<{^_^}>
#16528 (by vrthra, 3 years ago, open): boo language fails to build on HEAD
georges has joined #nixos
camsbury has joined #nixos
<camsbury>
hey all
<camsbury>
trying to pull some nixos-hardware modules via fetchFromGithub... any idea how to select files in subdirectories individually for top level configuration.nix imports?
<camsbury>
goal is to get my wifi card working on my new xps 7590
<camsbury>
and there is a PR out for nixos-hardware, but its basically just an overlay for the driver and some common imports
<Gigabitten>
When I tried the first method of installation, I got a super weird error wherein, despite "arch" resulting in "x86_64," I got an error which seemed to be exclusively associated with having an incompatible CPU architecture (usually the result of trying to run a 64-bit program on a 32-bit processor).
<Gigabitten>
So I went with the second. When I ran nix-shell, it started... something that looks like compilation. It's been going for a few hours now. What on earth is going on here?
<Ashy>
Gigabitten: probably the best way to get a good overview is to read the first 2 chapters of nix pills: https://nixos.org/nixos/nix-pills/
evax has joined #nixos
<Ashy>
basically nix is a package manager and nix-shell is a way to open a temporary shell with a bunch of packages available without "installing" them globally like apt or yum would
<Ashy>
there's also a whole Linux distro built on top of the nix package manager called nixos
<Gigabitten>
hmm, well that's pretty dang neat but I can't say I understand how that's possible
<Gigabitten>
as for nixos, I'm aware of its existence
<Ashy>
and the package definitions are written in a programming language also called nix
<Gigabitten>
I had to type the channel name after all
<Gigabitten>
how the heck
<Gigabitten>
so much nix
<Ashy>
yeah, heh
<Ashy>
nix pills is worth a quick read of the first couple chapters
<infinisil>
Gigabitten: Did you add the binary caches as the post suggests?
<Ashy>
it's a good overview and explanation
<Gigabitten>
indeed I did, though I had to create the relevant directory
<infinisil>
Gigabitten: It might be the case that they purged the needed paths from the cache
<Gigabitten>
who is "they?" is this compilation the result of Russian interference with Nix?
<infinisil>
Hehe, the reflex people who made the cache
<Gigabitten>
ah, I see lol
<infinisil>
But Nix should ensure that you'll get the same result, even when building locally
<infinisil>
Just takes a bit longer
<Gigabitten>
my guess is a lot longer
<Gigabitten>
it also got down to having used about 2.5 GiB of disk space before going back up to pretty close to where I started, and I don't have a lot more than that free on this drive lol
<Gigabitten>
...hmm. I was waiting on it and it stopped after a bit and then suddenly popped up with about a billion lines, and now I can't get back to a state where all that doesn't pop up immediately. Indeed, I tried closing and re-opening my terminal, and renavigating to where I was before, and somehow nix-shell doesn't do anything anymore.
<Gigabitten>
probably it was the try nix thing that allowed me to get where I was
<Gigabitten>
ok lemme go back through some stuff
<Gigabitten>
er, try reflex
<infinisil>
Gigabitten: Ah, you should set up your shell to source ~/.nix-profile/etc/profile.d/nix.sh automatically
<infinisil>
With .bashrc or so
<infinisil>
Or .profile
<Gigabitten>
ah I see
<Gigabitten>
not gonna lie, I know how to use .bashrc to set aliases and not terribly much else
jb55 has quit [Remote host closed the connection]
jb55 has joined #nixos
<Gigabitten>
but it looks like all I have to do is just place "source *long path*" in there
<Gigabitten>
so I should be fine
<infinisil>
Yeah, it's really just a file that gets sourced when bash starts, so you can put any bash command there (such as this `source` command which sources yet another file)
<Gigabitten>
"no manual entry for source" blegh
<Gigabitten>
from what I see from a search, would it just cause it to run nix.sh when I start up a bash environment?
<{^_^}>
[nixpkgs] @disassembler pushed 3 commits to master: https://git.io/fjNyn
<{^_^}>
[nixpkgs] @thefloweringash opened pull request #67374 → nmap: update clang specific patch → https://git.io/fjNyl
<iqubic>
Finally the latest versions of the nixos linux kernel modules work with ZFS. Those pesky spl issues are gone.
martyet-o has joined #nixos
vika_nezrimaya has joined #nixos
palo1 has joined #nixos
palo has quit [Ping timeout: 245 seconds]
palo1 is now known as palo
<{^_^}>
[nixpkgs] @lheckemann opened pull request #67375 → Merge openssl 1.1 into staging → https://git.io/fjNy2
UltraPhil has quit [Quit: Leaving]
suppi has joined #nixos
pjan has joined #nixos
___laika has joined #nixos
domogled has quit [Ping timeout: 244 seconds]
___laika has quit [Ping timeout: 248 seconds]
domogled has joined #nixos
bakakuna has joined #nixos
<{^_^}>
[nixpkgs] @oxij opened pull request #67376 → nixos: zsh: add more helpful documentation into generated files → https://git.io/fjNy5
drakonis has quit [Ping timeout: 244 seconds]
ashkitten has quit [Ping timeout: 276 seconds]
ashkitten has joined #nixos
bakakuna has quit [Ping timeout: 245 seconds]
domogled has quit [Ping timeout: 245 seconds]
orivej has joined #nixos
slack1256 has joined #nixos
orivej has quit [Ping timeout: 258 seconds]
Gigabitten has joined #nixos
avn has quit [Ping timeout: 246 seconds]
endformationage has quit [Quit: WeeChat 2.5]
oborot_ has quit [Ping timeout: 248 seconds]
oborot has quit [Ping timeout: 272 seconds]
dansho has quit [Ping timeout: 245 seconds]
<{^_^}>
[nixpkgs] @vcunat pushed 1000 commits to staging-next: https://git.io/fjNS6
nadley has quit [Ping timeout: 252 seconds]
dansho has joined #nixos
<Gigabitten>
So, independently of other stuff, I tried just installing the nix package manager on its own
<Gigabitten>
I'm on lubuntu. erm, it's not terribly new, but not more than a year old.
<Gigabitten>
Whatever I do, when I try to use nix-shell, it catapults into a whole hell of a lot of stuff that I'm pretty sure doesn't typically need to be done.
<srhb>
Gigabitten: Such as?
<Gigabitten>
I'll throw in a pastebin of the first screenful.
<Gigabitten>
This just goes until I run out of my pitiful few remaining gigabytes of disk space.
<EsperLily>
surely it only does this once for `nix-shell -p rogue`, no?
<Gigabitten>
No.
<EsperLily>
well if you run out of disk space maybe it's not managing to store them properly
<EsperLily>
in any case it looks like it's grabbing sufficient dependencies to compile rogue
<Gigabitten>
ah, okay.
<EsperLily>
The "warning: substituter 'https://cache.nixos.org' does not have a valid signature for path" lines are curious. I don't know if this is a problem
<srhb>
I wouldn't expect it to actually have to build rogue, as long as this is not some strange architecture
<srhb>
So I'm wondering if your substituters are setup wrong
<emily>
sounds like the trusted keys might be broken
<srhb>
nix show-config please?
<emily>
like, it's ignoring the binaries from the cache because the signatures are considered invalid?
<Gigabitten>
ah yeah the ryan trinkle thing is from the bigger purpose of getting nix running
<srhb>
I wonder if you maybe used reflex-platform installer and it ate the nixos key
<EsperLily>
Are you on NixOS or otherwise? If you're on NixOS then you need to edit your configuration.nix for this
<srhb>
It's lubuntu
<EsperLily>
ok
<Gigabitten>
I indeed used the reflex-platform installer.
<srhb>
Gigabitten: I've seen this happen before with it. I think there's some edge case where it ends up nuking the nixos cache, which is bad.
<Gigabitten>
oh geez
<Gigabitten>
will adding the nix key fix it
<srhb>
Yes :)
<Gigabitten>
excellent
<srhb>
But let us know anyway ;-)
<Gigabitten>
oh, what's the syntax for this? do I separate them with a comma or add a new binary-cache-public-keys entry?
<srhb>
space :)
<Gigabitten>
ah
<Gigabitten>
that'd also make sense
<srhb>
I don't know if it does, but that's the syntax regardless. :P
<Gigabitten>
whaddya know
<Gigabitten>
that did indeed fix it
<srhb>
Great!
<Gigabitten>
Do the Reflex people know about this? er, team obsidian or something?
<Gigabitten>
obsidian systems
<srhb>
Obsidian Systems. I don't know. Did you install nix first and then ran reflex-platform installer, or did you let it install Nix for you?
<Gigabitten>
I tried letting it install nix for me, and that didn't work. So instead I nuked my install of Nix and tried to install just Nix alone beforehand. Then I came here.
<srhb>
Hm. An issue with your steps might indeed help them find the issue :)
<Gigabitten>
I was also here earlier today trying to work out the same issue, but it would seem that nobody at the time had the particular knowledge I needed for this very easy fix.
<EsperLily>
since that's trying to modify an existing line
<srhb>
Gigabitten: Oh, did it leave a backup file?
<EsperLily>
though at first glance it looks like that should just append to the end of the line
iyzsong has joined #nixos
<EsperLily>
which is to say, it looks like that should work too
<Gigabitten>
oh, yes, there probably is a backup or two since I've been using emacs to add the keys in
<srhb>
Maybe it's something silly like it defaults to the hardcoded cache.nixos.org-1 key unless the option is specified (either at all or with a non-empty value) and once specified by reflex, the hardcoded one is no longer taken into consideration
<Gigabitten>
nothing terribly interesting though, just the two lines which the reflex tutorial I was using instructed me to add manually. before I put those there, the file didn't yet exist. It's been some hours now, with some other activities between, so I don't remember, but it's possible I did things a bit out of order and it failed to update things properly when it saw there was already something there.
<srhb>
Gigabitten: And the script didn't leave a backup in /etc/nix ?
<Gigabitten>
nope, just configuration.nix, nix.conf, and nix.conf~
<srhb>
Oh well.
suppi has quit [Ping timeout: 244 seconds]
<Gigabitten>
it wouldn't be hard to try to replicate
<Gigabitten>
lemme see if I can get it
<srhb>
Gigabitten: If you're up for it I'm sure it would improve the bug report.. :)
<Gigabitten>
since I know how to make it work I'll just nuke all the nix stuff I know of off my drive - it's actually said in the installation instructions for nix that you should just be able to `sudo rm -rf /nix` to uninstall it, so I'll do that
<Gigabitten>
*for the nix package manager
<srhb>
That's not quite true though
<Gigabitten>
What else should I clear out?
<srhb>
(There's still stuff in /etc/nix, various /etc/profile /etc/bashrc etc., /home/foo/.nix*)
<{^_^}>
[nixpkgs] @paulreimer opened pull request #67377 → vimPlugins.float-preview.nvim: init at 2019-04-07 → https://git.io/fjNSd
<srhb>
Depending on which installer you use (single or multi-user) you will need to clear out more
<Gigabitten>
hmm
<srhb>
For single user I think it should be enough with /nix and /etc/nix
<Gigabitten>
I didn't sudo at any point in the installation so I'm assuming I used the single-user installation process
<srhb>
if that has a process, you're on multi-user
iyzsong has joined #nixos
<srhb>
Alternatively nix-info :)
<Gigabitten>
nothing gets returned as a result
<srhb>
OK, single user it is then.
<Gigabitten>
nice
<Gigabitten>
okay, I've got that cleared, and I'll also comment out my modifications to .bashrc
<Gigabitten>
and grab a new terminal
<srhb>
You can leave those
<Gigabitten>
oh
<Gigabitten>
alrighty
<srhb>
The installer is smart enough to recognize those :)
<srhb>
/etc/nix and /nix should suffice
<Gigabitten>
okay then, I'm going to go back through the installation process from reflex real quick
<srhb>
Cool :)
knupfer has joined #nixos
<Gigabitten>
Okay, so the specific thing I did was, I clones the git repository then did ./try-reflex and it didn't work last time
M-Gregoire49 has joined #nixos
<srhb>
Interesting bit is whether it leaves a nix.conf in place.
<Gigabitten>
IRL problem
M-Gregoire49 has quit [Client Quit]
<Gigabitten>
I had a cup with water and a cup with something else
<Gigabitten>
I tried to stack the cup with something else into the cup with water and had to get a towel for the resulting mess
<Gigabitten>
it's too late in the night for this shit
<Gigabitten>
ok anywaay
<Gigabitten>
ok anyway
M-Gregoire49 has joined #nixos
<Gigabitten>
I remember this part
<Gigabitten>
"Add binary caches for reflexes to /etc/nix/nix.conf ? "
<Gigabitten>
"yes"
<Gigabitten>
1) Yes // 2) No // 3) Ask again next time
<Gigabitten>
#?
<Gigabitten>
yes
M-Gregoire49 has quit [Client Quit]
<Gigabitten>
#?
<srhb>
1, presumably? :P
<Gigabitten>
OH DAMN IT
<srhb>
lol
<Gigabitten>
okay I expect that this is where it went wrong
<srhb>
Probably :)
<Gigabitten>
when I failed to get past that I ctrl+c'd out, probably got distracted and ended up skipping figuring that bit out
<Gigabitten>
I'm going to see what happens in both the case where I do this right and the case where I do it wrong, to be sure
ottidmes has joined #nixos
<srhb>
Cool :)
<Gigabitten>
Okay - pressing 1, that looks like a correctly configured nix.conf to me
<Gigabitten>
so I'll nuke it once more and see what happens when I ctrl+c out
<Gigabitten>
Indeed, /etc/nix doesn't even exist
madhukar93 has joined #nixos
o1lo01ol1o has joined #nixos
<madhukar93>
Hi, I'm on nixos-unstable. On `sudo nixos-rebuild switch --upgrade` the default generation in the grub menu isn't pointing to the newest generation any longer
<Gigabitten>
alrighty, that's fairly conclusive. the main problem seems to be that it's not idiot-proof, though I will contend that it'd be better to accept "yes," "Yes," "y," and "Y" there, especially since the numbers for the options don't even show until you fail the first time
<madhukar93>
it's been this way for a while though
ng0 has joined #nixos
<madhukar93>
Also I want to speed up boot from hibernate, so I would like to not be presented with a choice of generation to boot into only if I'm waking up from hibernate. Would appreciate pointers with this
o1lo01ol1o has quit [Remote host closed the connection]
jgeerds has joined #nixos
orivej has joined #nixos
freeman42x has joined #nixos
o1lo01ol1o has joined #nixos
<srhb>
Gigabitten: Still not sure how you ended up with reflex key but not nixos key though
<srhb>
Gigabitten: There must be some real error when doing normal nix install, then try-reflex, I figure?
<srhb>
madhukar93: Are you out of space on your boot partition?
<Gigabitten>
oh no I know the specific reason
<Gigabitten>
it's because I canceled the installation and manually put the key for reflex because it was in the tutorial I was using to do so despite being able to be done automatically
<srhb>
Gigabitten: Ah, I see. :)
<Gigabitten>
uh, side question, how much disk space does reflex stuff take up?
<srhb>
Gigabitten: Well, usability issues are certainly worth an issue too, I feel
<Gigabitten>
yeah I suppose so
<srhb>
Gigabitten: Haskell is generally a otn of space, but we can find out more precisely..
alanz_ has joined #nixos
<srhb>
Let me just update my platform.
semigroupoid has joined #nixos
alanz_ has quit [Client Quit]
<{^_^}>
[nixpkgs] @vbgl opened pull request #67379 → ocamlPackages.cstruct: 3.1.1 -> 4.0.0 and related updates → https://git.io/fjN9L
o1lo01ol1o has quit [Ping timeout: 245 seconds]
alanz_ has joined #nixos
alanz_ has quit [Client Quit]
<Gigabitten>
I ask because as it does stuff it's hovering around 2/3 of a gigabyte left on disk setting up reflex for use lol
o1lo01ol1o has joined #nixos
<semigroupoid>
Hi does anyone here have experience with setting up a haskell environment? I can't seem to install any packages for ghc
<srhb>
Gigabitten: Looks like it places a gc root to the final nix-shell out in gc-roots/shell.out, we will be able to deterine the size of the runtime from that, and the size of the build env from the similar .drv root
ottidmes_ has joined #nixos
<srhb>
Gigabitten: Nix and especially nix + haskell is really not a low-disk-usage thing :P
<Gigabitten>
well, I guess I should have been more prepared probably
<srhb>
fwiw there are some things that help
orivej has quit [Ping timeout: 245 seconds]
knupfer has quit [Ping timeout: 264 seconds]
<srhb>
Placing the /nix/store on a compressed partition and allowing hardlink optimisation are great
<Gigabitten>
honestly
<srhb>
My nix store currently has a compression ratio of 2.18
<Gigabitten>
it's not that I don't have space anywhere, it's that I don't know how to put this stuff on my hard disk drive
<srhb>
Gigabitten: OK, built mine. Runtime env is 9.6G and build env is 142.6M
<Gigabitten>
oh god
<Gigabitten>
*ctrl+c*
<srhb>
:P
ottidmes has quit [Ping timeout: 245 seconds]
<srhb>
fwiw I bit the bullet long ago, my store is currently 184GiB compressed. >_>
<srhb>
Garbage collection is for people who like to sit around and wait...
jgeerds has quit [Ping timeout: 258 seconds]
<srhb>
You can nix path-info -Sh ./gc-roots/* yourself if you're curious, but only after the fact of course...
<Gigabitten>
I'm gonna need to figure out how to put this on sda1, a slow but usable terabyte hard drive with about a tenth of that free
<Gigabitten>
my 128 gb SSD just can't take it
<tilpner>
I didn't read all the context, but why can't you put it on your SSD?
<srhb>
:/
<Gigabitten>
honestly I just don't know how
<Gigabitten>
that's the only reason
<Gigabitten>
I'll figure something out lol
<tilpner>
I've used NixOS from an SSD and from a laptop HDD, and HDD is just painful
<Gigabitten>
I just want the package manager here
<Gigabitten>
and nix-shell
<srhb>
All you need to do is put /nix/store on something that has enough space for you.
<srhb>
Or just /nix itself
<Gigabitten>
huh
<tilpner>
Oh, you're not on NixOS!
<tilpner>
Well, then it might be fine, because your OS will still be on SSD
<Gigabitten>
look, I may have chosen Lubuntu hastily but I've got it set up the way I like
<Gigabitten>
switching costs are real
<Gigabitten>
:p
<tilpner>
No, it's fine, I just didn't realise from skimming half a screen of backlog
<Gigabitten>
heh
<Gigabitten>
understandable
<Gigabitten>
Okay, so, what gets put in /nix/store?
<Gigabitten>
Presumably lots of stuff.
<srhb>
Gigabitten: Everything, basically! every package, every build.
o1lo01ol1o has quit [Ping timeout: 246 seconds]
<Gigabitten>
nice.
<srhb>
Gigabitten: So all your haskell stuff will live there. gc roots, profiles, the database and some other housekeeping will live in ther places in /nix but will take up almost no space
<srhb>
(my /nix/var is 407M, compared to my /nix/store of 184G)
<Gigabitten>
Well I suppose I only need move it, then. How do I do that? I've got a dead link, a stackexchange dissertation from someone who, at a glance, didn't seem to solve the problem, and an article from 2014 that looks like it was written in 1993 based on the layout.
<Gigabitten>
So I figure I'll just keep probing here for answers.
<srhb>
In general you'd want to mount some partition on /nix
<Gigabitten>
That I can do.
alanz_ has joined #nixos
<Gigabitten>
Once I've got that mounted, how do I link it up?
<srhb>
Just move everything over.
<madhukar93>
@Gigabitten: doesn't seem that way `/dev/nvme0n1p3 510M 195M 316M 39% /boot`
<srhb>
madhukar93: Did you switch boot loader at some point?
<madhukar93>
oh sorry, I meant to reply to you
<madhukar93>
no I haven't
<srhb>
I figured :)
<srhb>
Rollback?
<madhukar93>
I did switch from 19.03 to unstable at that point, but that's not when this started happening
<srhb>
If none of those, I'd reinstall the bootloader with nixos-rebuild switch --install-bootloader, but take note that this is potentially dangerous and you should definitely have a live usb ready to unbreak your system.
slack1256 has quit [Ping timeout: 245 seconds]
M-Gregoire49 has joined #nixos
<madhukar93>
oh boy. Well this is a weekend
<semigroupoid>
alexarice: Do you know how I can make the package discoverable by HIE?
<srhb>
madhukar93: Heh, yeah..
hamishmack has joined #nixos
<Gigabitten>
So if I make an fstab entry like `/dev/sda1 /nix/hdd ntfs defaults 0 0` then that'd do? Does nix just notice I've got something mounted there and automatically start storing stuff there?
<Gigabitten>
(will it being ntfs be an issue? I hope not, but I could partition the hdd)
M-Gregoire49 has quit [Client Quit]
M-Gregoire49 has joined #nixos
<srhb>
Gigabitten: I definitely think that will not work
<Gigabitten>
nice
<Gigabitten>
ok so what will work?
o1lo01ol1o has joined #nixos
<srhb>
Most linux partitions.
m0rphism has joined #nixos
<Gigabitten>
haha yep I just tried the dumb way and it started eating disk space very quickly
<srhb>
ext, zfs, xfs..
<Gigabitten>
alrighty, gonna go do that real quick
<srhb>
Also it should be something like /dev/sda1 /nix, not /nix/hdd
<Gigabitten>
oh I see
<srhb>
The easiest approach is to nuke /nix first, then mount a new partition, then install Nix again
<srhb>
Otherwise you'll have to manually move things over.
<Gigabitten>
ohhhhhhhhhhh
o1lo01ol1o has quit [Remote host closed the connection]
<alexarice[m]>
semigroupoid: I think you could either install the package globally or use direnv/lorri/something similar
<alexarice[m]>
though I don't have much experience with that
<srhb>
If you already have a data partition, I believe a bind mount may work too, but I'm not completely sure. :)
<srhb>
(I'm also not sure it's worth the trouble even if it does work)
<Gigabitten>
I see exactly what's going on here now
<Gigabitten>
learning new stuff about how you can use mounting
<srhb>
:)
<tilpner>
Gigabitten, srhb: You probably don't want /dev/sda1 if you have multiple devices. Prefer /dev/disk/by-*
<Gigabitten>
ok so looking up what a bind mount is, what happens if I make a recursive bind mount
<Gigabitten>
does my computer catch on fire or what
<Gigabitten>
eh, never mind, I couldn't possibly remove it from the rest of the filesystem to do the confusing, useless garbage I had planned
<semigroupoid>
alexarice: That's what I did but for some reason HIE doesn't find certain packages. Thanks for the help anyways
<Miyu-chan>
Don't make recursive bind mounts. I accidentally deleted my / because of that lmao
<srhb>
Gigabitten: Not sure what a recursive bind mount is, but it sounds fun :P
<srhb>
tilpner: Good point :)
<alexarice[m]>
semigroupoid: when I say install globally I don't mean install the package I mean install it with `ghcWithPackages`
<Gigabitten>
lmao miyu I didn't think it would actually mess it up that badly
<semigroupoid>
whoa. you mean install hie inside `ghcWithPackages`?
<Miyu-chan>
TL;DR. I was just having my merry way, and `rm -fr /old-root`.
<madhukar93>
srhb: `sudo nixos-rebuild switch --upgrade --repair --install-bootloader` didn't work either. The point at which it started happening doesn't seem special. Maybe I can truncate older generations and it would be forced to reset or something
o1lo01ol1o has joined #nixos
<Miyu-chan>
Turns out that / is mount bound to /old-root/nixos.
o1lo01ol1o has quit [Remote host closed the connection]
<Miyu-chan>
Now I get why they don't allow directory hard links.
<srhb>
madhukar93: I've only really experienced this when I accidentally had two bootloaders, and I was booting using the one NixOS was no longer updating
<srhb>
madhukar93: Is this GRUB/MBR or something else?
o1lo01ol1o has joined #nixos
<madhukar93>
New generations are getting added, just that the default generation isn't the new one for some reason. It's stuck at one from 12 August
<srhb>
Ooh.
<srhb>
Sorry, i didn't realize that.
<madhukar93>
sorry I wasn't clear
<clever>
madhukar93: is /boot mounted? do you happen to have multiple boot partitions? what is in the /boot directory of / (un-mount /boot to see)
<srhb>
madhukar93: and which bootloader is it?
<madhukar93>
It's grub only. Haven't configured it myself
<madhukar93>
New to nixos, mostly running defaults
<srhb>
madhukar93: What does `mount | grep boot` say?
<srhb>
`df /boot` is probably better actually
<srhb>
(We're interested in the filesystem being used)
hamishmack has quit [Ping timeout: 245 seconds]
ee1943 has joined #nixos
<semigroupoid>
alexarice: I moved it inside ghcWithPackages but it still didn't work.
<semigroupoid>
Is there any other LSP for haskell?
<ivan>
oh, I figured out the cause of Hydra "Jobset contains a job with an empty name. Make sure the jobset evaluates to an attrset of jobs." after trying everything for several hours the other day
<ivan>
I typed ", lib" into the inputs for the release expression
<Athas>
I'm trying to do a 'nixos-rebuild switch', but I get the following message:
<Athas>
error: Package ‘cachix-0.2.1’ in /nix/store/qysm29q8mz84nmiyvbynadspr4c1a5r6-nixos-19.09pre190337.8f84220eb29/nixos/pkgs/development/haskell-modules/hackage-packages.nix:45806 is marked as broken, refusing to evaluate.
o1lo01ol1o has joined #nixos
<Athas>
It then tells me how I can work around it by enabling broken packages, but what does it even mean for cachix to be "broken"? Did a human set that somewhere? Can I figure out how or why? I can't find anything in the nixpkgs repository.
zfnmxt has quit [Quit: Bye!]
Soo_Slow has joined #nixos
<{^_^}>
[nixpkgs] @c0deaddict opened pull request #67393 → nixos/transmission: make user configurable → https://git.io/fjNd8
o1lo01ol1o has quit [Ping timeout: 248 seconds]
zfnmxt has joined #nixos
zfnmxt has quit [Client Quit]
<Miyu-chan>
Is there a more ergonomic version of readDir?
Chiliparrot has joined #nixos
<srhb>
Athas: It's a haskell package, so they're treated specially. Normally, yes a human does mark it as broken. In the case of the autogenerated hackage set, a separate CI job runs on Hydra for huge swathes of Hackage updates. Packages that fail to build (with the automatic update) are also automatically marked as broken, iirc.
<srhb>
Athas: The full and correct story is probably somewhere in the hackage2nix docs
<srhb>
Specifically, they're added to configuration-hackage2nix.yaml under the broken-backages: list
<srhb>
And then the resulting hackage-packages.nix file contains the meta.broken = true attribute
iyzsong has quit [Read error: Connection reset by peer]
<srhb>
Hm, I'm actually not sure where the scripts that I assume exist to mirror results of builds into configuration-hackage2nix.yaml... I'd be surprised if it really is manual though.
<srhb>
peti or basvandijk would know
<{^_^}>
[nixpkgs] @Ma27 opened pull request #67394 → nixos/treewide: drop dependencies to `keys.target` → https://git.io/fjNdg
<Miyu-chan>
I don't have nixpkgs.lib for this. Does anyone think this can be done better?
zfnmxt has joined #nixos
<srhb>
Athas: That said, I know it's safe to manually modify (so that the next regeneration takes into account now-working packages) -- so maybe it is an iterative and manual process.
<ivan>
apparently Hydra does not think this is a local store store_uri = file:///var/lib/hydra/cache?secret-key=/var/secrets/hydra/secret_key
<Miyu-chan>
joko: Well, that's 2 people that want it now. :)
<joko>
Miyu-chan: you mean this programmatic restrict-mode check? :)
<joko>
ivan: why not store-uri = file:///nix/store?secret-key=blabla?
<joko>
(warning: my Hydra deployment must be a bit old)
<m1crOman>
How do you deal with the likes of syslog + nscd that require bind to be online to resolve their target hosts? My boot is stalling because of it and I really don't want to hard code the IPs. Tried adding requiredBy rules to bind to no avail
<ivan>
joko: I don't think it can work like that
<ivan>
anyway I added an option so it'll probably work now
<ivan>
# Convince Hydra's isLocalStore that yes, we do have a local store we want to serve
<ivan>
server_store_uri = local
shibboleth has joined #nixos
<ivan>
oh, weird, Hydra _can_ put things like /nix/store?
<Athas>
srhb: is cachix particularly critical? I don't specify it myself in my configuration.nix AFAIK, so I'm a bit concerned that it is holding back my upgrade.
<{^_^}>
mozilla/nixpkgs-mozilla#91 (by flosse, 1 year ago, open): How to use rust overlay with additional targets?
<ivan>
I just had no clue what to do, now I have a hint :-)
<pie_>
oh its nixos-option i was trying config
<hpfr[m]>
Anyone have problems with pkgs.hplipWithPlugin? I tried running hp-setup after including that in my services.printing.drivers and I get something about python-dbus not being installed.
<joepie91>
right :P
ilmu has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<joepie91>
is audio in WINE broken for anyone else?
<joepie91>
hm. actually, it works in winecfg, just not in UT2004 and something else I tested a few weeks back?
<vika_nezrimaya>
Trying to set up uwsgi with several mountpoints. The first one loads successfully, but the second one crashes uWSGI with Fatal Python Error: initsite: failed to import the site module
cbarrett has joined #nixos
<dahirsch>
when i enable gpgagent gpg itself is not added as system package. is it a good idea to add that to the gpgagent module?
casaca has joined #nixos
<vika_nezrimaya>
dahirsch: Feels reasonable :3
<pie_>
tilpner: i figured it out, thanks <3 , i was making a really stupid mistake
<pie_>
tilpner: the problem was that im working on a local checkout but i prepended the existing NIX_PATH to my local path
<pie_>
(i didnt used to do that)
<pie_>
so i was working on the wrong stuff, i dont know how that resulted in this but yeah
knupfer has quit [Ping timeout: 264 seconds]
george has quit [Quit: leaving]
<pie_>
tilpner: ugh i really appreciate this, i wish i didnt have so many problems with stupid simple things like this
george has joined #nixos
<tilpner>
pie_: Syntax errors are the quickest way to tell if the file you're editing even matters. Why do you think I know that? c.c
oborot_ has quit [Ping timeout: 272 seconds]
oborot has quit [Ping timeout: 272 seconds]
jgt has joined #nixos
selfsymmetric-pa has joined #nixos
zupo has joined #nixos
<pie_>
tilpner: i know but i didnt think of this :/ maybe ill rememebr next time
selfsymmetric-pa has quit [Remote host closed the connection]
<pie_>
tilpner: another problem is i had it cached in my head that this _is_ getting evaluated
<pie_>
because it_was_ getting evaluated at one point, before i chanced the nxi_path
<tilpner>
Cache invalidation is hard :)
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
drakonis has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<betaboon>
hello all, my nix-store is huge even after nix-collect-garbage -d. any pointers where/what to look for ?
<GiGa>
Hello, having problems building my system again :'(
<srhb>
GiGa: What's up?
<GiGa>
Dependencies cannot be built, but I can't tell which packages they're for.
<GiGa>
It's not a 0 file size issue this time
<GiGa>
I suspect this one is the real culprit: "builder for '/nix/store/jb9k9spji6k37zxl7cbpi1pyn3iw2hgk-nvidia-x11-390.129-4.19.67.drv' failed with exit code 1"
fresheyeball has quit [Quit: WeeChat 2.4]
phdoerfler has quit [Quit: Leaving.]
<srhb>
GiGa: That's just a summary message, the real error is above.
<GiGa>
Ah plop, I can't scroll up as I'm in a screen session
<GiGa>
I also don't know why that always seems to mean I can't scroll up
Myhlamaeus has quit [Ping timeout: 246 seconds]
<srhb>
I'm pretty sure screen has scrollback
<GiGa>
Just running "nixos-rebuild switch" outside of screen
noudle has joined #nixos
<GiGa>
srhb: I think it should have, just not on my system for $reasons
<srhb>
Try ctrl+a->esc->arrow up
<srhb>
I'm guessing you just haven't integrated it with your terminal emulator yet..
<GiGa>
possibly issues patching Nvidia driver?: "1 out of 2 hunks FAILED -- saving rejects to file kernel/common/inc/nv-list-helpers.h.rej"
<GiGa>
?
<srhb>
Yup.
<srhb>
That'll do it.
<{^_^}>
[nixpkgs] @Alkeryn opened pull request #67404 → init: serialdv at 1.1.1 → https://git.io/fjNA0
<srhb>
GiGa: Did you make alterations to the kernel you use or the nvidia drivers?
pie_ has quit [Excess Flood]
<GiGa>
srhb: Not to my knowledge. I've moved home, so I've set up a new initfs crypt. I've also added swap (so I can hybernate). I did initially say "switch --upgrade" but that failed so I figured I'd accept just a switch of config (running unstable).
pie_ has joined #nixos
<srhb>
GiGa: --upgrade is just short for nix-channel --update
freedan42x has quit [Remote host closed the connection]
<srhb>
So that has already happened and you're on a commit that's bad with your config already.
<GiGa>
srhb: (Also ^A > Esc > cursors works.)
<GiGa>
Didn't know that, so presumably I don't need to do the "nix-channel --update" first?
<srhb>
GiGa: Not understood, first what?
<srhb>
As in, before what?
<GiGa>
srhb: Normally I do "nix-channel --update" then "nixos-rebuild switch --upgrade" to upgrade system packages.
<srhb>
Ah, right, no, both are not necessary
noudle has quit []
* GiGa
notes that for future
<GiGa>
As to the current problem, I really need to apply the new config for my change of disk (/home is on a failing disk, I've just installed a new one). Thoughts?
<srhb>
Roll back your channel to the previously working one
<srhb>
Then rebuild switch
<GiGa>
How please?
<srhb>
sudo nix-channel --rollback
<srhb>
Also man nix-channel is helpful :)
oborot has joined #nixos
o1lo01ol1o has joined #nixos
<GiGa>
Point taken ;)
oborot_ has joined #nixos
<GiGa>
That's now building so we'll see what it says.
<GiGa>
I might have specified my crypt wrongly, we'll soon see!
<srhb>
GiGa: btw, it's probably a good idea to not upgrade and change config in the same motion unless you have to. makes things easier to fix and debug :)
<srhb>
However, I can confirm the bug with nvidia locally.
<srhb>
Or I can try :-)
fendor_ has joined #nixos
<GiGa>
I didn't think that Nvidia driver was being updated any more, so was surprised there was a change.
<GiGa>
Does mean, however, that I can't upgrade my packages as that one will always fail :'( .
<GiGa>
srhb: noted about upgrade and change.
chiiba has quit [Ping timeout: 245 seconds]
fendor__ has quit [Ping timeout: 248 seconds]
<srhb>
GiGa: Why wouldn't it be updated anymore?
slack1256 has joined #nixos
<GiGa>
I was under the impression it was the legacy driver
<srhb>
Oh. Yeah, you're probably in bad shape then :P
<srhb>
Probably just need to adjust the patch though :)
<GiGa>
srhb: Definitely not something I'm qualified to do. Odd given it used to work ...
<srhb>
Presumably either it or the kernel got updated, and the patch no longer fits.
<GiGa>
Urgh, still problems even with just a rebuild switch
<srhb>
What's the problem now?
<GiGa>
Android studio, one dependency couldn't be built
<GiGa>
I could roll back again?
<srhb>
If you didn't just add that to your config, and it used to work, yes.
o1lo01ol1o has quit [Remote host closed the connection]
<GiGa>
Can confirm that's also not a new addition
<GiGa>
and I can totally see how changing config, testing and _then_ upgrading would be easier than my current predicament.
<srhb>
Indeed. :)
o1lo01ol1o has joined #nixos
samrose has joined #nixos
<srhb>
I wonder how you might have ended up with (at least) two channel bumps without a succesful build too..
o1lo01ol1o has quit [Remote host closed the connection]
fnords has quit [Quit: i have seen the]
o1lo01ol1o has joined #nixos
<GiGa>
srhb: make that more than two. Nvidia just failed to build again
fnords has joined #nixos
<srhb>
You might want to visit /nix/var/nix/profiles/per-user/root/channels and decide on a sufficiently old version that you know is good.
ryantrinkle has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
<GiGa>
I'm not sure how I'd determine a version that was good
<GiGa>
Unless I can determine what my system is currently running with?
<srhb>
You can check your booted version
<srhb>
Or actually, your running one will be good too, so nixos-version
<GiGa>
"19.09pre185259.362be9608c3 (Loris)"
<srhb>
GiGa: grep for that in each channels symlink's manifest.nix
<srhb>
Just the final number should suffice, that's the commit sha1
<srhb>
(that is, grep 362be9608c3 /nix/var/nix/profiles/oer-user/channels*/manifest.nix)
Gigabitten has joined #nixos
<GiGa>
OK, I'm guessing that "channels-23-link" gives me the number to roll back to?
<srhb>
That's correct.
<srhb>
(really, it's just the channels -> channels-23-link symlink that will be created)
<evanjs>
jtojnar: Re https://github.com/NixOS/nixpkgs/pull/67115 - I added wrapGAppsHook as postman uses gnome3, etc, but I'm still a little unsure if it's necessary at all, as it runs fine without it, and I noticed we still don't have it documented per #16285. Any rule of thumb on whether it should be used or not?
<evanjs>
jtojnar: per the warnings I'm getting, yeah, GTK3 for the interface. I wasn't sure if it was necessary as the themes are independent of my GTK theme settings, and it doesn't appear in dconf, d-feet, etc. I'll leave it in as it uses GTK3 for the UI. Thanks!
o1lo01ol1o has quit [Remote host closed the connection]
xkapastel has quit [Quit: Connection closed for inactivity]
o1lo01ol1o has joined #nixos
wildtrees has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
shibboleth has joined #nixos
o1lo01ol1o has joined #nixos
GiGa has joined #nixos
<GiGa>
srhb: It built fine, unfortunately when configuring /home's mount it seems I'm pointing at an LVM2_Member
<GiGa>
So the system keels over
lally has joined #nixos
<lally>
Hi, if anyone's about. I'd love to figure out how to enable the 'kvm' feature on my local nix install in ubuntu
<lally>
I don't have (or just can't locate) the global configuration.nix file that a proper nixos install would have
<GiGa>
lally: Sorry, I've only ever used Nix within Nixos, never atop another OS.
<lally>
GiGa: thanks, np. Any idea where I should look in the docs?
o1lo01ol1o has quit [Remote host closed the connection]
<GiGa>
lally: I don't even know if you can enable kvm in that scenario (not my area of expertise).
<srhb>
GiGa: An LVM2 member?
<srhb>
GiGa: As opposed to.. What's it supposed to be?
<GiGa>
lally: Bear in mind I've not read this through, but on https://nixos.org/nix/manual/#chap-installation I searched for "kvm" and found this config line: "requiredSystemFeatures = [ "kvm" ];" If that helps?
<srhb>
lally: Features are just arbitrary options in nix.conf, so you can just add eg. system-features = nixos-test benchmark big-parallel kvm
<srhb>
lally: The building user must be able to use the kvm device node.
<GiGa>
srhb: Well, it is an lvm2 volume (LUKS --> LVM --> my home partition), but I can't work out what to say in configuration.nix to get the LVM partition mounted to /home
<lally>
In a non-nixos install, where's nix.conf?
<srhb>
lally: /etc/nix/nix.conf if it exists, otherwise create it.
<{^_^}>
[nixpkgs] @duhirsch opened pull request #67409 → nixos/gnupg: Add gpg to systemPackages → https://git.io/fjNxl
<GiGa>
srhb: Certainly is. If I boot an old config I can luksOpen and authenticate
<srhb>
GiGa: If you mount it by hand you can try running nixos-generate-config (with the relevant options to not clobber your existing stuff)
<srhb>
GiGa: Not sure it works, it certainly breaks for zfs on luks.. Then again, if you know what fstab should look like it really should just mirror that
<srhb>
GiGa: You seem to have a dash where lvm reports a slash in /dev/aug2019home-home -- is that how it shows up in dev?
<GiGa>
srhb: It only shows under /dev/mapper and shows as "/dev/mapper/aug2019home-home"
<GiGa>
srhb: I have nothing at "/dev/aug2019home" as that path does not exist. Equally it didn't earlier when I had it mounted
<srhb>
But.. Isn't that the whole LVM thing
<srhb>
Because you called it: boot.initrd.luks.devices.aug2019home-home.device
* GiGa
isn't very good with LVM
<srhb>
Where augh2019home-home appears to me as the device you get from decrypting that device
<srhb>
Eg. NOT the lv.
<srhb>
But the entire lvm partition
<srhb>
Maybe I'm not either :P
<evanjs>
bah. needed the config and pkgs imported at the bottom. Solved.
<srhb>
GiGa: iow you should end up with a device node that looks like... /dev/$VGNAME/$LVNAME
george has quit [Remote host closed the connection]
<Miyu-chan>
Hm. More ideas.
<gentauro>
21:40 < srhb> gentauro: Great! Sad I didn't make it. :)
<gentauro>
yeah
<srhb>
GiGa: Do you have /dev/mapper/aug2019home-home/home? :P
<gentauro>
matk Korea in the calendar ;)
<srhb>
gentauro: Lots of travels for you :P
<Miyu-chan>
Oh wait, I can't just shebang SSH, because exec needs access to the file itself lol
<Miyu-chan>
(This is in reference to `nix run` shebang a while ago.)
<gentauro>
next year is NJ (States) which I will skip
jgt has joined #nixos
<gentauro>
but I think in 2021 it's Korea :D
abathur has quit [Quit: abathur]
phdoerfler has joined #nixos
phdoerfler has quit [Client Quit]
<GiGa>
srhb: Sorry, had to say good bye to some peeps here. One sec
<GiGa>
srhb: I do not. I can see it's a symlink to dm-1 though
<GiGa>
srhb: I mounted the filesystem earlier though, as I copied all my data to it!
wfranzini has quit [Remote host closed the connection]
<srhb>
GiGa: *what* is a symink to dm-1 ?
wfranzini has joined #nixos
<GiGa>
srhb: Sorry, that wasn't clear: /dev/mapper/aug2019home-home is a symlink to ../dm-1
<{^_^}>
[nixpkgs] @worldofpeace pushed to master « pantheon.elementary-gsettings-schemas: use File URI »: https://git.io/fjNxV
<srhb>
GiGa: Hm..
<GiGa>
srhb: These are the commands I ran earlier to create the LVM within the LUKS container: https://pastebin.com/LcUfCjwP
<GiGa>
and once mounted at /mnt/floppy I successfully copied all my data over
<GiGa>
So now I'm completely confused as to why this doesn't work. Equally I know I won't necessarily be able to boot to this environment again as my /home on the faulty disk is less and less stable.
dasj19 has quit [Quit: dasj19]
Chiliparrot has joined #nixos
<GiGa>
srhb: If I comment out the lines relating to sdf4 and /home and do a rebuild (without switch) that should give me a config I can boot into in the event the disk fails completely right?
<GiGa>
srhb: Ah, nixos-rebuild has to run with an argument. Just looking at the man page to see if there's one that doesn't switch to it but does make it available in the future
<srhb>
build.
<GiGa>
Yes, just came to the same conclusion
<srhb>
But that won't make it bootable.
<GiGa>
Just noticed that too
<GiGa>
there must be a way to get a bootable config
<GiGa>
?
<GiGa>
but not switch to it?
<GiGa>
Ah, boot
<GiGa>
OK, that's hopefully got me a fall back in case the drive fails
<Miyu-chan>
Interesting. One can do `st nix-shell -p bash`, that is, use nix-shell as an "interactive shell," for things that expect one, right?
<GiGa>
Hmm, this is odd. If I attempt to luksClose "crypttemp" it's not active
<Miyu-chan>
FOr example, Emacs scripts which use interactive processes.
<srhb>
GiGa: I think we may be coming at this wrong because we're analyzing it after you've done some manual things
<srhb>
GiGa: I suggest you boot into a new config where the device is luksOpened but you're still mounting your old home
* GiGa
thinks that through
<srhb>
GiGa: Then we should just have to determine how the lv shows up in /dev
<GiGa>
OK, I think I can build that
<srhb>
Because it's already luksOpened.
Guest49 has joined #nixos
<GiGa>
So I want two boot.initrd.luks.devices lines, one for the old home and one for the new
<GiGa>
but /home mounted on the old one
<Guest49>
Hi everyone. I'm a bit of a newb, but I managed to set up a declarative syncthing, a custom xmonad and some declarative users on a bare-metal machine.
<Guest49>
Now I'm trying to set up a local hydra server
<Guest49>
is there any way to declare the projects in the nix file where I define the hydra server?
<GiGa>
srhb: BRB, rebooting into that dual luksOpened config
GiGa has quit [Quit: Leaving]
Soo_Slow has quit [Remote host closed the connection]
<srhb>
Guest49: Alternatively I suppose you could write a systemd unit that connects to postgres and adds the project as necessary. That could become almost declarative.
<srhb>
Guest49: But the main issue is exactly that: It's a project defined in a database.
<Guest49>
It sounds like hydra is not really what I thought it was / what I need it to be
<Guest49>
I'll try again in about 3 years
<Guest49>
or build a replacement if I find a client to pay for that ;P
andreas303 has quit [Remote host closed the connection]
<srhb>
Guest49: I will add that once you have basically one project setup you can basically do the rest declaratively
<srhb>
Guest49: Even adding, removing packages, whatnot.
<srhb>
Guest49: But if you do manage to get something more declarative than Hydra written, please let us know :P
<Guest49>
forgive my non-familiarity with hydra
<Guest49>
but is a project 1:1 to a repo?
<srhb>
Nope.
<srhb>
It's just a logical grouping.
<Guest49>
so 1:N then?
<Guest49>
ok
andreas303 has joined #nixos
<srhb>
It has no inherent meaning outside of perhaps permissions to modify things inside it
jgeerds has joined #nixos
<Guest49>
so the setup is essentially: 1. set up project, point it to a repo and then do whatever necessary at that repo
o1lo01ol1o has quit [Ping timeout: 244 seconds]
<srhb>
Correct.
<Guest49>
hm
<Guest49>
that may still be worth it then
<Guest49>
the other thing that I did not like about hydra is this
<GiGa_recovery>
srhb, It's not going well. Currently the old drive is refusing to work so I've just booted a config with the new drive luksOpened by configuration.nix but not mounted
Havvy has quit [Read error: Connection reset by peer]
<srhb>
It contains a number of disparate things.
<Guest49>
$ hydra-queue-runner --helperror: unrecognised flag '--help'Try 'hydra-queue-runner --help' for more information.
<srhb>
Guest49: There's lots of things like that. If it bothers you already, maybe steer clear :P
<Guest49>
jup, will do
<Guest49>
must retain my sanity
<srhb>
GiGa_recovery: And where's the lv?
<GiGa_recovery>
srhb, "sudo lvdisplay" says /dev/aug2019home/home yet that path does not exist
ensyde has joined #nixos
<srhb>
GiGa_recovery: And if you explore /dev, do you find it?
<GiGa_recovery>
srhb, nope
LouisDK has joined #nixos
<GiGa_recovery>
There's dm-0, which would be the LUKS, but nothing under it
<LouisDK>
Is there a proper way inside a nix file to due an operation only if a specific package is installed?
Havvy has joined #nixos
<srhb>
GiGa_recovery: No idea then
<GiGa_recovery>
srhb, Is this better done in a /msg ?
<srhb>
LouisDK: No. That's not even something you can say in Nix :)
<srhb>
GiGa_recovery: sure, but I feel like I'm out of ideas :)
<Guest49>
@LouisDK that sounds like an XY problem
<Guest49>
what are you trying to do?
<tilpner>
srhb: What about a readDir on storeDir? c.c
* srhb
hands tilpner cone of shame
* tilpner
is sorry
<GiGa_recovery>
srhb, Makes two of us. I'm completely stumped as to how I was able to copy data to this LVM earlier yet now I can't see it. Interestingly we can see from my earlier pastebin that I never mounted it via /dev/aug2019home.
<srhb>
Messaging you, then
<LouisDK>
@Guest49 I'm trying to only enable speciific functions in a program upon compiling if a specific package is installed.
<{^_^}>
[nixpkgs] @charles-dyfis-net opened pull request #67411 → bees: Try all findmnt modes in service wrapper → https://git.io/fjNxD
<evanjs>
I believe that's typically done via flags (e.g. ripgrep has PCRE2 on by default now) but idk if stuff has "reflection" to that extent on a Nix system
<srhb>
LouisDK: The right way to do that is to declaratively specify what a package _should_ have activated, and then if it is, it will pull in the necessary dependencies
<srhb>
evanjs: No, that would be impure.
<srhb>
(Looking into the store, seeing what's there, that is)
<srhb>
That would undo nix entirely :)
<evanjs>
srhb: right exactly. The closest I can think of is things being installed/configured based on options that are set, e.g. What services.xserver.displayManager.gdm.enable does on top of just installing gdm
<evanjs>
But again, not the same thing
lally has quit [Ping timeout: 244 seconds]
<srhb>
evanjs: Indeed, yeah. :)
<LouisDK>
I have a configure script which inlcudes "if enabled xnvctrl; then" and eventrough it isn't enabled it continues therefore it'd be a good idea to test if nvidia-x11* is even installed on the system and if not just comment the hole segment.
<evanjs>
Though, that might actually be the proper approach, depending on the application. Making an option, that is. Then you could do cfg.mkIf, etc. infinisil has quite a few good examples in his config
<evanjs>
Yeah mkIf can just be given several things. If nvidia is enabled in the user's config you can assume it's going to be present
<Guest49>
srhb: Thanks for the help, I'm signing off!
<srhb>
Guest49: o/ good luck
slack1256 has quit [Ping timeout: 258 seconds]
Guest49 has quit [Remote host closed the connection]
o1lo01ol1o has joined #nixos
nabana has quit [Ping timeout: 264 seconds]
fusion809 has joined #nixos
o1lo01ol1o has quit [Remote host closed the connection]
jgt has quit [Ping timeout: 252 seconds]
<lejonet>
What the heck am I doing wrong if I get exec format error with ZFS and SPL modules for the kernel? I'm trying to use the linuxPackages_hardened kernel, and have boot.extraModulesPackages = with config.boot.kernelPackages; [ zfsSTable ]; and I can also see both zfs and spl being built for the kernel (I'm trying to deploy with nixops)
pie_ has quit [Ping timeout: 250 seconds]
jgt has joined #nixos
<lejonet>
Is linuxPackages_hardened simply just not compatible with the ZFS modules?
<ivan>
lejonet: possibly a version mismatch requiring a reboot
<lejonet>
ivan: well, its during reboot stuff goes south, because the modules cannot be loaded...
<ivan>
ah
<lejonet>
(due to exec format error)
<lejonet>
Which would mean that they were in fact built for another kernel than the one trying to boot, which feels odd. Did the build target the currently running kernel or?
Myhlamaeus has joined #nixos
<lejonet>
I use boot.kernelPatches to add 2 extraConfig things (EXPERT y and ARCH_RANDOM n, to disable RDRAND usage by the kernel)
<lejonet>
Is there something more than "just" adding them to boot.extraModulePackages that has to be done if you trigger a kernel rebuild with boot.kernelPatches that I'm missing?
domogled has joined #nixos
phreedom has joined #nixos
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
phreedom_ has quit [Ping timeout: 260 seconds]
wfranzini has quit [Remote host closed the connection]
wfranzini has joined #nixos
casaca has quit [Ping timeout: 245 seconds]
casaca has joined #nixos
<lejonet>
Imma try with linuxPackages_latest just to rule out linuxPackages_hardened being the culprit
<ivan>
do I need to DELETE FROM to make Hydra forget evaluations? (I changed where I store things and need to rebuild)
Guest49 has joined #nixos
eof has joined #nixos
<Guest49>
Hi again, I said I was gone but it turns out I lied
<Guest49>
because now I have an even stranger problem
<Guest49>
when I run `df -h`, I see that there is a partition, /dev/sdb2 that has size 50G, and is full.
GiGa_recovery has quit [Quit: Leaving]
<srhb>
Guest49: Yes?
<Guest49>
when I run `lsblk -fs`, it says that the thing disk is >900G in size
<Guest49>
so somehow they disagree...
o1lo01ol1o has joined #nixos
<Guest49>
the partition is 900G in size
<ivan>
can you paste the outputs
o1lo01ol1o has quit [Remote host closed the connection]
<lejonet>
Guest49: the filesystem doesn't have to utilize the full partition (even if that is rare and usually something you have to manually ensure)
GiGa has joined #nixos
GiGa has joined #nixos
GiGa has quit [Changing host]
<lejonet>
or if the filesystem was moved (i.e. dd'd) from another disk, that was smaller, to this new partition
<Guest49>
aha
<Guest49>
so I could just put the fs on it again
<Guest49>
and that should be fine then
<Guest49>
with mkfs.ext4 or so
<lejonet>
If you don't need the data on the partition, sure
<lejonet>
but I concur with ivan, please paste outputs of lsblk -fs and df -h
justanotheruser has quit [Ping timeout: 245 seconds]
<lejonet>
(throw in a fdisk -l /dev/sdb while you're at it)
LouisDK has quit [Remote host closed the connection]
<lejonet>
Guest49: I would guess that the filesystem on /dev/sdb2 was copied from another drive, where it was 50GB in size
<lejonet>
seeing as its ext4, you could just do resize2fs to enlarge it to the full size of the partition
mcmx has joined #nixos
<Guest49>
well, that was surprisingly easy
<Guest49>
holy smokes
<Guest49>
thanks
ilmu has joined #nixos
<lejonet>
Most FSes have a way to enlarge if the partition is larger than the FS claims of the partition (shrinking them is a completely different beast)
<lejonet>
Np
oborot_ has quit [Ping timeout: 248 seconds]
<srhb>
lejonet: Most definitely have a way of shrinking too! It's just lossy!
<Guest49>
lejonet and ivan, if you ever need any help, please contact me via cs-syd.eu (I'm sure you can find the address) because I'm only on irc to coerce other people into helping me
<Guest49>
srhb, the same holds for you ;)
<srhb>
o/
<lejonet>
srhb: True, lossy and/or offline :)
oborot has quit [Ping timeout: 248 seconds]
<lejonet>
I have done the stupid thing of shrinking the fs on /, it never ends well... :P
<GiGa>
srhb managed to fix my issue
<ivan>
man I sure hope "store_uri = daemon" is a working Hydra option
<GiGa>
For those following along, my LUKS mapper had the same name as my LVM volume group
<GiGa>
so the system freaked out essentially
<dahirsch>
thats never a good idea
<srhb>
I'd never seen that before
<srhb>
Doesn't look pretty xD
tg has quit [Ping timeout: 245 seconds]
<lejonet>
GiGa: that is fully understandable, because they will map to the same symlink name in /dev/mapper :)
<GiGa>
Indeed
<lejonet>
that is why many distros, like RHEL, always append "vg-" and "lv-" to LVM LVs and VGs
<lejonet>
prepend*
ilmu has quit [Ping timeout: 250 seconds]
<GiGa>
so I now know a lot more than I did about LVM (although still not a lot)
<lejonet>
LVM is... fun :P
<srhb>
lejonet: Doesn't help much if people follow the same logic for luks :P
<srhb>
luksOpen foo vg-bar
<srhb>
woo
<srhb>
:P
<lejonet>
:P
zupo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<lejonet>
luksOpen damnit whoops
tg has joined #nixos
<GiGa>
Which is kinda what I did
<GiGa>
So, to celebrate I think I'm now going to watch some TV, before I'm actually ready for sleep
<GiGa>
clever: Virtualbox now works too :)
<lejonet>
GiGa: :)
<GiGa>
So, today's NixOS lessons: enable virtualbox the correct way (not by declaring the package), set your luks node to a different name to your PV or LV
<GiGa>
The latter applying to other linuxes too
<GiGa>
Given the now significantly better performance of my system, that other disk has been failing for a while
<lejonet>
Its always good to namespace your different layers :)
civodul has quit [Quit: ERC (IRC client for Emacs 26.2)]
<GiGa>
lejonet: It really is
<lejonet>
GiGa: which is why I literally append the layer nr "-layer-1" "-layer-2" etc
<mcmx>
the installation instruction should really mention encryption, it's absolutely necessary if you're using an SSD and want to be able to actually delete data
justanotheruser has quit [Ping timeout: 268 seconds]
<lejonet>
Hmm, it would seem like it is indeed linuxPackages_hardened that is my problem with ZFS, using linuxPackages_latest at least boots like it should, ZFS and all
justanotheruser has joined #nixos
pie_ has joined #nixos
pie_ has quit [Ping timeout: 250 seconds]
UltraPhil has quit [Quit: Leaving]
lally has joined #nixos
<lally>
Hi, what do you folks use to build stack-based haskell apps into nix packages? stack2nix, stackage2nix, some version of cabal2nix?
<Guest49>
a custom-written nix overlay :+1:
dahirsch has quit [Ping timeout: 250 seconds]
<lally>
I tried buildStackProject, but I can't include that in a derivation that depends on it (namely, dockerTools.buildImage)
<lally>
I really want to minimize cognitive overhead for this going forward..
<infinisil>
lally: buildStackProject should work but I can't recommend it. stack2nix is the next best thing I can recommend, but you might want to try out a haskell.nix stack setup instead, which will probably work even better (I haven't tried it myself though)
Zer0xp has quit [Ping timeout: 248 seconds]
phdoerfler has joined #nixos
phdoerfler has quit [Client Quit]
growpotkin has joined #nixos
<lally>
buildStackProject is enough to build for stack's use, but isn't enough to let nix build it. It tries to do some network requests during build :/
work_ has joined #nixos
chr15 has quit [Ping timeout: 272 seconds]
<lally>
I'll have another look at haskell.nix then, but I was weary of (a) no release versioning, (b) a lot of raw downloads that seemed a bit sketch
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fjNhq
<Gigabitten>
ok so earlier I asked how much space reflex is supposed to take up and before something failed after a very long time my nix store broke 20 gigabytes so I'm pretty sure something's still broken
<Gigabitten>
`error: getting status of '/home/myusername/reflex-platform/gc-roots/shell.drv': No such file or directory` was the failure message
<Gigabitten>
22.9 gigabytes, it came out to
growpotkin has quit [Read error: Connection reset by peer]
<lally>
infinisil: Thanks, I'm taking a more serious look into haskell.nix now, and just found their irc channel.
Guest49 has quit [Remote host closed the connection]