<sviteri>
samueldr: but I think that you are saying that the path part creates a problem?
<samueldr>
yeah, it would need some cleverness to ensure I'm not using a path
silver has quit [Read error: Connection reset by peer]
<samueldr>
I'll see what the fine folks making nix think of it, I'll link this discussion
<sviteri>
samueldr: I think fundamentally the complexity increase comes down to the need to impose an ordering for the parse of attribute, which is currently not necessary
<sviteri>
samueldr: Then I can say -- take everything _after_ --keep-vars, and _before_ --*
<sviteri>
samueldr: this is easy to do with regex matching, but would require a rewrite of a significant amount of the nix-build file
<samueldr>
oh, pretty sure this can't happen (as it would be a breaking change)
<samueldr>
`nix-shell hello bash -p` works
<samueldr>
`nix-shell --keep-env foo hello bash -p` should still make `hello, bash` be selected for backwards compat
drakonis has joined #nixos
<samueldr>
--keep-var* obviously
<acowley>
I have a patch I thought I should perhaps put through nox-review, but nox-review is complaining that a package in one of my unrelated overlays needs an argument that I define in my user's packageOverrides. How should one navigate this?
<samueldr>
acowley: when I test PRs, I use `env -i nix-build` which blasts away all of my environment
<samueldr>
maybe it's possible to do the same for nox-review
<samueldr>
(let's hope it doesn't need an environment)
<sviteri>
samueldr: I'm exiting, thanks for your help today!
<acowley>
samueldr: I don't think I understand your response
sviteri has quit [Quit: Leaving]
<acowley>
samueldr: You run `env -i nix-build`, and then a nox-review command?
<samueldr>
acowley: `env -i` can be used to start another program without inheriting your environment
<samueldr>
so it would be `env -i nox-review`
<samueldr>
(I never actually used nox-review :/)
<acowley>
Oh, I see!
<acowley>
If I do that, it can't find nox-review
<acowley>
because I'm in a nix-shell I'd guess
<samueldr>
hm, could it be using PATH to call itself?
<samueldr>
env -i PATH="$PATH" nox-review
<samueldr>
(that's how you'd pass variables you need if needed)
<acowley>
nox-review itself isn't working when I try variations of that. I think it needs some nix things from the environment
<iqubic>
Has anyone run into issues with steam-run displaying all the text as black outlined rectangles?
<iqubic>
Anyone know how to fix that?
<acowley>
It seems as though something like nox-review would be very good to have. It's a shame it isn't a more active project.
<samueldr>
acowley: possibly, though for the record, works wonder with nix-build
<acowley>
I don't want to PR a change that breaks things for people.
<samueldr>
acowley: alternatively, mangle your NIX_PATH for that shel
<acowley>
Yeah, I think some surgery is needed, but I don't really want to do it as the less automated it all is the less likely it is to be a useful exercise
<acowley>
I guess if it's finding that overlay of mine, it's because it's checking $HOME for a config.nix. But if I reset HOME, then nox-review can't find a place for its own ~/.nox directory. Oy vey.
<acowley>
I feel like I'm going to end up grepping nixpkgs for downstream packages and just nix-building them manually
<acowley>
That's actually not the absolute worst thing in this case.
<sir_guy_carleton>
so for a pull request, should i commit the change to the master of my fork?
<iqubic>
Question: Can I do something like "nix-shell; echo $PATH; exit" in a bash script and have that echo the nix-shell $PATH, instead of the default $PATH from outside?
<Enzime>
sir_guy_carleton: doesn't matter, I personally make a branch in my fork
<Enzime>
so that my master tracks Nixpkgs master
<sir_guy_carleton>
Enzime: okay, good advice thanks
<iqubic>
looks like "nix-shell; echo $PATH; exit" drops me into a nix-shell without doing anything else. It didn't do what I thought it would. :(
<cmcdragonkai>
Is there a way to have variable attribute keys?
<cmcdragonkai>
`let a = "b"; in { "$a" = 3; }.b`
Guanin has joined #nixos
<voiceftp>
cmcdragonkai, see builtins.listToAttrs
MP2E has quit [Remote host closed the connection]
<ekleog>
iqubic: nix-shell --run 'sh -c "echo $PATH"' should work :)
MP2E has joined #nixos
<ekleog>
when you use ';' in the shell it's not equivalent to '<Enter>', it's first executing nix-shell, then executing 'echo $PATH', then executing 'exit'
freeman42x]NixOS has quit [Ping timeout: 240 seconds]
kyren_ has quit [Ping timeout: 260 seconds]
<ekleog>
and so you get dropped in a nix-shell, and when you'll exit it you'll get the PATH printed out and just after it'll exit
<ekleog>
,help
<{^_^}>
Use `,` to list all commands, `,foo = Foo!` to define foo as "Foo!", `,foo =` to undefine it, `,foo` to output "Foo!", `,foo somebody` to send "Foo!" to the nick somebody
<{^_^}>
let a="b"; in { "${a}" = 3; }.b: I can evaluate nix if you prefix your expression with "> ". Use it here only for short demos or helping others with the syntax. If you wanna spam me, do it in #bottest!
<ekleog>
> let a="b"; in { "${a}" = 3; }.b
<{^_^}>
3
<ekleog>
^ cmcdragonkai
<ekleog>
(easier than builtins.listToAttrs)
<ekleog>
(you can do any kind of string interpolation in here, btw)
<iqubic>
The list of commands has grown since I last checked it.
<cmcdragonkai>
Awesome
spear2 has quit [Remote host closed the connection]
spear2 has joined #nixos
kyren has joined #nixos
kyren has quit [Client Quit]
kyren has joined #nixos
ilbelkyr has left #nixos [#nixos]
<sir_guy_carleton>
so i can merge my branch in my forked repo to the master branch without any problems, right?
halfbit has quit [Ping timeout: 265 seconds]
tusj has joined #nixos
wigust has quit [Ping timeout: 240 seconds]
wigust has joined #nixos
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
drakonis has quit [Read error: Connection reset by peer]
<cmcdragonkai>
Given that I want to use `"a.b"` as a path string, is there a built in function that will split this into `["a" "b"]` so I can use the `lib.getAttrFromPath` and other attrset functions?
philippD has joined #nixos
<cmcdragonkai>
iqubic: Yes you can, but you need to set your `PIP_PREFIX` and `PYTHONPATH` appropriately.
<clever>
> lib.splitString "." "a.b"
<{^_^}>
[ <CODE> <CODE> ]
<iqubic>
I don't know what either of those do.
<clever>
cmcdragonkai: the above code should do what you want
<iqubic>
So I take it pip install is not the right way to install libraries on NixOS.
<iqubic>
What is the right way to get a python environment with the libraries I want?
<clever>
iqubic: python.withPackages in nix
<iqubic>
How does that work?
<clever>
iqubic: drawing a blank, can you name a python package?
<iqubic>
NumPy
<clever>
python.withPackages (ps: [ ps.numpy ])
<clever>
this generates a python that already has numpy in its search path
<clever>
stick it into the buildInputs of your nix-shell
<iqubic>
Why is ps used twice there?
<clever>
the 1st is defining a function that takes 1 arg, called ps
<clever>
the 2nd is using ps to get numpy
<iqubic>
python.withPackage is a function, right?
<clever>
yes
<clever>
its a function that takes another function
<iqubic>
Right. I see.
<{^_^}>
[nixpkgs] @cjpbirkbeck opened pull request #44334 → Added a link to the new Discourse forum. → https://git.io/fNKtm
Ariakenom has joined #nixos
crmlt has joined #nixos
endformationage has quit [Quit: WeeChat 1.9.1]
<jasongrossman>
Hi all. I'm idly thinking about getting a cheap tablet and putting NixOS on it. Is that crazy? Its only function would be playing music.
<clever>
jasongrossman: id use either nixops or not-os for something like that
<clever>
nixops would be simpler
* jasongrossman
looks up not-os
<jasongrossman>
My concern was whether there's enough touch-friendly UI software.
<jasongrossman>
Not-os looks fun.
<clever>
ah, both nixops and not-os entirely ignore the UI perspective
<clever>
either way, you have to find something to handle the ui and install it
<clever>
xorg is also untested under not-os, so getting graphics may be more difficult
<jasongrossman>
OK. Not worth it just for playing music then. Maybe a fun project for other uses.
<jasongrossman>
Thanks!
<clever>
nixops would let you trivially update the tablet from another system, so then you just need to services.xserver.enable = true; and install a graphical media player
iqubic has quit [Ping timeout: 265 seconds]
<jasongrossman>
Excuse my ignorance, but does NixOS work OK on the ARM chips that presumably tablets have?
<clever>
i have had it working on an armv6 and v7 raspberry pi before, and others have it working on aarch64
<jasongrossman>
Great. Thanks.
<jasongrossman>
I'll think about hardware now then and worry about software again later.
Ross has quit [Ping timeout: 260 seconds]
<clever>
sphalerite has also been doing some work with an android tablet i believe
<jasongrossman>
:-)
dnovosel has quit [Ping timeout: 276 seconds]
<clever>
when using nixops locally, you would start by just getting nixos to install normally, then you can upgrade ti to nixops
fragamus has joined #nixos
<sphalerite>
Chromebook, not android tablet
<sphalerite>
jasongrossman: nixos is based on mainline linux
goibhniu has joined #nixos
iqubic has joined #nixos
<sphalerite>
jasongrossman: android is a very different beast, and it's likely to be a massive pain to get a real linux working on a random android device
<sphalerite>
samueldr had a little project to marry nixos with https://postmarketos.org/ iirc, not sure what the state of that is
<jasongrossman>
Good advice. Thanks. I'll probably stick to lineageos for tablets.
<jasongrossman>
But maybe I'll get a Chromebook instead. :-)
<ekleog>
<sir_guy_carleton> so i can merge my branch in my forked repo to the master branch without any problems, right? <-- if your branch is based on a reasonably recent version of the nixpkgs repository, you can submit a pull request to the master branch of nixpkgs, yes
<sir_guy_carleton>
btw, has anybody here gotten nix on a *BSD (besides macOS, if you count it as a BDS)?
<ekleog>
<iqubic> Will "pip install" work in a nix-shell? <-- it does (already had to), but it's often regarded as a bad ideau
<iqubic>
ekleog: What should I do instead?
<ekleog>
oops
<clever>
iqubic: python.withPackages
<ekleog>
iirc the idea is that you should use python.withPackages
<jasongrossman>
sphalerite: Thanks! That's very good to know.
<jasongrossman>
sphalerite: As it happens, I wouldn't need wifi.
<jasongrossman>
sphalerite++
<{^_^}>
sphalerite's karma got increased to 10
<sphalerite>
jasongrossman: you will need some way to connect to the internet though presumably, and they don't have ethernet ports
<sphalerite>
so you'll need a USB wifi adapter or usb ethernet adapter
<sphalerite>
Biggest issue with the C201 is that it's ARMv7, so binary cache coverage is… limited
reinzelmann has joined #nixos
<sphalerite>
I don't have a graphical browser running on that. Barring that though, I could almost use it as a daily driver provided I have remote access to some more powerful machine
<sphalerite>
Of course aarch64 is supported better, so you might want to consider an aarch64 chromebook (but I'm not sure if those are as freedom-compatible)
<{^_^}>
[nixpkgs] @xeji pushed commit from @dywedir to master « libtorrentRasterbar: 1.1.7 -> 1.1.9 (#44335) »: https://git.io/fNKY9
Ross has joined #nixos
eeva has joined #nixos
troydm has quit [Ping timeout: 260 seconds]
page has quit [Ping timeout: 256 seconds]
kyren has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
page has joined #nixos
cfricke has joined #nixos
<jasongrossman>
unacceptable: And then you had to key those files in to the front panel on your mainframe.
<unacceptable>
It sometimes came on pre-punched cards attached to the magazine, but those were often tampered with by Punk Kids (TM)
troydm has joined #nixos
lpsmith- has joined #nixos
sheyll has quit [Ping timeout: 256 seconds]
lpsmith has quit [Read error: Connection reset by peer]
makefu has quit [Ping timeout: 256 seconds]
LnL has quit [Ping timeout: 256 seconds]
Unode has quit [Ping timeout: 256 seconds]
lvmond has quit [Ping timeout: 256 seconds]
LnL has joined #nixos
<ekleog>
actually, for just playing music, I guess the nixos live cd would make a good install source, by copying from the store of the CD? 😇
markus1189 has quit [Ping timeout: 256 seconds]
<ekleog>
DVD*, actually
ThatOtherGuy has quit [Ping timeout: 240 seconds]
Unode has joined #nixos
<srhb>
unacceptable: If you have a sufficiently fast system, you can just generate random bytes until you have the right hash for /run/current-system -- one of the nice things with NixOS.
kyren has joined #nixos
<ekleog>
errrhm… actually, no, you'll likely hit a hash collision long before that :p
<{^_^}>
[nixpkgs] @xeji pushed commit from @bignaux to master « photon: init at 1.0.7 (#44269) »: https://git.io/fNKOo
<jasongrossman>
ekleog: Good point, but there's a very high probability that the hash collisions will turn out not to be executable, so you can automate discarding the wrong ones.
<jasongrossman>
ekleog: Ooh, live NixOS USB stick. That is a good idea.
<ekleog>
jasongrossman: there's like 256bits in a nix hash, so if you just take a 32-char string that can be randomized to whatever by keeping the program executable, it'll still be executable
<ekleog>
including shell comments
kyren has joined #nixos
<ekleog>
I think you'll have at least several weird constants / character strings, not even counting modifications that are still executable :)
<TweyII>
What's the new-style equivalent to nix-build -Q --no-out-link foo, i.e. build foo and print the store path?
<srhb>
TweyII: There is none.
oida has quit [Quit: byez]
<srhb>
Build it, print the path, remove the path.
<srhb>
This is a funny failure, can someone explain it to me? It looks like "cat" is missing in the test environment, however I cannot reproduce that locally: https://hydra.nixos.org/build/78608184/nixlog/1
<srhb>
This is with all safety features that I know of enabled.
<clever>
srhb: do you have the nix sandbox enabled?
<clever>
srhb: grep sandbox /etc/nix/nix.co
<clever>
nf
<srhb>
I do :)
<clever>
and does your drv have the exact same hash as hydra?
<srhb>
Yeah, I did nix-build /nix/store/9s01s37zw7xzlmw2gs53adiiivhnw2fp-conduit-extra-1.3.0.drv --check afterwards to be sure, which is the hash the log reports at the top.
oida has joined #nixos
<clever>
thats strange then
<srhb>
Indeed!
hakujin3 has joined #nixos
<srhb>
checking the PATH with a pure nix shell reports that coreutils really is there.
<clever>
srhb: does it ever make use of /usr/bin/env?
<srhb>
Not as far as I can tell. It's the generic haskell builder after all.
<TweyII>
srhb: Oh okay :( Thanks
<TweyII>
I'll stick to nix-build, that seems suboptimal. Guess I could generate a unique tmp path or something
<srhb>
TweyII: There's an argument to be made that it's inherently tricky to do it any other way in face of garbage collection. But the old way isn't deprecated, so... :)
<TweyII>
Yeah :)
<TweyII>
srhb: The issue is potential race conditions if you GC and build at the same time?
<TweyII>
Since printing doesn't create a root?
<srhb>
TweyII: Right. If there's no gc root (./result)
<TweyII>
Mmh.
<TweyII>
Not a problem for my case, thankfully
<ng0>
what is the "sandbox" output of chromium in nix?
<srhb>
Yeah, most of mine not either. But I do see the argument. Safer to leave the root, do the thing, remove the root.
rprije has quit [Remote host closed the connection]
rprije has joined #nixos
toppler` is now known as toppler
<sphalerite>
TweyII: using the output symlink is the "right" way because of the GC race conditions you mentioned, and as such is also used by nixos-rebuild for example
<sphalerite>
TweyII: you could also use nix-shell as that will also create a temporary gc root for you
<sphalerite>
ng0: it contains chromium's wrapper for sandboxing
<ng0>
ok, thanks
<ajs124>
How does backporting to release branches work in NixOS? There is this commit in master to a module, that I would like to have in 18.03, so do I just create a PR against release-18.03 or how does this work?
<clever>
ajs124: yeah, thats how i believe it works, fork a branch from release-18.03, then file a PR back to it
<ajs124>
clever, ok thanks, will do
nschoe has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @volth opened pull request #44350 → xfce4-13.xfce4-panel: fix path to tzdata → https://git.io/fNKaD
ericsagnes has joined #nixos
r0bby has quit [Remote host closed the connection]
Wizek has quit [Remote host closed the connection]
philipcristiano has quit [Remote host closed the connection]
teozkr has quit [Remote host closed the connection]
nz has quit [Remote host closed the connection]
colemickens has quit [Remote host closed the connection]
nikki93 has quit [Remote host closed the connection]
vdemeester has quit [Remote host closed the connection]
themistocle__ has quit [Remote host closed the connection]
benkolera has quit [Remote host closed the connection]
fingerzam has quit [Remote host closed the connection]
sorear has quit [Remote host closed the connection]
codedmart has quit [Remote host closed the connection]
ilovezfs_ has quit [Remote host closed the connection]
etrepum has quit [Remote host closed the connection]
drag_ has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @Zimmi48 opened pull request #44353 → graphql-cli: init at 2.16.5 → https://git.io/fNKrz
<infinisil>
siers: I think the new fetchGit builtin does that now with nix 2.0
<siers>
Hah, ok
<qknight>
betaboon: thanks btw
vmandela has joined #nixos
<sphalerite>
infinisil: can you fetchGit a non-git directory with a .gitignire?
vmandela has quit [Quit: Leaving]
<siers>
Would you be able to do a check-out of a dirty work🌲?
<sphalerite>
infinisil: nix-gitignore would still be relevant I think, also for dirt
<sphalerite>
damn siers beat me to it :p
<sphalerite>
siers: also that is some next-gen spelling :D
Neo-- has joined #nixos
<siers>
Haha, android autocomplete gave the idea
aarvar has quit [Ping timeout: 260 seconds]
nschoe has joined #nixos
<betaboon>
qknight: you're welcome
jasongrossman has quit [Remote host closed the connection]
pie_ has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @xeji merged pull request #44351 → masterpdfeditor: fix broken source url. the old one 404s → https://git.io/fNKwW
<{^_^}>
[nixpkgs] @xeji pushed commit from @ajs124 to master « masterpdfeditor: fix broken source url. the old one 404s (#44351) »: https://git.io/fNKK2
<borodust>
so it should work for me, but it doesnt
<borodust>
i have it in the /etc/nixos/configuration.nix, so should be system-wide
<gchristensen>
no, configuration.nix doesn't set it system-wide _just_ for the packages described
<srhb>
borodust: It should work for things installed via nixos-rebuild switch only
<srhb>
eg. systemPackages and co.
<borodust>
aaaah
<borodust>
damn, i should have read the page fully, jeez, my apologies
<srhb>
The wikipedia is a bit misleading too... I'll try to edit it
<sphalerite>
wikipedia?
<srhb>
Err, wiki.
<srhb>
"When using NixOS, it is possible to enable unfree for eg. environment.systemPackages with the following setting in your /etc/nixos/configuration.nix: ..." <-- better?
<nly>
can anyone show me how to setup unbound dns cache please :)
<{^_^}>
[nixos-weekly] @garbas pushed commit from NixOS Weekly Robot to gh-pages « Preview of '5496f5601073142f74594f3c63d7e05384ac58f9' commit built by Travis-CI \#411257554 »: https://git.io/fNKiq
<{^_^}>
[nixos-weekly] @garbas pushed commit from NixOS Weekly Robot to gh-pages « Preview of '72bdcabde3cd00e15e5f481d9c9e2fbe0aa97dcc' commit built by Travis-CI \#411260771 »: https://git.io/fNKiX
<{^_^}>
[nixos-weekly] @domenkozar pushed 3 commits to production: https://git.io/fNKM2
<srhb>
davidak[m]: I believe usually, /nix/var/nix/profiles/per-user/root/channels and $HOME/.nix-defexpr/channels are included in the default NixOS NIX_PATH, so if a channel named nixos-unstable is there, it should work.
<{^_^}>
[nixos-weekly] @garbas pushed commit from NixOS Weekly Robot to gh-pages « Release of (commit '65117749ae02f274aee651e6ef23ccc553445884') built by Travis-CI \#411272538 »: https://git.io/fNKMo
<srhb>
And otherwise defining NIX_PATH yourself will also work, as long as the name is there.
<davidak[m]>
it should work, that's what i'm also thinking. but it doesn't. maybe i have done something wrong or something is wrong with NixOS :D
<davidak[m]>
it's probably some stupid mistake on my side since i havn't done it yet
<{^_^}>
[nixpkgs] @xeji pushed commit from @ktf to master « doing: init at 1.0.10pre (#44034) »: https://git.io/fNKSC
<davidak[m]>
ok. i will create an issue to discuss it
<srhb>
Cool :)
<davidak[m]>
how do you think we could get rid of them?
<srhb>
I personally feel they're a senseless abstraction compared to just checking out channels and managing them with git, or fetching them with the builtins, or ...
<davidak[m]>
i like to be able to have a stable system and use SOME packages from unstable
<srhb>
davidak[m]: As do I
<srhb>
davidak[m]: So I have fooBranch = fetch some revision; barBranch = fetch some other revision; ...
<davidak[m]>
inside configuration.nix?
<srhb>
There and in other places
<srhb>
I've also just used a local checkout for a long time
<srhb>
I could use more than one if I wanted to
<srhb>
eg nixpkgs=/home/sarah/nixpkgs
<davidak[m]>
and don't have nix-channel?
<srhb>
Nope, zero channels for me.
<davidak[m]>
wouldn't that make it more complicated for a normal user that don't want to change anything in nixpkgs?
<srhb>
davidak[m]: Possibly, yes. Every approach I've used relies on being able to use git for rollbacks.
<srhb>
Or some VCS.
<srhb>
I don't know of another approach that's simpler and doesn't rely on VCS. Channels are supposed to be it. :)
Dedalo has joined #nixos
<davidak[m]>
with channels you are always up to date with hydra
<srhb>
davidak[m]: No, you're up-to-date if you run nix-channel --update
<srhb>
davidak[m]: I'm up-to-date if I git pull :)
<thoughtpolice>
I don't use channels either, but I'd say I'm more of an upstream hacker (I have commit bits at least...) I more or less track master directly on all my machines, but yes this isn't for everyone.
<davidak[m]>
you check out the channel repo?
<srhb>
davidak[m]: It's one of my remotes, yes.
<thoughtpolice>
Normally I checkout something into /etc/nixos/nixpkgs as a repository clone, keep that in sync, push everything upstream as much as possible.
sorear has joined #nixos
<thoughtpolice>
the nix.nixPath option also helps a lot to make sure you always use a custom checkout
<thoughtpolice>
But! You can still use this method, and get the binary cache!
<srhb>
Yeah, I usually cherry onto nixos-unstable from there once I'm doing gooping around.
<srhb>
Er, done*
<thoughtpolice>
That repository is essentially a copy of the upstream nixpkgs. But it is fast-forwarded only when Hydra updates a channel, e.g. release-18.03
<thoughtpolice>
So this gives you the best of both worlds, if you want that. Caches, but some flexibility to modify nixpkgs, if you need, and if you set nix.nixPath it's more or less transparent for users, as well.
<thoughtpolice>
Best IMO, anyway
<srhb>
I enjoy that method at least :)
<makefu>
thoughtpolice: a lot of people pin their nixpkgs by checking it out in some state directory. it is a great way to keep everything working
<srhb>
But yes, it relies on git knowledge. I don't think basic use of that is far more difficult than understanding the channels mechanism though :D
<makefu>
i also have the nixpkgs channel checked out and i cherry-pick my changes onto this repo
<thoughtpolice>
Yeah. If you don't even need to hack nixpkgs, you don't even have to use git
<thoughtpolice>
You can set NIX_PATH=nixpkgs=https://github.com/nixos/nixpkgs-channels/archive/master.tar.gz or whatever I think and it'll download that for you, I'm pretty sure
Compaq-Local has joined #nixos
<sphalerite>
thoughtpolice: please don't suggest that for people to use master
<thoughtpolice>
Nix is pretty aggressive about checking for new versions though, when you use an HTTP url
<makefu>
but i think the master is like worst case scenario
<{^_^}>
[nixos-weekly] @domenkozar pushed to master « 2018/05: fix link »: https://git.io/fNKHY
<thoughtpolice>
sphalerite: In practice that method is useless anyway
<thoughtpolice>
Because 99.9999% of anyone, practically speaking, is always going to have to make a patch for something.
<sphalerite>
thoughtpolice: regardless, master we don't want people who don't know exactly what they're doing to use master
<{^_^}>
[nixos-weekly] @domenkozar pushed 2 commits to production: https://git.io/fNKHs
<sphalerite>
blargh typing with high-latency feedback is hard
<thoughtpolice>
Sure, that's fine, I'm just blabbing.
<sphalerite>
thoughtpolice: so please stick to recommending the channels :)
<{^_^}>
[nixos-weekly] @garbas pushed commit from NixOS Weekly Robot to gh-pages « Preview of '6593fc18c52dce94fe6960921ffb1a174390baa5' commit built by Travis-CI \#411291029 »: https://git.io/fNKHW
<thoughtpolice>
(It was more about the convenient HTTP mechanism, but even complaints of using master aside, it's still mostly useless. Just a nice feature to know, occasionally)
<sphalerite>
something like NIX_PATH=channel:nixos-unstable could work though :)
<thoughtpolice>
Actually it's not entirely bad.
<thoughtpolice>
I use it for nix-shell shebangs, which is nice for scripting
<davidak[m]>
as a user, i like to run just one command and have a fresh system like nixos-rebuild switch --upgrade
<{^_^}>
[nixos-weekly] @garbas pushed commit from NixOS Weekly Robot to gh-pages « Release of (commit 'c5f13f6ae3c1154cc9b9a22282efc201a00aeda7') built by Travis-CI \#411291482 »: https://git.io/fNKHr
<davidak[m]>
i just use git checkout for hacking on nixpkgs
<sphalerite>
maybe we should rename the master branch to "breack-my-sysstem"
<sphalerite>
minus typos
FRidh has joined #nixos
<makefu>
or "no-sleep-for-you-today" or "prepare-to-revert"
<davidak[m]>
when i work on different stuff, like editing videos, i just want a stable system and don't care about the details
<sphalerite>
makefu: no, prepare-to-revert suggests that it'll still be possible to revert
<thoughtpolice>
I surprisingly don't revert off master very much, but then again a lot of my systems are headless.
<sphalerite>
makefu: it has happened that master has broken the bootloader
<thoughtpolice>
Actually I've been running a custom patchset *on top of updated master* for weeks now with little issue. Daring.
<sphalerite>
thoughtpolice: you know there risks and are prepared to deal with the consequences I presume
<infinisil>
thoughtpolice: Yeah but that's the thing about master: You have no guarantees, it could break never or tomorrow
<sphalerite>
it's not a case of "this will immediately ybreak your sysstem" or "this will break your system in a week" or even "this will break your system in a year"
<thoughtpolice>
infinisil: Oh, I'm well aware of that much, believe me.
<sphalerite>
please pretend I'm not making so many ttypos >_>
<etu>
sphalerite: mosh is very nice if you have a slow connection and use a client over SSH :-)
* etu
is experiencing that at this moment
<etu>
s/slow/laggy/
<sphalerite>
yeah that is indeed the case
<sphalerite>
buuuut effort to set it up :p
<etu>
programs.mosh.enable = true;
<etu>
:p
<sphalerite>
and does the echoing help wihtin tmux as well?
<etu>
It does
<infinisil>
We're all lazy after all, just like Nix
<thoughtpolice>
tmux is weird if you do vertical splits (especially on latent connections) but otherwise it's fine and does help
<sphalerite>
good to know, thnanks!
<sphalerite>
p
<thoughtpolice>
That's because during latent spikes mosh will "push over" the columns in the line where you type, because it has to draw your keystrokes as they happen, not as the server responds.
<etu>
sphalerite: It even works in weechat in tmux (like I'm doing it right now. And I get some underlining for text that isn't transfered yet)
johnw has joined #nixos
<sphalerite>
perfect. I'm in weechat in tmux without mosh and it's not so much fun
<etu>
But yeah, with the quirks that thoughtpolice mention of course. But it's still better.
<thoughtpolice>
Also happens if your PS1 has something on the RHS for example like your CWD. But yeah the latency and mobile roaming are completely worth minor redraw issues
fiddlerwoaroof has quit [Read error: Connection reset by peer]
<thoughtpolice>
There was also allegedly work to help try and fix that I thought, and stuff like ssh-agent forwarding, but mosh development stalled out a while ago last I remember. It's stable and works fine though, don't know if it's picked back up. It basically does its job.
phreedom_ has joined #nixos
<sphalerite>
I need to use programs.mosh.emable on the server and just install mosh on the client?
phreedom has quit [Ping timeout: 250 seconds]
<thoughtpolice>
Yeah, mosh needs a custom set of UDP ports open through your firewall on the server, so programs.mosh is just a convenient wrapper for adding that to networking.firewall
hakujin3 has joined #nixos
fiddlerwoaroof has joined #nixos
<{^_^}>
[nixpkgs] @basvandijk merged pull request #44331 → lift version restrictions on language-ecmascript inputs → https://git.io/fNKvr
<thoughtpolice>
Basically mosh just first logs in with SSH in order to do key exchange securely (and for convenience), then once that happens it shares an ephemeral private key between client/server, server opens a UDP port, and tells the client to connect there with the key. That's how it bootstraps itself, and how stateless/mobile roaming works
<tilpner>
You shouldn't generalize that. It depends on your usecase. Do you need all dependencies to be the exact same as they were a while ago?
joehh has quit [Ping timeout: 264 seconds]
dbmikus has joined #nixos
<tilpner>
Importing an older channel is more expensive, in memory, storage, and network. But it's your only option if you need the exact same dependencies
Ariakenom_ has quit [Ping timeout: 240 seconds]
<haitlah>
tilpner: In the case I need a newer version of a dependency (x+1), which is not supported by my older package version
bfrog has joined #nixos
<haitlah>
tilpner: if not recursively pointing, what's the solution ? having multiple .nix for each versions ?
<maurer>
Usually people have multiple .nix files, and if there's a lot of duplication, give them a common.nix or something to import from
<tilpner>
haitlah - You can try updating just a dependency of your older package, but there's no guarantee it'll work
<tilpner>
haitlah - During callPackage, take the dependency from your newer nixpkgs
<haitlah>
okay guys thanks
Lears has quit [Remote host closed the connection]
endformationage has joined #nixos
<azazel>
does the test system keeps cache of the previous runs?
<johnw>
i have a nix-scripts expression I use to install it
<johnw>
it just needs to be on the PATH, though
<johnw>
or put it in ~/.nix or something and use an explicit path in .envrc
<iqubic>
How do you make sure that emacs can see that script and use it to find executables?
<johnw>
emacs sees it on the PATH
<johnw>
I don't think I'm understanding your question
<adisbladis[m]>
I cache my direnv environments
<adisbladis[m]>
Moving around buffers is a breeze :)
<johnw>
yeah, use_nix.sh caches, using Nix
<adisbladis[m]>
Exwm would be unusable without it
xy2_ has joined #nixos
<johnw>
and protects direnv environments from GC
<cransom>
gchristensen: if you have a shell.nix that's non trivial, direnv on a directory that you go into often is painfully slow. some caching though and it's instant
<gchristensen>
maybe could teach direnv or nix-shell todo better :)
Ariakenom has joined #nixos
<iqubic>
If I create a shell.nix that install some python related executables like jedi and flake8, will emacs be able to find those executables when it looks for them?
<iqubic>
adisbladis[m]: You use EXWM? So do I.
<johnw>
I'd like to teach nix-shell to do a lot of things better :)
<adisbladis[m]>
johnw: I use nix-instantiate to calculate my hash instead though
drakonis has joined #nixos
Supersonic has quit [Quit: Bye]
<johnw>
use_nix.sh uses nix-instantiate
<iqubic>
adisbladis[m]: What do you mean by, "Cache the results" so that buffer switching isn't a pain?
<adisbladis[m]>
I found all other approaches too brittle
<johnw>
iqubic: emacs-direnv uses direnv to find out what all the shell vars should be, and then changes Emacs to use that variables. When you change buffers, it changes them back.
<johnw>
the nice thing about direnv is that if it works at the shell, it will work in Emacs, and vice versa
<johnw>
one approach to rule them all :)
<johnw>
i basically never use nix-shell anymore
<iqubic>
But you still write shell.nix files to decribe projct deps, right?
<johnw>
I just write a default.nix that works for both nix-build and nix-shell
<johnw>
and with optional pinning for Travis
<iqubic>
I never use default.nix, because I never use nix-build.
<iqubic>
johnw: I did not know you were an op in #org-mode.
<{^_^}>
[nixpkgs] @Enzime opened pull request #44368 → mpv: Fix mpv 0.29.0 on macOS and disable OpenAL → https://git.io/fN6Ut
<johnw>
I didn't either
<iqubic>
Are you an op there?
<johnw>
yes
<iqubic>
How did you not know that?
jperras has quit [Client Quit]
jperras has joined #nixos
<maurer>
I don't know about org-mode, but sometimes channels will just give soft ops (opped but chanserv won't regive them ops) to long time contributors as a defense against low operator presence
<johnw>
iqubic: very good question
jperras has quit [Client Quit]
<maurer>
So if org mode only has like, 1 or 2 official ops, they may just unofficially op a bunch of semi-trusted people
<johnw>
maurer: sounds likely
jperras has joined #nixos
phreedom has joined #nixos
<gchristensen>
that said, #org-mode does have you listed as a permanent full-time op, johnw
<johnw>
yay
<johnw>
guess I should add that channel to my list of auto-joins
<gchristensen>
in fact, it sees you are the most powerful op of the channel, and only one other person with ops - kensanata
<gchristensen>
/msg chanserv access #org-mode list
<iqubic>
gchristensen: kensanata didn't even know they were an op either.
<johnw>
oh, hey, I'm the Founder even
<iqubic>
Really/
<iqubic>
??
<johnw>
I guess I forget things
phreedom_ has quit [Ping timeout: 250 seconds]
<tilpner>
johnw - You can transfer founder-ship to people that are more likely to be present (and remember)
<johnw>
i'll stay on channel (haha)
sir_guy_carleton has joined #nixos
xy2_ has joined #nixos
Sonarpulse has quit [Ping timeout: 260 seconds]
<tilpner>
Any ideas how to allow a (otherwise unprivileged) user to bind mount a directory as long as the mountpoint is in /var/lib/name/foo?
<tilpner>
I guess security.sudo.extraRules with a wrapper around sudo that only uses mount, and which checks for proper mountpoint location, but that seems fragile
<clever>
tilpner: if you add the user flag to an entry in fstab, then a user is allowed to mount it
halfbit has quit [Quit: WeeChat 1.9.1]
<clever>
if mount was setuid root i believe, which it isnt on nixos
<clever>
-rws--x--x 1 root root 39872 Dec 20 2014 /bin/mount
<clever>
on gentoo, mount is setuid, but on nixos it isnt
<clever>
so you need to make mount setuid, and use the user flag in fstab, and it will still limit the users to using the exact device+mountpoint in the config
<tilpner>
clever - I want one user to be able to bind-mount /nix/store/* to /var/lib/name/roots/*, and exactly nothing more
jperras has quit [Quit: WeeChat 2.2]
<clever>
what for?
<tilpner>
To then virtfs them into a vm. I used your *insane* method of generating disk images, which worked wonderfully on my laptop, but ended up hdd-bound when deployed to a server, so now I need to redo it without disk images
<guibou>
I want to overrideCabal an haskell package in nixpkgs because that packages depends on an external binary. Should I set the ovveride inside `development/haskell-modules/configuration-nix.nix` ?
<TweyII>
sphalerite: I'm actually doing this from within a shell anyway
<TweyII>
So I think it's fine
betaboon has quit [Quit: WeeChat 2.1]
FRidh has quit [Quit: Konversation terminated!]
<srhb>
TonyTheLion: It's trying to use the internet. That's forbidden.
Compaq-Local has quit [Ping timeout: 240 seconds]
<srhb>
(At least, that's what it looks like)
<guibou>
I have a hackage package with some non-haskell dependencies. Hence the automatically imported package in nixpkgs is broken. Should I hack nixpkgs to add the depency (in `haskell-modules/configuration-nix.nix`) or should I hijack the `extra-libraries` flag of my package's cabal?
<srhb>
guibou: Without more knowledge, can't tell! :-)
<TonyTheLion>
srhb: so that means I cannot run jekyll in a nix-build?
<srhb>
TonyTheLion: Not with the usual safeties in place, if it uses the network.
<TonyTheLion>
oh
<TonyTheLion>
and I'm guessing there's no easy way around these safeties?
<TonyTheLion>
is this in place for security reasons?
<srhb>
There is, but I'd say it's a very bad idea. Builds are supposed to have no side effects.
<srhb>
It's about determinism as well
sigmundv has quit [Ping timeout: 268 seconds]
<TonyTheLion>
srhb: hmmm, well I may need to do bad things then
<TonyTheLion>
is it documented on how to ignore that side effect?
genesis has quit [Remote host closed the connection]
__monty__ has joined #nixos
pie_ has joined #nixos
Thra11 has quit [Quit: WeeChat 1.4]
erickomoto has joined #nixos
<aminechikhaoui>
TonyTheLion: I guess if you don't care about reproducibility you can disable sandboxing ? but then you can just run a shell script :D
genesis has joined #nixos
bfrog has quit [Ping timeout: 276 seconds]
Denommus has joined #nixos
<d1rewolf>
are there any practical concerns for making part of a container's file system writable from an external host user?
<infinisil>
TonyTheLion: Don't abuse Nix for doing stateful stuff!
<d1rewolf>
or...is there a better way to let a non-root user edit a container's files?
<drakonis>
infinisil, why not :V
<infinisil>
Because it makes no sense, you can just run a shell script for that, why would anybody want to use nix for that?
<infinisil>
Nix is a tool for building derivations, and not for deploying stuff or other such matter
jperras has joined #nixos
galaxite has joined #nixos
<d1rewolf>
would it be harmful to change the group of a container's file system?
<d1rewolf>
from root to something else?
<d1rewolf>
i'd like to give non-root users a way to edit and access on particular directory within the container's filesystem
Fare has quit [Ping timeout: 276 seconds]
Fare has joined #nixos
orivej has joined #nixos
<TonyTheLion>
infinisil: its useful tho for building a website like that, cause all deps are taken care of by nix, and you don't have to do any of yourself
troydm has joined #nixos
<TonyTheLion>
just run 'nix-build'and its done
<TonyTheLion>
unless you have silly state issues
<srhb>
TonyTheLion: Use nix to build the _script_ that builds your website then
<srhb>
With all the dependencies available
<infinisil>
Yes exactl ^^
rardiol has quit [Ping timeout: 240 seconds]
<azazel>
how it's possible to delete a store path and all the other paths that depend on it?
erickomoto has quit [Ping timeout: 240 seconds]
<coconnor>
d1rewolf: non-root users on the container or host?
<d1rewolf>
coconnor: on the host
UNIcodeX has joined #nixos
<TonyTheLion>
shrb: building the website is just executing jekyll
<azazel>
d1rewolf: I've done that in the past, non on Nixos btw
<coconnor>
d1rewolf: as the container is a separate namespace, the users/groups would have to be exported into it
<coconnor>
d1rewolf: then, afaik, a directory on the host with group rw can be unioned onto the container
<coconnor>
err mounted into the container
<d1rewolf>
coconnor: ok, i'll take a look at doing that. thx
rardiol has joined #nixos
TweyII has quit [Ping timeout: 260 seconds]
<coconnor>
d1rewolf: if your are using declarative containers it's the bindMounts option. otherwise *shrug*
<clever>
i'm a bit fuzzy on the exact value, let me check
<infandum>
Another way would be to set the script as the windowManager -- is there a way to do this? Like, services.xserver.windowManager.custom = "run this script"?
<clever>
15719 ? S 0:00 /nix/store/8zkg9ac4s4alzyf4a8kfrig1j73z66dw-bash-4.4-p23/bin/bash /nix/store/3akbkdg5qzn8kc4dkmk7qydzkxdqybm6-xsession xfce
<clever>
infandum: ah, so $sessionType will just be whatever the 1st arg to xsession was
palo has joined #nixos
leothrix has quit [Remote host closed the connection]
leothrix has joined #nixos
pie_ has quit [Read error: Connection reset by peer]
sigmundv has joined #nixos
pie__ has joined #nixos
igo95862 has joined #nixos
iMatejC has quit [Remote host closed the connection]
<infandum>
clever: Works perfectly, thank you!
<phry>
heyho :) I'm trying to override my zfsUnstable kernel package. I tried using an overlay and just overriding "zfsUnstable", but of course that is not working as it is called another way :/
<phry>
is there a way to override a module for the current kernel?
<Denommus>
hey
bennofs has joined #nixos
<Denommus>
when using node2nix, it generates a package that includes only files from the same directory as package.json
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
mkoenig has quit [Remote host closed the connection]
<clever>
phry: all-packages.nix contains a linuxPackagesFor function, which takes a given kernel derivation, and generates module derivations for every module, including zfs
<phry>
clever: yes, I've seen as much. But that references a file from within a function, so there's no way to override that from outside. I was hoping that I could prevent to check out another copy on nixpkgs with some magic ^^
<clever>
phry: i think your best bet is to do something like, linuxPackages = self.linuxPackages // { zfsUnstable = super.linuxPackages.zfsUnstable.overrideAttrs (old: { src = ...; }); };
<{^_^}>
home-manager#327 (by ghost, closed): "/nix/store/...-news-info.sh: No such file or directory" with -2
<drakonis>
can't install home-manager
<drakonis>
the install instructions aren't working out
erickomoto has joined #nixos
freeman42]NixOS has joined #nixos
<erickomoto>
Hi, I am having troubles installing docutils. I use `nix-shell -p python37Packages.docutils` and an installation test fails with the following:
<infinisil>
drakonis: Still the same error?
<erickomoto>
Oh, I guess people have reported this already?
<erickomoto>
The error is unknown locale: UTF-8
<drakonis>
yah?
freeman42x]NixOS has quit [Ping timeout: 244 seconds]
<drakonis>
error: syntax error, unexpected $undefined, at /home/drakonis/.config/nixpkgs/home.nix:3:32
<drakonis>
/nix/store/byrdbpzz4grh4h52zxx5fcq4ap5lb8yf-home-manager/bin/home-manager: line 101: /run/user/1000/home-manager-build.3ekbwTJW6R/news-info.sh: No such file or directory
UNIcodeX has quit [Quit: Leaving]
<infinisil>
drakonis: Um, that's a syntax error in your nix file
<infinisil>
The first error at least
<drakonis>
yeah i know that
logzet has quit [Read error: Connection reset by peer]
<infinisil>
And why do you think that's relevant?
<drakonis>
hmm i figured the problem
ng0 has quit [Ping timeout: 250 seconds]
<drakonis>
seems like simply copying the text that should be cat'd into the file doesn't work
_cyril_ has quit [Ping timeout: 240 seconds]
<drakonis>
the famous problem between the chair and the keyboard
_cyril_ has joined #nixos
<infinisil>
If you did figure out the problem then tell the solution, and what file? You're being confusing
<drakonis>
its my fault
<drakonis>
there's this command cat > ~/.config/nixpkgs/home.nix <<EOF
<drakonis>
i copied the text that was supposed to be introduced into the file
<drakonis>
but it seems it needs the EOFs or it will error out
ng0 has joined #nixos
Denommus has quit [Ping timeout: 244 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed to release-18.03 « firefox-esr-unwrapped: bring back the alias »: https://git.io/fN601
<drakonis>
so, i'm curious about triton again, looks like its still going
<drakonis>
a one man job
<gchristensen>
the bittorrent test is CRAZY y'all!
<gchristensen>
this is the most wild test!
<drakonis>
wait what
<clever>
gchristensen: i always use it as the best example of how to excersise the vlan features of the nixos test framework
<gchristensen>
it starts a tracker, a machine which can talk to the tracker, a router, a machine which can only talk to the first client through the router and not the tracker ... then verifies the second client can poke a hole with upnp through the router to the client
erickomoto has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
shabius has joined #nixos
<{^_^}>
[nixpkgs] @ngortheone opened pull request #44378 → Fix broken packges that do not build → https://git.io/fN6zV
<{^_^}>
[nixpkgs] @basvandijk opened pull request #44379 → pg_journal: init at 0.2.0 & fixed some PostgreSQL plugins related issues → https://git.io/fN6zK
<phry>
clever: now I'm absoluely irritated. Nothing seemed to work, so I decided to remove all kinds of zfs packagesfrom kernelPackages. It doesn't stop working. This seems to take kernelPackages from I-don't-know-where o_O
<phry>
Here's what I did: boot.kernelPackages = lib.traceVal (lib.filterAttrs ( n: v: n != "zfsUnstable" && n != "zfs" && n != "zfsStable" && n != "zfsLegacyCrypto" ) ( pkgs.linuxPackages ));
<{^_^}>
[nixpkgs] @xeji pushed commit from @lsix to staging « gnupg: 2.2.8 -> 2.2.9 (#43431) »: https://git.io/fN6gU
eskytthe has joined #nixos
katyucha has quit [Ping timeout: 265 seconds]
erickomoto has joined #nixos
<phry>
clever: here you go - https://pastebin.com/8rQFwter - I commented all my other includes out, so this really is what is running for that `nixos-rebuild build`
<clever>
phry: yeah, i would expect that to hail fail
<ivan>
is there any way to audit changes coming into nixpkgs? neither `tig` or `tig --first-parent -m` do something particularly useful (the latter because of how merges are done in nixpkgs)
philippD has joined #nixos
<{^_^}>
[nixpkgs] @ngortheone closed pull request #44378 → Fix broken packges that do not build → https://git.io/fN6zV
<ivan>
other than you know rewriting the entire git DAG to get linear granular commits out
<phry>
error: attribute 'extend' missing, at /nix/var/nix/profiles/per-user/root/channels/nixos/nixpkgs/nixos/modules/system/boot/kernel.nix:39:31 - I'm looking into that file right now :)
<phry>
clever: that was a wild ride, thank you! I'm trying it right now :)
rouma has quit [Remote host closed the connection]
joko has joined #nixos
<clever>
so, breaking it even worse, is the way to find the answer! :D
patrl has joined #nixos
rardiol has quit [Ping timeout: 240 seconds]
<phry>
*gets out the chainsaw*
rouma has joined #nixos
rardiol has joined #nixos
<{^_^}>
[nixpkgs] @meditans opened pull request #44380 → Add missing dependencies for swi-prolog → https://git.io/fN6Vt
MP2E has joined #nixos
Thra11 has joined #nixos
<{^_^}>
[nixpkgs] @Ericson2314 pushed 96 commits to staging: https://git.io/fN6Vg
<obadz>
ekleog: I won't have time to write anything for the foreseeable future (also wouldn't know where to start because it's been so long since I've touched dotnet). But if you have specific questions, feel free to put them here or on a github issue and I'll respond as soon as I can (apologies in advance for the delays)
<obadz>
ldlework: yes?
<ldlework>
obadz: well you kinda just answered my question :(
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to staging: https://git.io/fN6KV
kyren has quit [Client Quit]
dtulig has quit [Ping timeout: 260 seconds]
infandum has quit [Remote host closed the connection]
<jabranham>
I'm thinking about (re)trying nixos over the weekend. Anyone have any experience with Steam? Have you been able to get it up and running? Games run fine?
<infinisil>
jabranham: Yee, steam runs almost perfect, never heard anybody complain about it not working, and am using it myself
<manveru>
i got a few games that don't run fine :|
<manveru>
but majority works
<jabranham>
thanks for the feedback :-)
<clever>
infinisil: when i first got into nixos, i had a weird font problem with KSP
<phry>
okay, I've got a simpler question ^^ is it possible to get the .drv from a built package in /nix/store ? I think I remember that was possible but I can't remember where I read it -.-
<clever>
infinisil: a patchelfed KSP, the steam sandbox KSP, and even windows KSP under wine all had the identical bug!
<clever>
infinisil: oh, and originally, steam had trouble with my gpu drivers, but its been fixed long ago
<clever>
infinisil: now its just weird things like steam claiming the controller is not detected, yet the controller can ack the warning :P
<infinisil>
I also need some manual udev rule to get my steam controller working
<manveru>
for me, all the zachtronics games don't work still
<infinisil>
Alright, I guess it's not "almost perfect" hah
<phry>
manveru: clever thanks you too, both commands do great stuff xD
<clever>
infinisil: i'm using an xbox 360 controller, and the receiver does some funky things
<clever>
infinisil: the usb receiver will dynamicaly (de)register HID devices, based on the controller(s) being wirelessly connected
<clever>
infinisil: so the ev input device doesnt exist until i turn on the controller
<clever>
and some games only check for that at startup
<clever>
most of my other wireless usb devices (keyboard, mouse, headset) claim to always be present, even if the wireless link is broken
<aristid>
what's the option i can pass to nix / nix-build to make sure the package is rebuilt even if it's already in the cache?
<clever>
aristid: --check, and why do you want to rebuild?
rardiol has quit [Ping timeout: 240 seconds]
<aristid>
clever: a weird failure in one commit, but not in the cached output of the other
<clever>
was the source changed to fix the failure?
<aristid>
no
<clever>
so its an intermitent failure, that caused it to not actually fail the build?
<aristid>
i know about source hashing :)
<aristid>
clever: i don't know yet what's going on
<clever>
ah
<aristid>
ok, the --check actually does NOT give an error
<aristid>
so something else is going on? weird
ris has joined #nixos
aarvar has quit [Ping timeout: 268 seconds]
<aristid>
clever: lol, looks like i just did a git pull in my first nixpkgs directory just before the fix was pushed, and in the second nixpkgs directory just after the fix was pushed
<aristid>
gchristensen: 700 dollars a month is a lot of money
<clever>
aristid: lol
<gchristensen>
it isn't if it saves each developer 4hrs/day
<LnL>
there are no magical gnomes that populate the nixos cache for free
<LnL>
(that we know of)
<gchristensen>
haha
<ggVGc>
well
<ggVGc>
there should be :(
<aristid>
gchristensen: the EPYC seems to be cheaper than the Xeon from packet.net, so that's not what i mean. it's just i took the $1/hr (which doesn't _sound_ like a lot of money) and multiplied it by 24*31 and then the number sounded big, that's all :D
<drakonis>
not yet at least :V
<gchristensen>
ok dinner time, hopefully people find that article to be interesting.
<gchristensen>
aristid: I know, it is a lot of money
<gchristensen>
easy to justify though if your devs charge $60/hr+ and have long builds
<drakonis>
free gnomes with provider companies
<gchristensen>
back in a few hours
<drakonis>
are there any hardware provider companies deploying nixos yet?
<gchristensen>
packet.net
simukis has quit [Ping timeout: 240 seconds]
Denommus has quit [Remote host closed the connection]
philippD has joined #nixos
Anton-Latukha has quit [Quit: Leaving.]
<aristid>
gchristensen: i think the key is that packet.net donates the server time to nixos anyways. because otherwise, i would probably try to deal with the lower quality network and support and pay ~200€ for a EPYC server from hetzner :D
<{^_^}>
[nixpkgs] @Infinisil merged pull request #44365 → nixos/tomcat: correct type specification for virtualHosts → https://git.io/fN6Jk
ambro718 has quit [Quit: Konversation terminated!]
<makefu>
infinisil: i think it is not always a sane default. people have an easy way to change the behavior via the option which is fine imho
<infinisil>
Hmm.. What to use as defaults is always a hard question, and when to change them
<worldofpeace>
Noticing that any time someone mentions "is default behavior in x, x, and x" it's considered very harshly or with derision
<infinisil>
I mean, that is a good point, is it not?
<worldofpeace>
Well then we get into a discussion if x and x are sane anyway.
massimo_zaniboni has left #nixos [#nixos]
<worldofpeace>
I guess some people don't think so or need lots of convincing.
dbmikus has quit [Quit: WeeChat 2.2]
<makefu>
it may not be very clear when your ram runs out just because tmp is full. at work, literally nobody uses /tmp as tmpfs, some apps even require /tmp to be available forever (which is absoutely insane imho)
<samueldr>
conversely, I overshot when shopping for my workstation; I thought that all the building shenanigans I would do could eat so much ram... never hit more than 20GB from my 64GB :(
<worldofpeace>
mkaito: Thanks for you slaving on #44383 :)
<rain1>
I desire to no longer use "npm", is nix a tool that I may be able to use for that?
__monty__ has quit [Quit: leaving]
Acou_Bass has quit [Ping timeout: 260 seconds]
Acou_Bass has joined #nixos
<mkaito>
worldofpeace: pshhh only took ~40 hours of constant recompiling
<worldofpeace>
mkaito++
<{^_^}>
mkaito's karma got increased to 1
<worldofpeace>
mkaito++
<{^_^}>
mkaito's karma got increased to 2
bennofs has quit [Quit: WeeChat 2.0]
<mkaito>
<3
Fare has joined #nixos
martingale has quit [Read error: Connection reset by peer]
igo95862 has joined #nixos
ntqz has quit [Remote host closed the connection]
ntqz has joined #nixos
ntqz has quit [Excess Flood]
Sonarpulse has quit [Ping timeout: 268 seconds]
contrapumpkin has quit [Ping timeout: 240 seconds]
ntqz has joined #nixos
<pmade>
On 18.03 if I manually configure an interface its broadcast address is set to 0.0.0.0 instead the correct prefix + 255. I have prefixLength set to 24. Any idea how to fix that?