cabal-install doesn't seem to build for me on nixpkgs 18.03
<srhb>
gchristensen: Thank you, and thanks for poking freenode. :)
<gchristensen>
glad to help -- +o's in here can unkline anyone klined here as long as they do it fast enough, otherwise you have to +o and ask for help :)
<Taneb>
Because hackage-security fails, overriding that to a later version causes cabal-install to fail with missing dependencies
<gchristensen>
makefu: unfortunately spam degrades the commons :(
<makefu>
just like every time something bad happens to an otherwise open community ...
<gchristensen>
can I specify a partition ID to fdisk?
<gchristensen>
like, create a partition with uuid X
<alex``>
emily: You use a display manager or tty > startx?
<srhb>
Taneb: Making 8.4.3 fully functional on 18.03 might be a bit tricky. Do you really need to?
<srhb>
Taneb: (As in, anything that isn't haskellPackages for a particular release might take a lot of work)
<LnL>
gchristensen: watcha doin? :p
<Myrl-saki>
srhb: How was uh... lie on the other side
<gchristensen>
LnL: my questions are usually fun, huh?
<srhb>
Myrl-saki: :|
<Myrl-saki>
srhb: :P
grp has joined #nixos
<alex``>
emily: occivink tells me a nice thing to do; keep default shell as login shell, then in the TTY, manually elvish if I want it, then set the default program for the terminal, Alacritty to open Elvish
<LnL>
gchristensen: I think that's based on the disk serial, is this is a vm?
<gchristensen>
it is an ARM system which starts up with 0 entropy. though d'ezgeg had a suggestion which I forgot about, and am trying now
<Taneb>
srhb: I'd rather not wait until 18.09 is out for doing this
<Taneb>
Even if that's only going to be a week or so now
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @Ma27 to release-18.09 « python3Packages.py3exiv2: fix build »: https://git.io/fAbs2
Rusty1 has joined #nixos
<srhb>
Taneb: Then I suggest using 18.09 for it today. :)
<srhb>
Taneb: Or using an earlier GHC and 18.03
silver has joined #nixos
<srhb>
Taneb: And if neither of those work, there's nothing to do but continue following the errors and doing overrides as you already were. :) Might go quite deep though.
<sphalerite>
gchristensen: filesystem uuid or partition uuid?
<Taneb>
I'm up to my neck in overrides already...
<srhb>
Taneb: Yup, well, that's to be expected..
<gchristensen>
dunno but I think I sorted it somehow??
<srhb>
Taneb: Every set that isn't haskellPackages is really just haskellPackages projected onto another version of ghc.
<{^_^}>
[nixpkgs] @Mic92 merged pull request #47148 → eiskaltdcpp: mark linux only → https://git.io/fA5Hs
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @LnL7 to release-18.09 « eiskaltdcpp: mark linux only »: https://git.io/fAbGO
ericsagnes has joined #nixos
<gchristensen>
awww it faulted again
<sphalerite>
gchristensen: in any case, tune2fs can adjust filesystem uuids for ext[234]
<sphalerite>
oh there's also the partition *table* UUID
<sphalerite>
but sfdisk can set both partition table UUID and partition UUID
<kai_w>
am I going to run into trouble if I use parallel building when testing changes to nixpkgs?
<kai_w>
i.e. cores=0, not max-jobs
<srhb>
kai_w: If you do run into problems, you should disable parallel building for the derivation :)
<sphalerite>
kai_w: typically not, because they only build with more than one core if they're explicitly marked to, and one would assume that someone who marks them as parallel-capable would have tested them in parallel :)
<kai_w>
awesome. gcc takes a while single-threaded
<sphalerite>
yes. Yes it does.
tobiasBora has joined #nixos
<tobiasBora>
Hello,
<tobiasBora>
I just got on my rasp an error "nixos GC Warning: Failed to expand heap by 238452736 bytes"
<tobiasBora>
do you have any idea why I get this error?
init_6 has quit []
<sphalerite>
running out of memory. If it doesn't crash you're still fine
<sphalerite>
but even if it does properly run out it shouldn't break disastrously
<tobiasBora>
sphalerite: it's RAM memory?
<sphalerite>
yes
<gchristensen>
,locate lspci
<{^_^}>
Found in packages: busybox, pciutils
<tobiasBora>
sphalerite: ok thank you. I'm curious to see what's using that much memory...
<tobiasBora>
Ouch... Gitlab uses 30% of the memory
<tobiasBora>
60%*
<tobiasBora>
and the swap is nearly full
Izorkin has quit [Read error: No route to host]
<sphalerite>
yeah the pi isn't exactly a heavy-duty server :')
<srhb>
And gitlab uses bizarre amounts of resources.
hakujin has joined #nixos
<andi->
gitlab on a RPi. I wouldn't have thought that is still possible :)
<typetetris>
I did set 'sandbox = true' for a multiuser nix install on a debian 8. nix-daemon is running as root. When I try to build something I get errors, because the mounting of /proc in the sandbox seems to be not permitted. What am I doing wrong?
<sphalerite>
typetetris: is it actually connecting to the daemon? Try passing --store daemon
<gchristensen>
kiloreux: no, s3 doesn't support that
<kiloreux>
Thank you gchristensen .
<wpcarro>
is there a way to get nix to evaluate the `shellAliases` after something like the `zsh.interactiveShellInit`? It seems like some of my aliases are being overwritten
<sphalerite>
typetetris: that's weird, I don't know what else to try :/
<typetetris>
sphalerite: But it's definitely the sandboxed build. I am using a vm (don't know which technology, it department just hands me a login for ssh) here, so maybe thats why.
johanot has quit [Quit: leaving]
<{^_^}>
[nixpkgs] @jtojnar pushed 134 commits to gnome-3.30: https://git.io/fAb41
<{^_^}>
[nixpkgs] @Vskilet opened pull request #47270 → roundcube: init at 1.3.7 → https://git.io/fAb4y
<d1rewolf>
is it correct to say that "environment.systemPackages = with pkgs; [ mpv ]" is the same as "environment.systemPackages = [ pkgs.mpv ]"?
<sphalerite>
d1rewolf: yes
<sphalerite>
d1rewolf: *unless* mpv has been defined by a let outside that expression
<d1rewolf>
sphalerite: k, thought so...thx very much
<sphalerite>
typetetris: hmm, maybe try virt-what to find out?
<typetetris>
sphalerite: its vmware, I don't know about virtualization technologies, is it plausible for sandboxed builds to fail on a vmware virtual machine?
iyzsong has quit [Read error: Connection reset by peer]
<sphalerite>
typetetris: I don't think so, but maybe it uses a weird kernel?
<sphalerite>
although I think if lack of namespacing supportwere the problem you'd be getting a different error
<tobiasBora>
I have an error: "Package ‘cups-gutenprint-binary-5.0.1’ in /nix/store/imar2zpr687zw5i4v0jcidpanwykilgq-nixos-19.03pre152634.218ce4de508/nixos/pkgs/misc/drivers/gutenprint/bin.nix:63 is not supported on ‘aarch64-unknown-linux-gnu’, refusing to evaluate"
<tobiasBora>
and nix tells me that I can fix it with nixpkgs.config.allowUnsupportedSystem = true;
<tobiasBora>
But it seems strange to me
<tobiasBora>
because if the app is only available for x86, I don't see the point of forcing it
<sphalerite>
I mean it's fine, just that there's a bot in here which will kick anyone who writes too much in a short period of time off the whole network :p
<devhell>
is networking.interfaces.eth0.ipv4.routes = [ { address = "10.100.0.0"; prefixLength = 24; via = "10.99.0.253"; } ]; not correct or am I missing something?
<devhell>
there is no infromation via `ip route`
<sphalerite>
devhell: does eth0 exist?
<devhell>
sphalerite: yes
<devhell>
it's a RPi3
<d1rewolf>
sphalerite: it is big. pastebin rejects it. any ideas of a pastebin that'll accept it?
<d1rewolf>
556k
<sphalerite>
d1rewolf: ooft. Maybe grep it for PATH yourself
<devhell>
there's also nothing in /etc/systemd/network/40-eth0.network other than the IP address and DHCP being off
<{^_^}>
"Whether we should use networkd as the network configuration backend or\nthe legacy script based system. Note that this option is experimental,\nenable at your own risk.\n"
<{^_^}>
[nixpkgs] @edwtjo pushed to master « pythonPackages.sharedmem: init at 0.3.5 »: https://git.io/fAbuP
<sphalerite>
d1rewolf: the line in the output mentioning terminator
<typetetris>
sphalerite: If I want to use sandboxes do the build users need special permissions?
hyper_ch2 has quit [Quit: Page closed]
<sphalerite>
nothing else there looked like it might affect your shells' environment
<sphalerite>
typetetris: no, root does the mount magic before dropping to a build user
mayhewluke has joined #nixos
<d1rewolf>
sphalerite: ok. thanks. do you still think my path looks wonky as well?
<sphalerite>
d1rewolf: yes, but it's just a result of terminator. Compare what it looks like in alacritty
<d1rewolf>
sphalerite: you're right. that's bizarre. is this worthy of filing a bug report with nix? seems like this sort of unexpected behavior could really throw people
<typetetris>
sphalerite: Can I somehow see a list of executed commands?
hotfuzz_ has joined #nixos
<sphalerite>
typetetris: it's not running shell commands at that point, but you could use strace to see the system calls
trcc has joined #nixos
ThatDocsLady_ has joined #nixos
<typetetris>
sphalerite: But I would need to trace nix-daemon I guess.
<typetetris>
sphalerite: It is a kernel bug. Something used `proc_mkdir` in the kernel instead of `proc_create_mount_point` or some such. Installing a newer kernel made it work.
<{^_^}>
[nixpkgs] @dtzWill opened pull request #47278 → e2fsprogs: patch out glibc instead of local fix so headers are usable → https://git.io/fAbw0
<wchresta>
How do I install firmware similar to copying to /lib/firmware/i915 in ubuntu?
Mateon2 has joined #nixos
Mateon1 has quit [Ping timeout: 244 seconds]
Mateon2 is now known as Mateon1
<andi->
wchresta: hardware.firmware or hardware.enableAllFirmware or hardware.enableRedistributableFirmware ..
<andi->
or various other firmware options
<wchresta>
andi-: Cool, thanks!
<{^_^}>
[nixpkgs] @romildo opened pull request #47280 → shades-of-gray-theme: init at 1.1.1 → https://git.io/fAbo2
Kelppo has quit []
<lejonet>
How do I force nix-shell to re-make a drv? I seem to have gotten a .drv file that is a zeroed file where the nix-shell derive should have been (nix-store --repair-path <path> said it couldn't fix the path)
<manveru>
rm?
wchresta has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @xeji merged pull request #47255 → boost: limit concurrent jobs to the maximum supported number → https://git.io/fAFwm
<{^_^}>
[nixpkgs] @xeji pushed commit from @andrew-d to master « boost: limit concurrent jobs to the maximum supported number (#47255) »: https://git.io/fAbKg
Ariakenom has joined #nixos
<{^_^}>
[nixpkgs] @dtzWill opened pull request #47281 → llvm7: update sanitizer patch to fix w/musl → https://git.io/fAbKV
haslersn has joined #nixos
<{^_^}>
[nixpkgs] @xeji pushed commit from @andrew-d to release-18.09 « boost: limit concurrent jobs to the maximum supported number (#47255) »: https://git.io/fAbKS
<lejonet>
manveru: that just makes it complain that it cannot find the drv file at all :P
<manveru>
heh
<manveru>
how do you run the nix-shell?
<manveru>
hm, wasn't there a reduce function in lib?
<manveru>
ah, fold
coot has joined #nixos
teej has quit [Quit: Connection closed for inactivity]
<ajs124>
I'm having the weirdest issue right now. One of my virtual machines fails to find its root device, but I know it's there. In fact, old configurations boot fine.
<{^_^}>
[nixpkgs] @edwtjo pushed commit from @r-ryantm to master « jackett: 0.10.160 -> 0.10.198 »: https://git.io/fAbd9
<worldofpeace>
sphalerite: any ideas?
<sphalerite>
nope sorry
<samueldr>
worldofpeace: `nix-build nixos/release.nix -A ova.x86_64-linux` on my end, on nixos-unstable (just pulled nixpkgs-channels) seems fine (still building though)
<{^_^}>
[nixpkgs] @andir pushed 4 commits to release-18.09: https://git.io/fAbFB
<samueldr>
though it's true, on master it fails
<worldofpeace>
hah that's not good
<samueldr>
oh, fancy, the last commit on the build-support/vm/default.nix seems relevant f777d2b719be0db4cb7819d5b6dbdd121db29a37
<samueldr>
today I tested: searching for a commit ID in the github PR search box will find the appropriate PR
orivej has joined #nixos
<worldofpeace>
reading the pr now
lopsided98 has quit [Quit: Disconnected]
_d0t has joined #nixos
<_d0t>
hi! Where can I read about installing udev rules in a nix package?
sir_guy_carleton has joined #nixos
<sphalerite>
_d0t: basically you don't
<cransom>
_d0t: thats not an option in a normal package. if you are running nixos however, you can add https://nixos.org/nixos/options.html#udev.extrarules to your module
<_d0t>
cransom, what if the rules are in a file?
lopsided98 has joined #nixos
<cransom>
you can read the file in your module, but it's still not an option in a package.
<_d0t>
though, I'm probably asking the wrong thing
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
<sphalerite>
_d0t: you can also have it symlinked into /etc/ by putting the path to the rules file in envrionment.etc."udev/rules.d/99-foo.rules".source
<cransom>
it can define them, but the system isn't installing them or looking at them in anyway.
<sphalerite>
_d0t: packages can't install that sort of thing, it's only applied by nixos
<_d0t>
sphalerite, i'm using nixos
<_d0t>
so this applies
<sphalerite>
that is, packages can put them in their store paths
<sphalerite>
but they won't get automatically picked up
<_d0t>
sooo... how do I make the rule picked up if it's in store?
<sphalerite>
oh wait actually it looks like the rules from systemPackages do get copied..?
Izorkin has joined #nixos
<_d0t>
copied to where?
<sphalerite>
/etc/udev/rules.d
<_d0t>
doesn't seem to be the case here
<_d0t>
I have 88-nuand.rules in the package's store but not in /etc/udev/rules.d
<cransom>
how would it copy them? /etc/udev/rules.d is in the nix store, woudl be unwritable.
<sphalerite>
aah ok it's not automatic from systemPackages
<sphalerite>
but various modules add rules via boot.initrd.extraUdevRulesCommands by the looks of it
<{^_^}>
"Shell commands to be executed in the builder of the\nudev-rules derivation. This can be used to add\nadditional udev rules in the initial ramdisk.\n"
<sphalerite>
aaah ok no
lord| has quit [Quit: WeeChat 2.2]
<sphalerite>
the *real* way to do it is to add the package to services.udev.packages
lord| has joined #nixos
<_d0t>
sphalerite, in configureation.nix?
<sphalerite>
_d0t: so if you put services.udev.packages = [ pkgs.libbladeRF ]; in your configuration.nix and rebuild it should work
<sphalerite>
yes, or in another module
<_d0t>
thanks
<_d0t>
how come it's not done out of the box?
<_d0t>
if a package installs udev rules, they should be picked up
<sphalerite>
to avoid packages installing god-knows-what udev rules I suppose
<sphalerite>
nixos tends to limit the scope of what a package can do to only what's explicitly specified, rather unlike traditional distros
<{^_^}>
[nixpkgs] @domenkozar pushed to master « nixos tests: move common configuration into separate file »: https://git.io/fAbNS
<_d0t>
isn't that supposed to be checked by maintainers?
<{^_^}>
[nixpkgs] @domenkozar pushed to release-18.09 « nixos tests: move common configuration into separate file »: https://git.io/fAbNN
<sphalerite>
what suits the maintainer might still not be what you want ;)
<_d0t>
probably true, but then again, i wouldn't install the package then
<_d0t>
and for the many packages rules are installed by default
<sphalerite>
what do you mean by that?
<_d0t>
so the distinction between the packages with rules installed and everything else seems to be a bit arbitrary
<sphalerite>
("for the many packages rules are installed by default")
<sphalerite>
ah
<_d0t>
I mean the /etc/udev/rules.d isn't empty
<sphalerite>
no
<sphalerite>
those are all specified explicitly by the various nixos modules that make up your system
<sphalerite>
each of them is created by a services.udev.packages setting somewhere in <nixpkgs/nixos>
astronavt has joined #nixos
<_d0t>
oh
<_d0t>
got it
<_d0t>
another thing
<_d0t>
bladerf udev rules specify that device permissions should be set for access from plugdev group, but this group is absent
<_d0t>
is this a bug?
<cransom>
packages don't define groups either, that would still be needed in your nixos configuration.
<_d0t>
cransom, but shouldn't a package specify some pre-existing group?
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « Update overrides for ghc-8.6.x. »: https://git.io/fAbAw
<cransom>
no, because, how would it?
<cransom>
a package only defines software, not the system it runs on.
<_d0t>
well, I suppose there should be a list of groups pre-defined in the system for various purposes
<PolarIntersect>
,locate pkg-config
<{^_^}>
Found in packages: pkgconfig, bashCompletion
<infinisil>
Wow, never seen that error: "acquiring/releasing lock: Resource deadlock avoided"
<sphalerite>
_d0t: no, plugdev is just a group that exists by convention on debian systems iirc.
<sphalerite>
_d0t: so those rules are for debian. A good reason not to use them verbatim ;)
<cransom>
there are some defined groups, but they are all defined by modules that are using them. otherwise they don't exist.
<{^_^}>
[nixpkgs] @peti pushed 11 commits to haskell-updates: https://git.io/fAbAF
<_d0t>
should I change the group in bladeRF for something that already exists in NixOS?
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « haskell-hspec: fix build with ghc-8.6.x »: https://git.io/fAbxz
civodul has joined #nixos
trcc has joined #nixos
alex`` has quit [Quit: WeeChat 2.2]
alex`` has joined #nixos
<cransom>
if you are going to create a module for it, you could create your own bladerf group. otherwise, i don't see an advantage to update the package.
notlar has joined #nixos
<_d0t>
cransom, then that's what I should do, perhaps
lopsided98 has quit [Quit: Disconnected]
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « haskell-doctest: fix build with ghc-8.6.x »: https://git.io/fAbpm
lopsided98 has joined #nixos
orivej has quit [Ping timeout: 252 seconds]
ixxie has quit [Ping timeout: 252 seconds]
georges-duperon has quit [Ping timeout: 252 seconds]
<sphalerite>
(bladerf reads really nicely as "bluh-durf")
<lostman>
is it possible to make ghc ignore user's package database when running in nix-shell? can I somehow tell ghc/cabal to only use the package database provided by nix?
zduch4c has joined #nixos
<{^_^}>
[nixpkgs] @7c6f434c pushed 4 commits to release-18.03: https://git.io/fANf0
<{^_^}>
[nixpkgs] @xeji pushed commit from @dywedir to master « fzy: 0.9 -> 1.0 (#47295) »: https://git.io/fANUl
<emily>
how on earth does your terminal echo your password
<zduch4c>
It’s easy to remember lol
<emily>
oh, eshell
<zduch4c>
Still doesn’t work _d0t
Henson has joined #nixos
<zduch4c>
I’m going to try with a different hotspot
<_d0t>
zduch4c, that's because your wifi auth failed
<_d0t>
swi
<_d0t>
sorry, wrong window
<zduch4c>
-
<zduch4c>
So what does auth failed mean?
<_d0t>
Probably means that you provided the wrong passphrase.
<_d0t>
Though wpa_supplicant is a weird thing, so one can't be sure.
<zduch4c>
Mhm
<zduch4c>
Okay I need to change my passwords now on another note
<zduch4c>
Unless you guys promise not to tell anyone
<Henson>
is it a good idea to run a nix program from within nix-shell for purposes other than development, if one is interested in having a pure environment?
<Henson>
or is there a more proper way of doing that?
<{^_^}>
[nixpkgs] @xeji pushed commit from @veprbl to master « lhapdf: fix "lhapdf list" interactive tool (#44905) »: https://git.io/fANTi
sir_guy_carleton has quit [Quit: WeeChat 2.0]
<{^_^}>
[nixpkgs] @Mic92 merged pull request #47269 → dotnet-sdk: Reduce closure: Unpack to a subdirectory to prevent copying build files → https://git.io/fAblO
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @baracoder to master « dotnet-sdk: Reduce closure: Unpack to a subdirectory to prevent copying build files (#47269) »: https://git.io/fANTD
<zduch4c>
It associated! But disconnected afterwards
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @baracoder to release-18.09 « dotnet-sdk: Reduce closure: Unpack to a subdirectory to prevent copying build files (#47269) »: https://git.io/fANTH
<{^_^}>
[nixpkgs] @xeji pushed commit from @fdns to master « compactor: 0.11.0 -> 0.11.1 (#47271) »: https://git.io/fANqa
<{^_^}>
[nixpkgs] @Ericson2314 opened pull request #47299 → lib: Deprecate `overrideScope` in lieu of `overrideScope'` taking arguments in the conventional order for 18.09 → https://git.io/fANmX
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #47238 → lib: Deprecate `overrideScope` in lieu of `overrideScope'` taking arguments in the conventional order → https://git.io/fAF4G
<{^_^}>
[nixpkgs] @Ericson2314 merged pull request #47299 → lib: Deprecate `overrideScope` in lieu of `overrideScope'` taking arguments in the conventional order for 18.09 → https://git.io/fANmX
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to release-18.09: https://git.io/fAN3q
graphene has quit [Remote host closed the connection]
mayhewluke has joined #nixos
graphene has joined #nixos
<catern>
is there a way to enforce at the daemon level, no local builds, only pull from the binary cache?
<{^_^}>
[nixpkgs] @xeji pushed 47 commits to staging-18.09: https://git.io/fAN3w
Fare has quit [Ping timeout: 245 seconds]
<catern>
also, I'm a little unclear: nix-build has a --fallback option and claims that falling back to a local build is not default, but it seems to fall back all the time
<catern>
and I'm not passing --fallback
<catern>
I'd rather it only use binary caches and fail if it needs to build locally
emerson has quit [Ping timeout: 630 seconds]
ma27 has joined #nixos
<elvishjerricco>
catern: It will always prefer the binary cache if the cache has the derivation you're asking for. The only case I've seen this fail is when you try the build before it's on the cache, kill it, and then try again after it's been uploaded to the cache. Nix caches which builds are not on which cache so that it doesn't have to recheck them. But I think this might have been changed in Nix 2.0... Not sure. Regardless, this
<elvishjerricco>
cache is stored in ~/.cache/nix/binary-cache-v5.sqlite{,-journal}
<catern>
elvishjerricco: I don't think that is an answer to my question: I know that Nix will not build if the artifact is in the cache, but I want it to also not build even if the artifacts is not in the cache: rather it should just fail
<elvishjerricco>
catern: Ah. I'm not sure then... Maybe `--max-jobs 0`?