<{^_^}>
[nixpkgs] @copumpkin pushed to staging « qt48: fix on darwin »: https://git.io/fAMfX
hamishmack has joined #nixos
patrl has quit [Ping timeout: 244 seconds]
edwtjo has quit [Ping timeout: 252 seconds]
halfbit has quit [Quit: WeeChat 1.9.1]
<__red__>
btw - once I've resolved all my issue with my PR and renamed it from WIP - is there anything else I need to do to bring it to the attention of someone with commit access? ... or is it just a case of letting the peeps work through the 800+ PRs in front of it?
<samueldr>
you pretty much explained it :)
<samueldr>
though, not all PRs are in front, but giving it an appropriate amount of time is often needed
<__red__>
looks like being a committer to nixpkgs could be a fulltime job :-/
<samueldr>
(a part of the 800+ PRs are either stuck, or WIP)
<samueldr>
sponsorships opportunities for deep-pocketed companies with interest in nixpkgs could help
<__red__>
yeah
<samueldr>
(not an official voice here)
<__red__>
so does a yellow label indicate stuck?
<samueldr>
no, nothing really indicates PRs as being stuck (except WIPs one)
<__red__>
so success on the three architectures, and "All checks have passed (8)" is all I need to do ?
<__red__>
I guess I just want to be sure that I've done everything that I need t odo
<samueldr>
haha +1 for using unicode tick marks instead of ticking the boxes :)
<samueldr>
#46753 for those wondering which was __red__'s PR
<__red__>
yeah - I didn't realize it was markup until it got created
<__red__>
:-)
<samueldr>
as far as github is concerned, putting a "x" there checks it
<{^_^}>
[nixpkgs] @washort opened pull request #46765 → [WIP] Bump AMDGPU-PRO to 18.03 for use with 4.18 kernel → https://git.io/fAMJk
jasongrossman has left #nixos ["ERC (IRC client for Emacs 26.1)"]
<samueldr>
(personally, I replace the checkboxes with those emojis ⬜✔️❌ as I dislike having "tasks" opened on a PR)
<samueldr>
it just *ticks* me wrong
<__red__>
I'm about to start a gnarly microprocessor project in the next few days and I really wanted to do it from my nixos workstation.
<__red__>
and arduino is already there, as is all the atmel stuff.
<__red__>
parallax isn't, so figured I'd add it for others who need it.
<samueldr>
that's good :)
<__red__>
I'm enjoying it, although this next package is causing me grief
<samueldr>
s/grief/learning experience/ ;)
<__red__>
they made the build directory outside of the source directory so I'm hacking the shit out of it
<samueldr>
oh
<__red__>
in hindsight, I probably should have just made a subdir for it and cd'd into it and then let it run make
<samueldr>
how "outside" is it?
<__red__>
then the "outside" would be "inside"
<__red__>
one directory down
<samueldr>
probably easier the way you described
<__red__>
$(realdir ..)/build
<__red__>
yeah- they also do annoying things like creating directories with architecture names etc
<__red__>
a real pita
<__red__>
but, like I said earlier this eve I kinda screwed myself as I can't commit this one until my other PR is accepted as the one I already submitted is a build dependancy for the second
<__red__>
kinda funny
<__red__>
but this is the way we learn
<samueldr>
how's your git? if you create a branch from your first commit, you can always rebase later on master once it reached it
<samueldr>
from your first PR*
wpcarro has joined #nixos
kreetx has joined #nixos
edwtjo has joined #nixos
<__red__>
I've 'cheated', so I can still work on it
<{^_^}>
[nixpkgs] @Infinisil opened pull request #46766 → haskellPackages: Fix ListLike for 8.02 and 8.22 → https://git.io/fAMJA
Kim has joined #nixos
<Kim>
Has anyone gotten chromium working with plasma-browser-integration? So far I've installed `plasma-browser-integration` system-wide and set up an override to include the package in chromium's `buildInputs`
worldofpeace has quit [Remote host closed the connection]
alex_giusi_tiri has left #nixos [#nixos]
kreetx has joined #nixos
Kelppo has quit [Ping timeout: 244 seconds]
kreetx has quit [Ping timeout: 245 seconds]
<Enzime>
,tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
LnxBil has joined #nixos
<LnxBil>
Alⅼɑһ іs doing
LnxBil has quit [Killed (Sigyn (Spam is off topic on freenode.))]
phreedom__ has quit [Remote host closed the connection]
phreedom has joined #nixos
lividsloth has joined #nixos
b has quit [Quit: Lost terminal]
mayhewluke has quit [Ping timeout: 272 seconds]
mayhewluke has joined #nixos
<infinisil>
Yay, I succeeded in getting rid of a problem for once!
kreetx has joined #nixos
<infinisil>
My streak of endlessly running into new problems without ever solving one is over
<infinisil>
Actually I even solved another thing, namely taffybar constantly crashing with segfaults. The solution to that one is to compile it with GHC8.2 instead of 8.4, I think
<infinisil>
(I think because the segfaults happened rather randomly, so I'm not 100% sure, but it hasn't crashed once with 8.2
worldofpeace has joined #nixos
<lividsloth>
Oh nice. I had never heard of beets. Gonna have to play with it.
kreetx has quit [Ping timeout: 240 seconds]
<emily>
ooh I'd heard of beats before but forgot about it, thanks for the reminder!
<infinisil>
Heh sure, I'm not entirely sure if I can recommend it, I'm not using it to its full potential, and I've gotten into some problems
<lividsloth>
Is there anything special I should know about installing NixOS on a headless system where I need the serial console?
<emily>
infinisil: *nods*
<emily>
to be fair I find that my needs in practice these days are served sufficiently by mpv and a shell
<infinisil>
lividsloth: Not really, by default it doesn't have any DE or so, so you should get to the console directly
<infinisil>
emily: Okay that would be a bit too minimalistic for me
<infinisil>
emily: Hold on, let's go to #nixos-chat before I start babbling uncontrollably
<{^_^}>
[nixpkgs] @peterhoeg merged pull request #43812 → epmd: Introduce erlang port mapper daemon service → https://git.io/fNn6d
<worldofpeace>
samueldr: I think fetchpatch produces a normalized patch so I usually use a placeholder hash and round robin. Though I think fetchurl could be safe too
<{^_^}>
[nixpkgs] @copumpkin pushed to staging « valgrind: don't force LLVM 3.9 »: https://git.io/fAMIh
<clever>
Myrl-saki: probably static vs dynamic linking of the haskell code
<clever>
Myrl-saki: it is also possible to mix it some, static haskell but dynamic everything else
<Myrl-saki>
True.
<Myrl-saki>
Just for funsies.
<jluttine>
where can i find documentation about some config value in nixpkgs? in particular, i'd like to know what i can put in `meta.platforms` in mkDerivation
<dhess>
I'm getting "fork... resource temporarily unavailable" errors when I try to build terraform+plugins on a NixOS build host. Any suggestions?
<colemickens>
I am not sure why I can't figure this out, but I want to have a systemd script call nixos-rebuild. What pkg do I need to make avaible to the path?
<colemickens>
no, I'm trying to script a cloud VM, I can provide a custom nix config, but not execute commands before/after. So, I'm having it run a service that will checkout nixpkgs, my nixcfg, symlink configuration.nix and then "switch" into it.
kreetx has joined #nixos
<colemickens>
I guess it would easier if I could just get the images to accept #!/usr/bin/env bash, in addition to #!nix.
<v0|d>
morning, i'm on nixos-unstable. I have a trouble w/ nvidia drivers/cudatoolkit. My driver is 390.77, cudatoolkit is 9.1.85. Sample in the kit tells me cudaGetDeviceCount returned 35 -> CUDA driver version is insufficient for CUDA runtime version Result = FAIL. any ideas?
baimafeima2 has joined #nixos
Tucky has joined #nixos
brejoc has joined #nixos
hiratara has joined #nixos
<Aleksejs>
Hello, has anyone fixed skype yet? I'm still getting an error "cannot download from any mirror"
Boomerang has joined #nixos
<MarcWeber>
Aleksejs: web.skype.com might work good enough
<MarcWeber>
I have some extra PHP code so I need cfg@{ all the options } and access cfg, but it only contains version and hash, not the fmpSupport like flags.
<MarcWeber>
At least that's what traceVal shows - shouldn't all { foo ? ...} like arguments be populated?
oida has quit [Ping timeout: 256 seconds]
oida has joined #nixos
civodul has joined #nixos
cement has quit [Ping timeout: 246 seconds]
musey has quit [Ping timeout: 272 seconds]
<Aleksejs>
I've actually noticed that they have a 'latest' link for skype now
<Aleksejs>
like this repo.skype.com/latest/skypeforlinux-64.deb
xt26 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
Mateon1 has quit [Ping timeout: 272 seconds]
Mateon2 is now known as Mateon1
<hyper_ch2>
this one was even able to post twice :)
<rsa>
hmm, i thought having a package in system.extraDependencies and then adding it to environment.systemPackages would work without downloading additional stuff
<rsa>
turns out i was wrong
rauno has quit [Ping timeout: 250 seconds]
<rsa>
maybe i'm missing outputs and need to add those explicitly
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « matomo: 3.5.1 -> 3.6.0 (#46242) »: https://git.io/fAMsM
phyber14 has joined #nixos
<phyber14>
Allah iѕ doⅰng
phyber14 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
rauno has joined #nixos
baimafeima2 has quit [Quit: Leaving]
<Aleksejs>
srhb: in general it works: IM and video are working, but sound/mic does not, it shows "ALSA lib pulse.c:243:(pulse_connect) PulseAudio: Unable to connect: Connection refused"
<{^_^}>
[nixpkgs] @xeji pushed commit from @r-ryantm to master « libcouchbase: 2.9.2 -> 2.9.4 (#46254) »: https://git.io/fAMGf
<melleb>
Hi all, I'm trying to set-up C# development on NixOS from a nix-shell env. Does anyone have any experience with this? I'm able to install the dotnet-sdk and mono, but integration with vscode isn't running smooth. It seems the debugger binary is generated, but when I'm trying to run it from the command line it sais 'file not found'. I don't know how to deal with this... Any pointers would be greatly
<sphalerite>
Aleksejs: are you actually running pulseaudio?
klntsky has quit [Ping timeout: 256 seconds]
<Aleksejs>
sphalerite: yes. I use bluetooth headset and after couple more tests I've noticed that sometimes skype sees this device in a/v settings but sometimes it doesn't. Also pavucontrol doesn't show me that skype uses sound devices
<sphalerite>
ah ok
klntsky has joined #nixos
jasongrossman has joined #nixos
fendor has joined #nixos
<Aleksejs>
when it sees then test sound does work but there's no sound during call and mic doesn't work as well
alexherbo2 has joined #nixos
<alexherbo2>
Hello
<alexherbo2>
I prepare my first NixOS install
<{^_^}>
[nixpkgs] @xeji merged pull request #46744 → exim: add optional support for PAM → https://git.io/fA1Me
<{^_^}>
[nixpkgs] @xeji pushed commit from @pacien to master « exim: add optional support for PAM (#46744) »: https://git.io/fAMZw
<etu>
alexherbo2: Nice, have fun! :)
<alexherbo2>
Is there recommended size for the UEFI partition and position (sda1 or sda2)?
<{^_^}>
[nixpkgs] @xeji pushed commit from @pacien to release-18.09 « exim: add optional support for PAM (#46744) »: https://git.io/fAMZD
<srhb>
alexherbo2: You should probably make your boot partition a lot bigger than what you're used to, because the rollback function relies on old kernels being available
* goibhniu1
thought that microsoft dropped linux support for skype entirely
<Ralith>
alexherbo2: I'd go for 1GB or so, to be very generous
<srhb>
alexherbo2: I recommend 500MiB-1GiB
kisik21 has quit [Ping timeout: 250 seconds]
<Ralith>
running out of space to store kernels is a pain
<etu>
alexherbo2: My /boot is also my UEFI partition. It's 1GiB
<alexherbo2>
In the install guide, I've see 512M, and it's on sda2
<Ralith>
goibhniu1: they replaced the functional Qt app with a janky electron crapheap
<goibhniu1>
ah, thanks ... good to know
<Ralith>
but it runs
<alexherbo2>
These two choice are arbitrary or there is actual rationales?
<Ralith>
stripped out all the settings too :(
<srhb>
alexherbo2: sda2? Doesn't matter.
<alexherbo2>
I usually see the boot partition always on sda1
<alexherbo2>
Ouki :p
<srhb>
alexherbo2: I don't see that in the manual though
<srhb>
alexherbo2: Where did you see it?
<hyper_ch2>
I use 1GB on /boot (bios, note efi) so that I could put an nixos-iso on it and boot from it - if needed
<adisbladis>
I typically make mine 512M, but I do gc every so often
<{^_^}>
[nixpkgs] @mcmtroffaes opened pull request #46773 → brotli: ensure CMAKE_SYSTEM_NAME=Windows on mingw cross builds → https://git.io/fAMZd
<etu>
alexherbo2: You're probably fine with that :)
<Ralith>
probably
<elvishjerricco>
alexherbo2: Yea, you'll need about 15M per kernel version IME.
<elvishjerricco>
So that'll last you quite a large number of NixOS upgrades
<alexherbo2>
Thanks :)
<elvishjerricco>
And the just GCing removes the old ones I think
<etu>
My raspberry had 120MiB. That was enough for 2 generations at the time.
<Ralith>
mine's 64MiB and it's mildly annoying
<elvishjerricco>
etu: I think the situation is different on RPi
<elvishjerricco>
Not sure
<etu>
I did a repartition when I moved it to a new SD-card and gave it 1GiB
<etu>
elvishjerricco: One kernel on there takes about 50MiB
<etu>
I could do two generations at once :D
<etu>
Unless it was the same kernel version
<elvishjerricco>
etu: Odd. 11M per initrd, 4M per kernel on my x86_64 desktop
<alexherbo2>
I see on the manual we must execute `nixos-rebuild switch` as root; `sudo nixos-rebuild switch` works? or it needs `-i` flag
<elvishjerricco>
I think initrd is much more bloated in the RPi install
<alexherbo2>
It's the first time I've see the `-i` flag, I'm not sure I get the difference
<srhb>
alexherbo2: Where do you see that flag reference?
<adisbladis>
elvishjerricco: Presumably rpi has far more firmware blobs
<elvishjerricco>
alexherbo2: `-i` makes it do a login shell for root, which sets more environment variables and stuff. It's more like `ssh root@localhost` than typical `sudo`.
<srhb>
alexherbo2: If you mean for sudo, that just gives you an actual login shell.
<sphalerite>
alexherbo2: search on the options page I linked above or in the configuration.nix manpage
<alexherbo2>
Thanks
phreedom has quit [Ping timeout: 256 seconds]
<elvishjerricco>
LnL: It always annoyed me that environment variables don't go to a special argument to mkDerivation. Right now you just get no errors if you type an argument name wrong.
<LnL>
you mean like stdenv.mkDerivation { name = "foo"; variables.FOO = 42; } ?
phreedom has joined #nixos
<hyper_ch2>
I finally managed to access my qemu vm behind nat via spice through webbrowser
<sphalerite>
elvishjerricco: !!
<sphalerite>
big +1 to that
<sphalerite>
never occurred to me x)
<sphalerite>
hyper_ch2: woooo congratulations
<elvishjerricco>
LnL: I would have chosen `environment.FOO`, but yea
<LnL>
yeah, that's a pretty good idea
<hyper_ch2>
sphalerite: now I just need to test if all https works also.. and then undo the unnecessary changes :)
<LnL>
stdenv.mkDerivation changed a lot over the years
<elvishjerricco>
LnL: What are the chances we can introduce `environment` now, and phase out the classic way over the course of some years?
<sphalerite>
hyper_ch2: and publish your nixos module for it? ;)
<hyper_ch2>
not really sure how to publish that as nixos module
<elvishjerricco>
TBH I wish all of nixpkgs were defined a la the module system.
<sphalerite>
elvishjerricco: I suspect pretty low, but opening an issue for it might be a good start
<hyper_ch2>
basically: use qemu with spice; download the spice-html5 client; setup nginx as remote proxy, download websockify and run it, open/forward ports on nixos and router
Fare has quit [Ping timeout: 252 seconds]
<LnL>
things like hello = pkgs.hello; in combination with $hello/bin/hello used to be common instead of the buildInputs, etc. abstractions we have now
<etu>
elvishjerricco: As long as both works people may accept it? This is a guess :)
<elvishjerricco>
i.e. `options.packages = mkOption { type = attrsOf (submodule { ...mkDerivation args as a type... }); };`
<{^_^}>
[nixpkgs] @xeji pushed commit from @rembo10 to master « postman: 6.2.5 -> 6.3.0 (#46771) »: https://git.io/fAM89
phreedom has joined #nixos
Asmadeus has joined #nixos
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos
trcc has quit []
<v0|d>
how do I learn the path of an output of a derviation? ie echo $(nix-build -E '${pkgs.fooPkg}')
<sphalerite>
v0|d: nix-build '<nixpkgs>' -A foo
<elvishjerricco>
LnL: Can't say I'm liking that gist so far...
<v0|d>
are there any generic ways to evaluate and return?
carlosdagos has quit [Quit: Connection closed for inactivity]
<v0|d>
maybe using nix repl
Avaq has joined #nixos
<elvishjerricco>
Hard coding the meta data into the language and the global name spacing... Plus I can't imagine how you'd do NixOS style merging. And the disallowing of dynamic definitions makes it totally useless
<nwspk>
I was just checking the NixCon-page and - after clicking over 9k times on uMatrix - it seems like there are no more tickets, am I right?
<sphalerite>
v0|d: the old UI is a bit of a mess (and the new one isn't complete), so you currently still need to use a variety of different commands with different flags to achieve similar thigns :/
<v0|d>
:) see. thnx.
<sphalerite>
nwspk: yep that's what it looks like to me
<nwspk>
sphalerite: too sad, just got time today ;_;
<sphalerite>
nwspk: you may still have a chance at buying a ticket off someone else who bought one but can't go after all
<sphalerite>
nwspk: maybe post on discourse asking
<nwspk>
sphalerite: thanks, I'll see if I can still get one, otherwise I'm aiming for 2019
<sphalerite>
nwspk: good luck, I'll see you there if you can make it I guess ;D
<etu>
nwspk: I missed this year because I waited too long to decide :(
<nwspk>
sphalerite: if I'll make it, I'd be happy to talk to you over a beer or something (:
chpatrick has joined #nixos
<nwspk>
etu: kind of same here.. wasn't sure if I'll have time and now when I'm sure they are out
<sphalerite>
well, often people buy tickets in advance, not knowing that they won't have time, and then sell them on afterwards. So I'd say you both still have a chance!
<nwspk>
sphalerite: then I still need to buy the flight and an accomodation, which won't be cheap if I can't buy in advance
<melleb>
Hi all, I'm trying to fix rpath using patchelf, but I can't seem to get it to work... I'm following the steps described here: https://nixos.wiki/wiki/Packaging/Binaries
djinni13 has joined #nixos
<djinni13>
Аllaһ is ԁoіᥒg
<djinni13>
moοᥒ ⅰs ᥒot dοіᥒg Aⅼlaһ is doiᥒg
<djinni13>
ѕuᥒ iѕ ᥒot ⅾoing Allɑh іs ԁoinɡ
djinni13 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<DigitalKiwi>
3...bots slacking
tzemanovic has quit [Ping timeout: 252 seconds]
<sphalerite>
melleb: how is it not working?
<melleb>
For some reason I'm not able to set the rpath
<sphalerite>
what command are you using, and how are you determining that it's not being set?
<melleb>
sphalerite: I need the binary to point to the correct stdc++.so.6
kreetx has quit [Ping timeout: 244 seconds]
mayhewluke has quit [Ping timeout: 240 seconds]
phreedom has quit [Quit: No Ping reply in 180 seconds.]
<sphalerite>
Anybody know of something like websockify that can wrap unix sockets instead of TCP sockets?
kreetx has quit [Ping timeout: 250 seconds]
<atu>
adisbladis: Do you know who's in charge of tickets? Due to a communication mistake we bought 5 extra tickets by mistake, and are still waiting for the secretary to try to cancel them, so that other people can get them instead
<srhb>
atu: Oh wow :P
ciil has quit [Quit: Lost terminal]
alexherbo2 has joined #nixos
<atu>
srhb: We bought early bird tickets, but then the secretary also bought tickets when booking plane and hotel. :-/
<srhb>
atu: How many of you are actually going?
<srhb>
atu: Aaah, I see how that could happen.
phreedom has quit [Ping timeout: 256 seconds]
<atu>
srhb: All 5
<srhb>
atu: Nice!
<alexherbo2>
Re
<atu>
srhb: I'm sure you could swing by and snatch one of the extra tickets ;-) ;-)
<Taneb>
...if there's still these 5 tickets maybe available I should try and go
<srhb>
Haha, yeah, that would be swell...
<alexherbo2>
I got an error at my first nixos-install
<alexherbo2>
about /boot
<alexherbo2>
> File system "/boot" has wrong type for an EFI System Partition (ESP).
<{^_^}>
error: syntax error, unexpected ')', expecting ID or OR_KW or DOLLAR_CURLY or '"', at (string):195:1
phreedom has joined #nixos
<sphalerite>
etu: nwspk: would you look at that, 5 tickets!
<srhb>
alexherbo2: When using it as a efisys, it needs to be formatted as fat(32) I believe.
kreetx has joined #nixos
<sphalerite>
srhb: are you going?
orivej has quit [Ping timeout: 252 seconds]
<srhb>
sphalerite: Afraid not. :/ Hopefully next year...
<alexherbo2>
I used GParted, created a GPT table, then added 2 partitions with name and label, sda1 fat32 and sda2 ext4, then mounted fat32 to /mnt/boot
ciil has joined #nixos
<etu>
sphalerite: What stopped me is the prices of the hotels :(
<srhb>
alexherbo2: That does sound correct.
<nwspk>
sphalerite: uff, the pressure of going is back .__.
fendor has quit [Remote host closed the connection]
<etu>
alexherbo2: That should work fine :)
<nwspk>
sphalerite: but thanks a lot (:
<sphalerite>
etu: :/
<sphalerite>
nwspk: hahaha
<srhb>
alexherbo2: Can you pastebin the output of "mount" ?
<Taneb>
atu: if you can't cancel but can transfer the tickets I'd be interested
<Taneb>
(I'd also be interested if you can cancel them, but I'd need to talk to someone else in that case)
<atu>
Taneb: I'm trying to move this along as quick as possible. I'll try to get back with an update tomorrow. If we can't cancel, we'll just give them away I think
* etu
has also had an UEFI system that only booted the bootloader that was named "Microsoft something something" so I had to copy my arch bootloader to that name to get it to boot :p
<atu>
etu: If it doesn't read the specific fat32 variant UEFI requires, then it's not an EFI firmware. So..
<v0|d>
how do I install nix-mode 4 emacs?
ckrn_ has joined #nixos
<ckrn_>
Aⅼlɑһ ⅰs ԁⲟiᥒg
ckrn_ has quit [Killed (Sigyn (Spam is off topic on freenode.))]
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
<atu>
Is there a systemd target you can depend on if you want /sys/class/net to be ready (non-empty?) before running the service? I think it's mounted by sysfs, but I'm not sure which target includes the sysfs mount. Any suggestions?
<alexherbo2>
srhb, etu: It works! >:)
<melleb>
v0|d: Thx ;)
<srhb>
atu: Morally I *think* that's network-pre, but honestly that area is a bit of a mess and in need of a redesign.
<catern>
Hey #nixos, I think I want to integrate Nix into a certain build system, such that the build system can build outside of Nix, pulling some of its dependencies from Nix and some from not-Nix. Has anyone done this before? Any advice?
<atu>
alexherbo2: That's basically all I do as well
Kelppo has joined #nixos
<catern>
(for example, one place where this would be useful is Python: if you wanted to hack on multiple Python packages at once, it would be nice to pull all those packages from impure directories outside Nix so you could just modify them in place; but the package you're not working on, you could pull from Nix)
worldofpeace has joined #nixos
<catern>
What I'm wondering is how to express the build environment given that some of the deps are coming from Nix and some from non-Nix; should I generate a shell.nix or something for just the deps that I want, and build inside that reduced nix-shell?
Dedalo has joined #nixos
<goibhniu>
catern: can't you do exactly that with virtualenv and nix-shell? (for python)
<catern>
goibhniu: No, I don't believe so, you can have only one package in developer mode at once
<goibhniu>
catern: I think you can specify multiple repos in a requirements.txt which you can install into the virtualenv using pip
<{^_^}>
[nixpkgs] @baracoder closed pull request #46516 → fetchNuGet: Add support for sha512 → https://git.io/fAwMI
<catern>
goibhniu: yes, sure, but I don't think the virtualenv knows how to pull packages from Nix
<catern>
if you believe it does, then please point me to something explaining the technique
fendor has quit [Remote host closed the connection]
<catern>
since the documentation in Nixpkgs itself says there's no way to have multiple packages in developer mode at a time :)
fendor has joined #nixos
phreedom has quit [Ping timeout: 256 seconds]
<goibhniu>
catern: maybe I'm missunderstanding ... let's say you want to work on a project which requires lxml, you can install that using nix
<catern>
What does "install" mean?
<goibhniu>
then you can create a virtualenv, using this python with lxml and use pip to install multiple development packages, from github
<exarkun1>
There's the `--system-site-packages` virtualenv option, I guess.
<goibhniu>
catern: nix-shell
<exarkun1>
I can't quite see what benefit you get from mixing Nix-packaged Python software with non-Nix-packaged Python software
<goibhniu>
catern: with import <nixpkgs> {}; (python36.withPackages (ps: with ps; [requests lxml virtualenv])).env
<goibhniu>
for example
<catern>
exarkun1: so you can modify the latter packages in place
<goibhniu>
so, what am I missing?
<catern>
exarkun1: and thereby immediately see feedback rather than needing to rebuild
<exarkun1>
catern: You get that if you just install everything with pip
<exarkun1>
catern: in a virtualenv
<catern>
exarkun1: yes, but what if some things can't be installed with just pip? such as native dependencies
<alexherbo2>
Time to reboot to try, etu srhb Thanks you much for your help o/
<goibhniu>
then you can create the environment you need to build them
<catern>
goibhniu: i'm not sure you are understanding but I do think you might be right that virtualenv using something like --system-site-packages to make the virtualenv use packages from outside the virtualenv, would work in the Python case
alexherbo2 has quit [Ping timeout: 252 seconds]
<goibhniu>
you can specify which python to use: `virtualenv . --python=my-python-with-lxml`
vmandela has joined #nixos
<catern>
now, what if I wanted to build tooling to make this to be automatic? for example, I have some list of packages, and I want to mark them as either coming from Nix or coming from some local checkout of that package; should I then generate the appropriate shell.nix on the fly?
<catern>
goibhniu: interesting, and I guess that will automatically have the Nix PYTHONPATH first, so use the Nix Python packages first. that's actually really cool! have you used this technique/heard about this technique before? it should be put in the Nixpkgs documentation!
<goibhniu>
I think it's in the docs already
<goibhniu>
there's a nice section on python development
<catern>
Where? I don't believe it is
<catern>
I'm familiar with that section and I'm looking at it and it doesn't seem to be in there
<haslersn>
in paths = [ ... ] you can add as many packages as you want
lopsided98 has quit [Client Quit]
<goibhniu>
that's not really what we're talking about though
<sphalerite>
Is there a way in nixos config to create tun/tap devices? I see there's a thing for networkd, but I'm not sure what else switching to networkd will entail
<haslersn>
goibhniu: oh, then sorry
<sphalerite>
If not I'll just create a systemd service that uses ip tuntap to create one
<goibhniu>
haslersn: thanks though ... we're talking about combining that stuff with development
iyzsong has quit [Read error: Connection reset by peer]
<haslersn>
Well, you can place all the tools you need for development in there
lopsided98 has joined #nixos
<{^_^}>
[nixpkgs] @vdemeester opened pull request #46781 → git-appraise: init unstable at 2018-02-26 → https://git.io/fAM6u
<catern>
Unrelated question: Is there a trivial easy hack to make all the buildInputs of a derivation "propagate" in the same way as propagatedBuildInputs?
<catern>
Given an existing derivation, I mean
<catern>
(And also operating on the entire tree, so doing this recursively for each of the buildInputs)
<gchristensen>
you could dump `env` to a file in $out
<gchristensen>
but what does it mean to propagate
tzemanovic has quit [Remote host closed the connection]
<haslersn>
catern: such that you can combine two or more packages into one with the combined functionality?
<catern>
well, I want to make a nix-shell that has the buildInputs of all my tree of dependencies, aka the union of all buildInputs over the tree; it seems like propagatedBuildInputs is the way
<haslersn>
catern: Do you realize that Nix is trying to prevent what you are trying to do?
<catern>
yes, that's why I said "trivial easy hack"
<{^_^}>
[nixpkgs] @jtojnar pushed 136 commits to gnome-3.30: https://git.io/fAMiL
<haslersn>
catern: I don't know of any transitive way, but I'm new to Nix ^^
<catern>
gchristensen: dumping env will work for the level-1 case but won't work if I want to do it recursively (dump the whole tree of build inputs), actually
reinzelmann has quit [Quit: Leaving]
kreetx has joined #nixos
<haslersn>
I think you could roll it yourself: Write a recursive function collecting all the dependencies into a list and make buildEnv { path = that-list; ... }
<srhb>
catern: You might take inspiration from the haskell lib for this
<catern>
yes, I'm asking if something like this already exists
<srhb>
catern: There is a shellFor function and a haskellPathsInDir (or something like that)
<srhb>
When combined, it can create a shell that has all the dependencies of all the packages in a directory
<catern>
Interesting, interesting...
<srhb>
The approach is.. Get all the directories that have .cabal, run callCabal2Nix on those, feed the result to shellFor
<srhb>
Pretty simple conceptually :)
<srhb>
And if it already has a default nix, skip the call-stuff.
<catern>
Oh, that last bit is what I want :)
<{^_^}>
[nixpkgs] @globin merged pull request #46655 → CodiMD: init at 1.2.0, add module & test → https://git.io/fAXeu
<catern>
Since effectively I'm talking about a scenario where each package always has a default.nix
<srhb>
catern: ok, grep for haskellPathsInDir and shellFor in a recent nixpkgs :)
<srhb>
Should get you going.
civodul has joined #nixos
<infinisil>
catern: there is something like this, hold on
kreetx has quit [Ping timeout: 272 seconds]
<catern>
what my true purpose is here, is, as I said earlier, I'm trying to integrate Nix with an existing build/deployment system called SWMI, which has its own notion of packages; I want to be able to depend on Nix derivations from those SWMI packages, and use the Nix derivations at SWMI-package build time, and also deploy the Nix derivations alongside the SWMI-packages at run time
<catern>
(it's a proprietary system)
captain4219 has joined #nixos
<captain4219>
Αllаһ іs doiᥒg
captain4219 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<infinisil>
closePropagation is like a manual way to collect all propagated build inputs, it's used a couple times through nixpkgs (shouldn't be in deprecated.nix)
<catern>
ultimately what I want to do is replace this system with Nix entirely, so I'm thinking right now that I'll have a tree of Nix files, just like with Nixpkgs, one per SWMI package, which describe the Nix derivations that each SWMI package uses (and I can omit the Nix file if I don't use any Nix derivations in that SWMI package); that will work fine at build time I think, I'll just do the SWMI package build inside a nix-shell, but I'm not
<catern>
sure how to extend that to working at run-time
<gchristensen>
the static one is also available on your local computer as part of the install phase etc.
<rawtaz>
infinisil: a webpage like that can (and by default should) be made such that if javascript is not enabled, it behaves like the former, and if javascript is enabled, it applies the filtering feature. that is, the best of both worlds. regardless, it doesnt make sense that there's two different options pages, because people get confused when they reach them through DDG or google searches (e.g. they arrive at the former and wonder where the heck the filter t
<rawtaz>
gchristensen: that part makes more sense, for local installs
<infinisil>
Okay well actually, the filter one should work without javascript as well
<rawtaz>
i havent even checked if it does, to be honest :>
<rawtaz>
anyway, to be clear, im not complaining, but i have multiple times found it a bit confusing when i reached the non-filter page through a DDG search and was missing the filter functionality (i use that all the time, i dont think i ever searched on the non-filter page)
<infinisil>
Maybe a link from the former options page to the latter saying "click here to get to the interactive option search"
<infinisil>
would be nice
<gchristensen>
rawtaz: yeah, maybe what should be done is have the static page -- which is produced as part of the documentation generated from NixOS -- be more clearly marked as "this is documentation" or something
<rawtaz>
yeah maybe, so people have a choice, and it also clarifies that theres an intention to have two different "versions" of it, so instead of people wondering why things are inconsistent they just hopefully figure there's a reason, and click the link
<rawtaz>
gchristensen: yeah, good idea, similar note to infinisil's suggestion
<rawtaz>
something like "Click <here> to go to the web based searchable options list"
<rawtaz>
and maybe also with "This page is statically generated, please click ..."
<gchristensen>
the trouble is that link will lead to a different version of the configuration options than the static page describes
<gchristensen>
because the static page on your computer is for the specific version you have, and the website options.html is not
<v0|d>
gchristensen: much appreciated, I'm trying to build a FHS for cuda yet no luck, samples complain that they cannot find libGL libGLU etc. any ideas?
<atu>
Why is it that dhcpcd isn't part of the network-online.target?
atu is now known as adamt
<adamt>
I mean, yes, it's not easy to answer what exactly constitutes being online (but having a default route might be a fairly reasonable criteria(?)), but reaching network-online.target without waiting for the dhcp daemon just seem wrong.
<adamt>
srhb: Thanks. That's a fairly recent discussion.
<rawtaz>
infinisil: if you have a pointer to which file should be edited to just add that little text and link on the non-filter page, i can file a PR? otherwise ill file an issue, if you think its better someone else does the coding
<srhb>
adamt: Yes, it's ongoing for pretty much a year if you follow all the threads. I think there will be an inevitable networking redesign. :)
vmandela has quit [Remote host closed the connection]
tzemanovic has joined #nixos
vmandela has joined #nixos
<adamt>
srhb: Yeah, sounds like it. :-)
mayhewluke has quit [Ping timeout: 240 seconds]
<srhb>
adamt: I also recently had to fix some surprises dhcpcd (I think) could end up in a race with predictable interface renaming despite the dependence on various units. Fun.
vmandela has quit [Remote host closed the connection]
<infinisil>
rawtaz: Yeah I have no idea
<srhb>
Let's just go back to a linear init model. :-)
<srhb>
Much easier.
vmandela has joined #nixos
<rawtaz>
infinisil: ok :)
haslersn has quit [Ping timeout: 252 seconds]
mayhewluke has joined #nixos
<johanot>
lets make a new target along the lines of "google-online.target"
<srhb>
haha
cgay28 has joined #nixos
<cgay28>
Aⅼⅼah іs ⅾoⅰng
cgay28 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<rawtaz>
no he isnt
<rawtaz>
actually last i heard Allah was a woman
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
acarrico has joined #nixos
<exarkun1>
don't engage
<rawtaz>
exarkun1: no worries, im one of the first to ignore trolls
<rawtaz>
it's the only sane solution
<jasongrossman>
/leave
<greymalkin>
What's the "default" version of ghc/cabal that's installed when using emacsWithPackages [ haskell-mode ]? I'm constantly getting "Unexpected response from haskell process" when trying to compile in emacs, and I'm about at my wit's end with finding errors on the command line.
<infinisil>
greymalkin: I don't think it installs ghc/cabal
vmandela has quit [Remote host closed the connection]
vmandela_ has joined #nixos
<{^_^}>
[nixpkgs] @exarkun opened pull request #46784 → Correct Tor config "transparent" typo → https://git.io/fAMDC
<infinisil>
greymalkin: I think it uses whatever you have in your PATH / exec-path
phreedom has quit [Ping timeout: 256 seconds]
bitmapper has joined #nixos
phreedom has joined #nixos
<greymalkin>
In that case, I'm actually still using ghc 8.0.2 -- and the only recent incarnations of this error I can google up deal mostly with >= 8.2. Has anyone else dealt with this problem?
<infinisil>
Well there should be a way to actually look at the error, "unexpected response from haskell process" is rather general. Maybe it's displayed in some buffer
<{^_^}>
[nixpkgs] @xeji pushed commit from @danbst to master « zoom-us: 2.3.128305.0716 -> 2.4.129780.0915, QT downgrade (#46778) »: https://git.io/fAMyY
<LnL>
would be nice if hydra keeps track of the actual username
<copumpkin>
yup
<greymalkin>
infinisil: I agree, but unfortunately, neither *Messages* nor *haskell-process-log* give further details.
Rusty_1 has quit [Quit: Konversation terminated!]
<infinisil>
greymalkin: Maybe look at the source code to see how it calls ghc and try to reproduce it in the terminal
<infinisil>
(Or maybe there's some process monitoring trick that lets you see arguments of very short-lived commands)
<{^_^}>
[nixpkgs] @xeji pushed commit from @danbst to release-18.09 « zoom-us: 2.3.128305.0716 -> 2.4.129780.0915, QT downgrade (#46778) »: https://git.io/fAMyP
<greymalkin>
Huh... according to https://github.com/haskell/haskell-mode/issues/1553 It seems that it's mostly related to the parsing of the "modules loaded" lines -- the cases there use digits, but my process log is showing "two" instead of "2"
<{^_^}>
haskell/haskell-mode#1553 (by ruomad, 1 year ago, open): Parsing problem with ghc-8.2.1
<greymalkin>
Okay, looks like upstream. I'll see myself out.
<{^_^}>
[nixpkgs] @globin pushed commit from @exarkun to master « nixos/tor: Correct "transparent" typo »: https://git.io/fAMSf
<rawtaz>
"amirite" that filesystems.?.options does not automatically include "defaults" when i set e.g. options=["foo=bar"] - i do need to include "defaults" myself if i want to keep it?
orivej has joined #nixos
ma27 has quit [Quit: WeeChat 2.2]
lawlesseel has quit [Remote host closed the connection]
lawlesseel has joined #nixos
phreedom has quit [Ping timeout: 256 seconds]
<srhb>
rawtaz: Yes. Option values are merged by the highest priority. Default values have a lower priority than the value you are setting
phreedom has joined #nixos
<rawtaz>
srhb: where can i see the priority values for different options and values?
<srhb>
rawtaz: Usually default values are created with mkDefault which has a set value for the priority
<srhb>
rawtaz: mkDefault = mkOverride 1000;
<srhb>
rawtaz: lib/modules.nix
<srhb>
and mkOptionsDefault as well.
tzemanovic has quit [Ping timeout: 252 seconds]
<rawtaz>
interesting, thank you
<srhb>
rawtaz: And bt the way, you can get the default value (if it's in the option declaration) with options.foo.option.default
<srhb>
rawtaz: So you can use it in your new value.
<{^_^}>
[nixpkgs] @aanderse opened pull request #46785 → php: added argon2 option for php 7.2+ → https://git.io/fAMHY
<rawtaz>
oh, so like options = [ options.something.option.default, foo=bar ] instead of options = [ "defaults", foo=bar ] ? or maybe options = options.something.option.default ++ [ foo=bar ] ?
<v0|d>
gchristensen: make EXTRA_CCFLAGS="-I/usr/include" solvd the problem.
<LnL>
aaah
<gchristensen>
v0|d: ah! want to send a PR?
<LnL>
impurities!
<v0|d>
sure.
<v0|d>
gchristensen: why the repo is not on wiki or anywhere else, if I may ask?
<v0|d>
I rlly stumbled upon it by accident.
<gchristensen>
I made it a long time ago, before the wiki was editable
<v0|d>
ah, see.
<v0|d>
do you have any easy way to omit this EXTRA_CCFLAGS?
<gchristensen>
no, I don't do anything with CUDA / GL / etc anymore
<v0|d>
I have added /lib to LD_LIBRARY_PATH in fhs env.
<v0|d>
do I rlly have to this like this?
<v0|d>
or are there any better ways? (this is not rlated to cuda I guess)
<v0|d>
it searches for libGLU etc and they are in /lib.
<v0|d>
I do export LD_LIBRARY_PATH=${pkgs.linuxPackages.nvidia_x11}/lib:/lib in fhs atm.
<teto>
I am trying to bump gtkspellmm from 3.0.4 to 3.0.5 . I just changed the version number/sha yet I can't download it. The archive seems available https://sourceforge.net/projects/gtkspell/files/gtkspellmm/ . I've been trying to download mirror://sourceforge/project/gtkspell/gtkspellmm/gtkspellmm-3.0.5.tar.xz from terminal but curl doesn't work with mirror:// . What's the trick ?
drakonis has quit [Read error: Connection reset by peer]
<tilpner>
manveru - Can you describe how your wishlist item differs from Nix's current concept of a derivation?
andrewrk has joined #nixos
<manveru>
well, we could start by `mkPackage = mkDerivation`
<manveru>
i don't think derivations are bad, just not very easy to understand for newcomers...
<worldofpeace>
^
adamt has quit [Ping timeout: 252 seconds]
<manveru>
the addition of `mkShell` was really nice in that regard
melleb has quit [Ping timeout: 272 seconds]
<alex``>
I noticed there is no Ruby / Python / pip packages
<alex``>
nor npm
<manveru>
yeah
<alex``>
What is the Nix way?
<alex``>
to get them
<manveru>
bundix/pypi2nix/yarn2nix, etc
erasmas has joined #nixos
Ariakenom has quit [Ping timeout: 245 seconds]
<v0|d>
gchristensen: sent PR.
<alex``>
After installing bundix, I will have ruby / irb / gem binaries?
<tilpner>
manveru - Users coming from other package managers associate certain properties with a "package", properties one wouldn't find in the build result of writeText for example. In that regard, introducing a new term ("derivation") may have reduced conflicts of expectation and overall confusion
<manveru>
tilpner: i guess you're right, since derivations are still functions, while packages in other managers don't take arguments...
<tilpner>
manveru - I'm not against introducing a nixpkgs-level concept named "package", but one should be very careful not to deviate too much from what packages look like in other systems. A mkPackage could indeed just be a mkDerivation that adds a tag for nix tools to use
<teto>
Problem seems to be that jaist.dl.sourceforge.net is listed in the sourceforgejp mirrors and not sourceforge mirrors so I never get my package when running nix-build -A gtkspellmm ~/nixpkgs
<manveru>
i just think calling `nix` a "package manager" is funny, since when people start using `nix-env --install` like they'd use `apt-get install`, they usually run into trouble at some point
<teto>
I beleive merging sourceforgejp and sourceforge mirrors should fix it but not sure how that works
lassulus has quit [Ping timeout: 240 seconds]
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
mizu_no_oto has joined #nixos
lassulus has joined #nixos
<v0|d>
are there any docs relatd to cross compiling to windows on nixos?
<lesh>
hey, I'm exporing nixops for managing some servers, what would be a proper way to deliver my own static files, apart from packages and system configurations to machines?
<lesh>
would I use a separate channel for this, or should I build custom nixos packages?
Dedalo has joined #nixos
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
cement has joined #nixos
<clever>
lesh: depends on where the files need to be, what they are used for, and if they contain any secrets
<lesh>
clever: let's say I'm deploying a static website, what would be the invision method for transfering these static files? or for example a nodejs app, I guess that would be a custom package?
<lesh>
invisioned*
<clever>
lesh: i would do something like services.nginx.virtualHosts."www.example.com".root = ./html_root;
<clever>
nix will then copy the entire html_root dir over, and configure nginx to use that copy
wpcarro has quit [Ping timeout: 252 seconds]
<lesh>
aha you'd keep it on the management machine, I see
<lesh>
what about nodejs app or like running a luks encrypted container (via /dev/loop)? would this entail custom packages? (given there isn't already some high level package I can configure to do these things)
<cransom>
lesh: not necessarily a custom package, but at last a custom service that would run the app/mount the encrypted file system
<lesh>
(just trying to get a feel on how you'd cover different scenarios with nixops)
periklis has joined #nixos
simukis has quit [Quit: simukis]
abueide has joined #nixos
<lesh>
aha via systemd.services, I see, tnx!
wpcarro has joined #nixos
<rawtaz>
so i notice that my / is listed in hardware-configuration.nix. now i want to add additional fstab entries. whats the syntax for adding this to the configuration.nix file and not replace but append to the fileSystems hash?
<gchristensen>
it will add by default
<clever>
rawtaz: a: nixos always merges them
<clever>
rawtaz: b: mount it manually, and re-run nixos-generate-config to update hardware-configuration.nix
<rawtaz>
clever: but why would i want to keep this in hardware-configuration, when i have all of my other stuff in configuration.nix?
<rawtaz>
seems uncomfy to have it spread out
<clever>
mostly personal preference, i just ditch hardware-configuration.nix entirely
<v0|d>
rawtaz: more than one machine, copy configuration.nix, but not hardware.nix.
<rawtaz>
for new installs it would make sense (IMO) to let hardware-configuration be generated, so you can leave it there and it just covers the absolute basics to get the minimal system running, then you complement with stuff in configuration.nix
periklis has quit [Ping timeout: 252 seconds]
<rawtaz>
v0|d: exactly :)
<cransom>
i usually merge all of the auto-generated options from hardware-configuration.nix if i have a very long running system. if it's a lab machine where i switch stuff up a lot, i don't bother and let it regenerate if i install again
<rawtaz>
i see. well im looking to keep configuration.nix in my dotfiles repo
<rawtaz>
regardless, thanks again for your valuable help, ill add it to configuration.nix
tzemanovic has quit [Ping timeout: 264 seconds]
periklis has joined #nixos
<teto>
nevermind the problem stem from the change in extension tar.gz became tar.xz and I failed to notice :'(
<xok>
clever: thank you very much, it worked just fine...
sigmundv__ has quit [Ping timeout: 246 seconds]
patrl has joined #nixos
patrl has quit [Client Quit]
polman has joined #nixos
bitmapper has joined #nixos
Thra11 has joined #nixos
haslersn has joined #nixos
philippD has joined #nixos
trevthedev has quit [Ping timeout: 252 seconds]
sbdchd has quit [Remote host closed the connection]
<teto>
to address "Settings schema 'org.gnome.gimagereader' is not installed", is wrapGappHook enough/the solution ?
<shreyansh_k>
Hi guys, where can I find docs for the "imports" statement?
jperras has joined #nixos
<{^_^}>
[nixpkgs] @ju1m opened pull request #46792 → nslcd: restart when nslcd.conf changes and let pwdutils tools use libnss_ldap.so → https://git.io/fADti
<samueldr>
(in that case it's not a statement, it's a value of the attrset, like the other options, but treated differently in the modules)
phreedom_ has joined #nixos
<rawtaz>
better read what samueldr suggested.
<samueldr>
rawtaz: depends on whether it's import (without s) or imports (with an s), but yours is an accurate explanation for import (without an s)
<rawtaz>
that clarifies it :)
<{^_^}>
[nixpkgs] @globin pushed commit from @aanderse to release-18.09 « php: add option to enable argon2 »: https://git.io/fADqk
<shreyansh_k>
Ok! Thank you. I'm looking at these things.
<samueldr>
to be less confusing (though annoying to type) the modules system one could have been called `otherModulesToImportAndMerge` ;)
orivej has joined #nixos
<rawtaz>
hehe yeah
<exarkun1>
is it reasonable to want nixops to be able to generate a Tor Onion Service key for the Tor Onion Services I am deploying with it? It looks like maybe it's just letting Tor generate a key (which will then be lost if I lose the machine).
<samueldr>
then yes, shreyansh_k, it should act accordingly to "34.6 Importing Modules"
ixxie has quit [Ping timeout: 252 seconds]
<xok>
does anyone know what should I do put theere?..
<xok>
the localhost didn't work...
trevthedev has joined #nixos
<rawtaz>
so multiple filesystem options can be specified as [ "rw" "nodev" ] , but is there a more proper way or is that the right syntax? no , i suppose?
trevthedev has quit [Changing host]
trevthedev has joined #nixos
<samueldr>
shreyansh_k: any particular question? (maybe there's something missing or you have an issue with something else)
<samueldr>
srhb: as far as I'm concerned, no reasons
<srhb>
samueldr: OK, thanks. :)
<samueldr>
possibly thinking that backports were strictly for software revisions at that point in time
<srhb>
How time flies... :D
reinzelmann has joined #nixos
vmandela has quit [Remote host closed the connection]
vmandela has joined #nixos
vmandela has quit [Client Quit]
phreedom_ has quit [Ping timeout: 256 seconds]
winem_ has quit [Ping timeout: 246 seconds]
kreetx has joined #nixos
trevthedev has quit [Quit: WeeChat 2.0]
haslersn has quit [Ping timeout: 252 seconds]
<exarkun1>
How do I manage an arbitrary file on the filesystem with nixops?
melleb has joined #nixos
melleb has quit [Changing host]
melleb has joined #nixos
<infinisil>
exarkun1: Define "manage"
ambro718 has joined #nixos
<exarkun1>
specify the contents of
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Tobba has joined #nixos
<judson>
I stumble over this occasionally and am wondering where would be best to document it:
derped has joined #nixos
<tilpner>
exarkun1 - systemd.tmpfiles or something likes nur.repos.tilpner.modules.files
<judson>
A nix derivation puts files into $out. nix-env builds a link forest out of $out files, but which files and how?
<tilpner>
I have been meaning to port that module to tmpfiles, to see if it'd work well
<judson>
(There's a whole project around "asking clever on IRC is not a documentation strategy :) )
waleee has quit [Quit: WeeChat 2.2]
<exarkun1>
tilpner: systemd.tmpfiles doesn't seem to be the sort of thing I mean. What's `nur`?
<judson>
In this case, I'm trying to set up a package for a server app. I only really expect it to get run by systemd via NixOS. Is it appropriate to set it up such that no files would be "installed" by nix-env, and how best can I accomplish that?
<exarkun1>
tilpner: My mistake. I accidentally read the documentation for "systemd.tmpfiles.rules" on https://nixos.org/nixos/manual/options.html and it seemed unrelated. I don't see any documentation for "systemd.tmpfiles" at all?
<symphorien>
man tmpfiles.d
sigmundv has joined #nixos
<exarkun1>
I don't see how systemd-tmpfiles is related. I want to specify that a file exist and that it have certain contains. `nur.repos.tilpner.modules.files` seems to be about right but why is it "unsafe"?
<symphorien>
The systemd.tmpfiles option precisely configures which files and directories systemd-tmpfiles will create
<symphorien>
Every string in the list is a line of the configuration file
alfie has quit [Ping timeout: 246 seconds]
<tilpner>
exarkun1 - Let me assure you, systemd.tmpfiles it is not unrelated. "F /home/exarkun/foobar 0755 exarkun users - quux" will create a file in your home directory with the specified permissions and content
<exarkun1>
I don't care about systemd temporary files, I don't understand how that is related my question at all.
<tilpner>
exarkun1 - It can also be used to create a symlink into a Nix-generated store file
ris has joined #nixos
<exarkun1>
Uh ... ok
rihards has joined #nixos
<tilpner>
exarkun1 - The problem with my module is that you have to be careful using it. I have not had any problems with it, but that's because I wrote it. If you're aware of how it works, you can do damage to your installation
milhouse13379 has joined #nixos
<tilpner>
There are no safeguards
<milhouse13379>
Alⅼаh is doinɡ
milhouse13379 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
Burrr15 has joined #nixos
<Burrr15>
ѕun iѕ not doіᥒɡ Alⅼаһ іѕ ԁoinɡ
<Burrr15>
mooᥒ is ᥒot doing Ꭺⅼⅼah iѕ doiᥒɡ
<Burrr15>
Alⅼɑh iѕ ԁⲟiᥒg
Burrr15 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<exarkun1>
reading the systemd tmpfiles documentation more carefully now
<tilpner>
TBF, I have not tested if "F /etc/passwd ..." does what it says it does
<tilpner>
It might be similarly dangerous to abuse
Izorkin has quit [Read error: Connection reset by peer]
sigmundv has quit [Ping timeout: 252 seconds]
Anton-Latukha has joined #nixos
Izorkin has joined #nixos
alfie has quit [Ping timeout: 252 seconds]
tzemanovic has joined #nixos
johanot has joined #nixos
orivej has quit [Ping timeout: 252 seconds]
<johanot>
if the python builder says: invalid command 'bdist_wheel', what is missing? Whell is missing in the build environment?
<johanot>
wheel*
alfie has joined #nixos
humanoyd has quit [Quit: WeeChat 2.2]
<Mic92>
yes, wheel is a python package
Neo-- has quit [Ping timeout: 252 seconds]
<johanot>
Mic92: I know.. Just don't understand why it's missing at the build stage. I tried supplying python.pkgs.wheel as buildInput. doesn't work :(
alex`` has joined #nixos
<Mic92>
johanot: try format = "wheel"; in case you are using buildPythonPackage or *Application
<exarkun1>
I tried to generate a private key using https://nixos.org/nixops/manual/#idm140737316403040 to link to with the systemd.tmpfiles features but the private key is empty...? It seems like nothing is generated but the deploy runs to completion and indicates success, creating an empty file where I told it the key should be.
<Mic92>
johanot: wait a sec, this is incorrect
<exarkun1>
`resources.sshKeyPairs.x = { };` and later `text = resources.sshKeyPairs.x.privateKey`... What'd I miss?
<Mic92>
johanot: do your python version matches?
<johanot>
Mic92: format = "wheel" actually worked.. I didn't wait, sry :P
<Mic92>
fair enough
sanscoeur has joined #nixos
<Mic92>
johanot: but did it the right thing? According to the code I see it just unpacks stuff.
alfie has quit [Ping timeout: 276 seconds]
<johanot>
Mic92: Before it failed to build. now it runs pass the buildPhase and fails in the install phase..
<johanot>
"you must give at least one requirement to install"
bitmapper has quit [Remote host closed the connection]
alfie has quit [Ping timeout: 244 seconds]
rauno has joined #nixos
reinzelmann has quit [Quit: Leaving]
cement has quit [Ping timeout: 245 seconds]
abueide has joined #nixos
Ericson2314 has quit [Quit: Leaving]
<rawtaz>
clever: uh, ok, so what can i do about nixos-rebuild test/switch throwing me into emergency mode? all i did was add a fileSystem entry in configuration.nix.
Ericson2314 has joined #nixos
orivej has quit [Ping timeout: 264 seconds]
jedahan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Ericson2314 has quit [Changing host]
Ericson2314 has joined #nixos
<clever>
rawtaz: check the journal, systemd failed to mount the fs, and it assumes the fs is mission critical
Growpotkin has quit [Ping timeout: 244 seconds]
reinzelmann has joined #nixos
<rawtaz>
im not seeing an option i can specify in configuration.nix to say that this FS should be mounted after everything else and/or that it's not critical or so on..
<rawtaz>
when i try to boot this configuration, it hangs instead of going into emergency mode
<gchristensen>
use the nofail option
<rawtaz>
it's a vmhgfs filesystem ive added
<rawtaz>
nice, looking it up
alfie has joined #nixos
kreetx has quit [Ping timeout: 252 seconds]
dmc has quit [Quit: WeeChat 2.2]
<rawtaz>
gchristensen: that solved the emergency thing
<rawtaz>
cheers
<rawtaz>
when i boot it fails to mount the vmhgfs though, hmm
<rawtaz>
seems fstab doesnt like the .host:/ i put there
rauno has quit [Ping timeout: 252 seconds]
<rawtaz>
even though that syntax works fine in other distros apparently
<exarkun1>
is there a base64 decoder in builtins or nixpkgs somewhere?
<shreyansh_k>
samueldr: Thank you for asking. I'm trying to understand how Nix uses the configuration.conf file to generate config and I think I'm good for now.
<symphorien>
I think nix strings can't contain nul bytes, so it seems unlikely
<exarkun1>
ugh. is nul the only byte forbidden? or is it more like "printable ascii only"?
Rotab20 has joined #nixos
<Rotab20>
Аlⅼɑh ⅰs doing
Rotab20 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
dmc has joined #nixos
ixxie has quit [Ping timeout: 272 seconds]
ixxie has joined #nixos
<symphorien>
exarkun1: you can try by reading binary files with builtins.readFile
jedahan has joined #nixos
reinzelmann has quit [Quit: Leaving]
matthewbauer has joined #nixos
<exarkun1>
I just tried reading a binary file with builtins.readFile, that's why I started wondering about using base64 to smuggle arbitrary bytes through
<exarkun1>
but if you can't even represent arbitrary bytes with a nix string I don't see how you could make it work
<symphorien>
what are you trying to achieve ? do you need the binary string in a nix string ?
<symphorien>
could a derivation containing the string be enough ?
<exarkun1>
the binary string is part of a secret I'm trying to use `deployment.keys...` to manage
<symphorien>
ah
Dedalo has joined #nixos
<symphorien>
your best bet is probably to have the end user application use the base64 secret instead of a binary form then
<exarkun1>
if I were in the business of going around and rewriting all terrible software in the world then I probably wouldn't begin with this one
<exarkun1>
nixops started throwing `xml.etree.ElementTree.ParseError: not well-formed (invalid token): line 320, column 37` at me. where is there even any xml to be parsed?
civodul has joined #nixos
timokau_ has joined #nixos
<Mic92>
exarkun1:this is internally used between nix and python
<Mic92>
could be maybe replaced with json these days.
mulderr has joined #nixos
<exarkun1>
so this is a nixops bug?
wpcarro has quit [Remote host closed the connection]
<Mic92>
exarkun1: for your interestest, nix uses null-terminated strings. So it is treated more like bytes then printable ascii characters.
<romildo>
tilpner, this is not dual license. According to the nixpkgs manual it express that (different) parts of the project are licensed differently.
mkoenig has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @vaibhavsagar opened pull request #46803 → all-cabal-hashes: update snapshot to Hackage at 2018-09-12T08:26:27Z → https://git.io/fADEF
<tilpner>
romildo - Then I guess look at some software you know to be packaged in nixpkgs and dual-licensed, and see how it was done for that?
<tilpner>
"Rust is primarily distributed under the terms of both the MIT license and the Apache License (Version 2.0), with portions covered by various BSD-like licenses."
<tilpner>
I think this is what you call dual-licensed?
<romildo>
Yes.
realrokka has joined #nixos
matthewbauer has quit [Ping timeout: 250 seconds]
<tilpner>
Then that's your precedent. If you think that's wrong, go file an issue and ping people who know more, but if you're just looking for an example... I'd say try it like this (if you're doing a PR). Maybe also draw attention to this, if it's a PR
<romildo>
But this rustc derivation seems to use a list of licenses as you first suggested. That does not expressions that one can choose one or the other.
<{^_^}>
[nix] @ariutta opened pull request #2429 → Upgrading docs: command -> screen & more info link → https://git.io/fADur
mkoenig has joined #nixos
<globin>
romildo: our meta does not distinguish between those two options
<romildo>
If one of the licenses in the list is unfree. nix consider that the software is unfree.
<tilpner>
Relevant: aseprite has license = if unfree then licenses.unfree else licenses.gpl2;
<{^_^}>
[nixpkgs] @Ericson2314 pushed 2 commits to release-18.09: https://git.io/fADuS
<rawtaz>
grr. theres something seriously broken with open-vm-tools in nixos. the vmware service is running, yet when you try to mount stuff through fstab you get "No such device" for the .host:/ entry in there
<rawtaz>
been looking at it now for a while and having a hard time pinpointing what's causing the problem.
<romildo>
I maintain a package (https://github.com/shimmerproject/Greybird). Greybird is dual-licensed as GPLv2 or later and CC-BY-SA 3.0 or later. The first is free, and the second is unfree in nixpkgs. Currently both licenses are listed in the package derivation. And nix takes that it is unfree.
<rawtaz>
listing the shared folders from within the VM works, *mounting* the shared folders from within the VM works when doing it manually. it's when it's in fstab that it doesnt work >_<
<{^_^}>
undefined variable 'viewSoruce' at (string):194:12
<tilpner>
> viewSource viewSource
<{^_^}>
value is a function while a set was expected, at (string):175:32
<tilpner>
Aww, only derivations?
sir_guy_carleton has joined #nixos
<infinisil>
tilpner: All attrs in pkgs
<ldlework>
I forget who it was, probably sphalerite, who wagered that like, polling window-manager-bar-widgets would probaby contribute to battery loss and other effects, and i wagered probably not to the extent that you'd ever notice
<infinisil>
tilpner: There's also the bot command :v to view the source of a definition in the bot:
<ldlework>
i think I definitely have noticed my laptop running hotter, the fans spinner more, and the battery seeping quicker - with just two widgets running at 0.5 so, yeah, good wager
<romildo>
tilpner, I have done that. But I would like to know if there is any way to express the idea of dual licensing in nix.
<tilpner>
romildo - I don't know one. Drawing attention to an issue describing your problem might help
<infinisil>
ldlework: I once polled 10 times a second from a script to put something up in the status bar. For some reason it was very very laggy. Turns out I used a nix-shell shebang in the script, and nix-shells are notoriously slow to start..
sanscoeu_ has joined #nixos
kreetx has quit [Ping timeout: 244 seconds]
fenedor has joined #nixos
<ldlework>
one widget was just using pgrep
<ldlework>
the other was uh
<ldlework>
using playerctl to talk to spotify
<ldlework>
they both used python to do the job
<ldlework>
since it is qtile the window manager
<ldlework>
i bet all the cost is in qtile unfortunately
<ldlework>
anyway, just wanted to swing back on that to lose the wager :) (i disabled them and now my laptop is cool as a kitten )
sanscoeur has quit [Ping timeout: 250 seconds]
tzemanovic has joined #nixos
jedahan has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<rawtaz>
can someone take a look at https://pastebin.com/HzYR6aFB and see what they think can be done about it? or should i just create an issue for this?
abueide has joined #nixos
<infinisil>
,tofu alex``
<{^_^}>
alex``: To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
<{^_^}>
vmware/open-vm-tools#199 (by kbreit, 49 weeks ago, open): Debian 9 - VMware Fusion 10 won't work with shared folders
<cransom>
or also `nix-prefetch-url -A kakoune.src` in your nixpkgs checkout.
<ogle>
I am trying to get mesos working on my nixos machine with gpu support, but running into this issue. "Failed to create a containerizer: Failed to NvidiaVolume::create: Failed to ldcache::parse: Failed to open file: No such file or directory" I suspect this is because nix does not have and ld.cache.so. And sure enough I see this in the mesos src - ./src/linux/ldcache.hpp:Try<std::vector<Entry>> parse(const
<ogle>
std::string& path = "/etc/ld.so.cache");
<ogle>
is there a way I can fake this?
alfie has quit [Quit: Bye.]
wpcarro has quit [Remote host closed the connection]
<cransom>
rawtaz: does vmhfq-fuse create a `/mnt/hgfs`? mount doesn't do that.
<cransom>
*vmhgfs rather.
<rawtaz>
cransom: there is already a /mnt/hgfs created
<rawtaz>
it just needs to mount it there
<rawtaz>
im gonna try the things in that issue i linked to above
cement has joined #nixos
<romildo>
Is there any golang developer or package maintainer here? I am needing help in package some software from deepin.
phiofx has joined #nixos
<phiofx>
Ꭺⅼⅼaһ iѕ doⅰᥒg
phiofx has quit [Killed (Sigyn (Spam is off topic on freenode.))]
wpcarro has joined #nixos
<ogle>
anyone have ideas regarding ld.cache.config?
<ogle>
correction ld.cache.so
wpcarro has quit [Remote host closed the connection]
kisik21 has quit [Ping timeout: 246 seconds]
mayhewluke has quit [Ping timeout: 252 seconds]
xok has quit [Ping timeout: 252 seconds]
mayhewluke has joined #nixos
<alex``>
infinisil: But how to tweak the Kakoune package?
<infinisil>
alex``: Not sure what you mean by "tweak"?
<alex``>
modify
wpcarro has joined #nixos
wpcarro has quit [Remote host closed the connection]
wpcarro has joined #nixos
xok has joined #nixos
sanscoeu_ has quit [Remote host closed the connection]
flokli has joined #nixos
wpcarro has quit [Ping timeout: 244 seconds]
tzemanovic has quit [Ping timeout: 252 seconds]
<rawtaz>
does anyone know how to get the path of a unit file from systemctl?
<tilpner>
systemctl cat shows it as the first line
mkoenig has quit [Read error: Connection reset by peer]
<rawtaz>
perfect answer tilpner !
<{^_^}>
[nixpkgs] @Ma27 opened pull request #46806 → nixos/testing: disallow `-` as machine names in network expressions → https://git.io/fADwk
<rawtaz>
oops, just found an inconsistency - there's one option with z in virtualization, all the other virt options are with s (virtualisation)
camsbury has quit [Remote host closed the connection]
<srhb>
rawtaz: Perfect opportunity for a pull request! :D
<rawtaz>
srhb: yep, if i had any idea what to change :D i suppose i should simply look at the .nix file that is referenced in/for that option, or could there be other places too?
<rawtaz>
srhb: that said, it should probably be with z, not s.. :P
<srhb>
rawtaz: For that sort of thing it sounds like a grep -r might find things for you :)
<srhb>
rawtaz: I think we're trying to use 's' consistently
<srhb>
Emphasis *think*
* rawtaz
wonders if it's an english (s) / american (z) thing
<rawtaz>
srhb: imagine i did change z to s - wouldnt that be considered breaking BC?
<srhb>
rawtaz: Yep. :) But we have a renaming strategy for that
jperras has quit [Ping timeout: 250 seconds]
<srhb>
rawtaz: nixos/modules/rename.nix
Thra11 has quit [Ping timeout: 244 seconds]
<rawtaz>
oh, so that's handled automatically?
TecnoBrat16 has joined #nixos
<TecnoBrat16>
Аllɑh is ⅾoіng
TecnoBrat16 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<rawtaz>
err, let me check that
<srhb>
rawtaz: Yeah, it can generate a warning automatically and then later on the old option can be removed
civodul has quit [Quit: ERC (IRC client for Emacs 26.1)]
Ariakenom has quit [Read error: Connection reset by peer]
jperras has joined #nixos
<rawtaz>
seems to me that the latter list there represents the current name of that option, so the former is the old one?
Neo-- has joined #nixos
<rawtaz>
if this is the case, then the incorrect option name (the one with z) doesnt need changing, as it's already replaced with a new one
<srhb>
rawtaz: Here's a neat trick: Because nix functions are defined as name = arg1: arg2: ...
<srhb>
rawtaz: You can easily grep for "name =" to find its definition
<srhb>
rawtaz: So try finding mkRenamedOptionModule via that method :)
tzemanovic has joined #nixos
rob``27 has joined #nixos
<rob``27>
Αllaһ iѕ dоiᥒɡ
<rob``27>
ѕᥙᥒ is not ԁoiᥒg Ꭺⅼlah ⅰs doiᥒg
rob``27 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<rawtaz>
srhb: right, yes
<srhb>
rawtaz: (Your analysis is correct - just wanted to show you where to find it to confirm what you already suspected) :)
<rawtaz>
srhb: that's highly appreciated, teaching someone to fish is the best way forward (although it doesnt hurt to give them some fish as well, especially when they're already burning hours trying to debug stuff :)
<rawtaz>
so keep having at it :)
<srhb>
rawtaz: Belated fish for you. :P
tzemanovic has quit [Ping timeout: 244 seconds]
<rawtaz>
drippin' fish
<rawtaz>
i should get a copy of the repo, then
<rawtaz>
grab me self a clone
<JonReed>
There is no way to call shell inside the nix expression to return some value, right? Like, to use `find` command to return a string of paths it found and that string would be a nix string.
<srhb>
rawtaz: Yeah, that's very useful, also for testing changes :)
lopes has joined #nixos
<srhb>
JonReed: See IFD and runCommand
<srhb>
find will be a problem though
<srhb>
But we have readDir
<rawtaz>
srhb: can you check real quick how big the repo is?
<srhb>
JonReed: The thing is, you can only operate on paths within the nix store anyway. So you will need to import whatever path in order to search through it. That's doable though, as long as it's something like a source tree.
<srhb>
JonReed: Not so much if it's your entire system.
<rawtaz>
srhb: are you one of the core devs? how many core devs are there anyway?
camsbury has joined #nixos
<srhb>
rawtaz: Um, I don't think I'd classify myself as "core" but I am a maintainer of nixpkgs. And we're a whole bunch, I think you can see the list on github.
<{^_^}>
JonReed: import-from-derivation (IFD) is when you evaluate nix from a derivation result, for example `import (pkgs.writeText "n" "1 + 1")` will evaluate to 2
wpcarro has joined #nixos
hiroshi has joined #nixos
lopes has quit [Quit: WeeChat 2.2]
lopes has joined #nixos
wpcarro has quit [Remote host closed the connection]
<JonReed>
srhb: Thanks.
lopes has quit [Client Quit]
wpcarro has joined #nixos
ixxie has quit [Ping timeout: 245 seconds]
Rusty1_ has joined #nixos
<JonReed>
srhb: IFD will create something in store? Or will it just create a temporary out which will be automatically deleted after the expr will be evaluated?
jedahan has joined #nixos
<srhb>
JonReed: In the example above, the n file _will_ be written to store
<srhb>
If there's no gc root, it will be deleted when you garbage collect.
<srhb>
It's treated as a normal build dependency in that regard.
<infinisil>
ryantm: You can use `options.foo.highestPrio` and if that's lower than 1000, it's (probably) user-set
bydo1 has joined #nixos
<bydo1>
Αllɑh iѕ doing
bydo1 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
realrokka has joined #nixos
<infinisil>
catern: Not sure what would happen then
andrewrk has left #nixos ["Leaving"]
<rawtaz>
ok guys, i have a detail question.. in my fstab i now have a filesystem with fsType=fuse and device=vmhgfs-fuse . this is all proper. systemd-fstab-generator then generates a unit file for this point, but when that unit is started, i get this error: mount[6776]: /bin/sh: vmhgfs-fuse: command not found
<catern>
infinisil: what about for Python?
<catern>
If I have two different versions of a Python package in my propagatedBuildInputs, what happens?
<ryantm>
I want to force option A to be true only if option B is true. Do I need to use mkIf and mkMerge along with mkForce?
<samueldr>
rawtaz: systemd units start without PATHs
<rawtaz>
it looks to me like a shell script is trying to run vmhgfs-fuse, but fails to find that command. if i run vmhgfs-fuse myself in the shell, it works fine. so, theres a path issue
<samueldr>
it's by design, so even if it's in systemPackages, it wouldn't work
fenedor has quit [Read error: Connection reset by peer]
<rawtaz>
samueldr: yeah.. so, the question becomes, how can we fix this? it looks to me like an incompatibility between nixos and systemd, so to speak. we must somehow make systemd run the full path to the command or whatever is needed to get to the correct binary
<infinisil>
ryantm: mkIf b { a = true; } should work
<Mic92>
or systemPackages
<rawtaz>
samueldr: if i can get this to work i think im on the right track to fixing the shared folder mounting problem in nixos under vwware fusion
<Mic92>
rawtaz: you can extend the path of a systemd unit
<rawtaz>
samueldr: cant be the first time something needed to be run by systemd :)
<samueldr>
if you were to add `path = [ pkgs.hello ]` to the service definition, hello would be made available to the service
<rawtaz>
ok so i should do e.g. systemd.services.?.path = open-vm-tools or something like that
<samueldr>
(as documented, it adds the bin and sbin subdirectories of packages)
<rawtaz>
yeah
<samueldr>
rawtaz: almost, a list of, so [ open-vm-tools ]
<rawtaz>
ah, right
wpcarro_ has quit [Remote host closed the connection]
wpcarro has joined #nixos
<rawtaz>
a related question; to do this, one needs to know the service name. when you do e.g. fileSystems."/mnt/hgfs" = { .. } the service name becomes mnt-hgfs.mount - i.e. the service name is dynamic and depends on the target mount path of your filesystem. i suppose the cleanest would be if the services.vmwareGuest stuff somehow looked through fileSystems and added entries for systemd.services.?.path as needed?
<rawtaz>
(for vmhgfs and/or fuse filesystems, of course)
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<rawtaz>
doesnt seem great that people have to make sure to have corresponding filesystem and service names
ma27 has joined #nixos
<samueldr>
hmm, most of the more complex behaviours will be made by implementing a module
<samueldr>
so the module will then somehow manage the names
ma27 has quit [Client Quit]
<rawtaz>
yeah. in this case, this is all about vmware guest tools, which is usually enabled by services.vmwareGuest.enable=true
<samueldr>
(though, this is an unhelpful answer)
<rawtaz>
so that module should handle this, im thinking
johnnyfive has joined #nixos
<samueldr>
I don't know how the vmware stuff is handled in any situations (nixos or not) so can't answer with certitude
<rawtaz>
(i might be talking something else than modules here, i realize - but i guess you get my point, that whatever needs doing should be triggered by vmwareGuest)
<samueldr>
it's possible
wpcarro has quit [Ping timeout: 245 seconds]
<rawtaz>
yeah. well thats for later anyway
<rawtaz>
ill try your suggestion first
<Ashy>
NixOps arranges the necessary network configuration to ensure that these machines can communicate securely with each other (e.g. by setting up encrypted tunnels).
<Ashy>
Wireguard would be nice but I guess as always, patches welcome right? :P
nckx has quit [Ping timeout: 240 seconds]
zanxis has left #nixos [#nixos]
nckx has joined #nixos
<rawtaz>
i added open-vm-tools to the systemd.service.mnt-hgfs.path, and it seems to have added the open-vm-tools binaries to the service, but for some reason the unit/service still doesnt find that vmhgfs-fuse binary: https://pastebin.com/QLtQY2EC
<Dezgeg>
apparently ntfs-3g does this: <nixos/modules/tasks/filesystems/ntfs.nix>
<Dezgeg>
er, system.fsPackages = [ pkgs.ntfs3g ];
fragamus has joined #nixos
<Dezgeg>
that would apply to vmhgfs-fuse as well, I presume
<rawtaz>
hm. ill try it just for kicks. i dont understand why the service doesnt already find the binary when it's right there in the bin folder which is in its PATH though
fragamus has quit [Read error: Connection reset by peer]
<rawtaz>
nope, same symptom, "mount[13005]: /bin/sh: vmhgfs-fuse: command not found"
<Dezgeg>
try fsType = "vmhgfs";
fragamus has joined #nixos
<rawtaz>
quick question first; there's no system.fsPackages in the options list afaict
<rawtaz>
yeah but this one seems like a dead end. the fuse approach is more likely to succeed, if only i can get the unit/service to call the vmhgfs-fuse binary properly
<Dezgeg>
it is using the fuse approach
<Dezgeg>
or I would presume so
<Dezgeg>
I mean now you successfully have mount.vmhgfs in PATH
wpcarro has joined #nixos
<rawtaz>
Dezgeg: well, sure, but that simply doesnt work. i havent been able to get past the "No such device". if you have ideas on how to fix that, im all ears
<rawtaz>
Dezgeg: and also the fact that manually doing e.g. mount -t vmhgfs or mount.vmhgfs also just errors out with "No such device", while doing vmhgfs-fuse .host:/ /mnt/hgfs mounts just fine :)
tertl3 has joined #nixos
<rawtaz>
ill try that :)
mizu_no_oto has joined #nixos
haslersn has joined #nixos
<rawtaz>
no change with fsType = "fuse.vmhgfs-fuse" , same error: Sep 18 00:28:57 nix mount[16960]: /bin/sh: vmhgfs-fuse: command not found
<rawtaz>
and that file, vmhgfs-fuse, is right there in the bin folder that is in the PATH of the unit
<rawtaz>
so why the heck doesnt it execute it successfully
<Dezgeg>
huh
ma27 has quit [Quit: WeeChat 2.2]
hamishmack has joined #nixos
erasmas has quit [Quit: leaving]
ma27 has joined #nixos
<Dezgeg>
it's strange that it can somehow find mount.vmhgfs but not the other binary
<rawtaz>
Dezgeg: see line 65 in https://pastebin.com/QLtQY2EC - it shows the first entry in the PATH for the systemd unit being the open-vm-tools bin folder. and line 71 shows that folder having the vmhgfs-fuse file that the unit fails to find/execute
<rawtaz>
so afaict it shold find it, cuz it-s right there
fragamus has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
ma27 has quit [Client Quit]
<rawtaz>
im not sure but i think that perhaps mount.vmhgfs and mount -t fuse are two different things
brejoc has quit [Ping timeout: 240 seconds]
ogle has left #nixos [#nixos]
ma27 has joined #nixos
<Dezgeg>
yes, apparently they are
<Dezgeg>
but mount.vmhgfs was getting found since you got this error: Share name must be in host:dir format
<rawtaz>
thats true..
<rawtaz>
so SOMETHING that understands what vmhgfs is was actually run at that point
<rawtaz>
so far so good. the problme is that when i give it the correct format, i hit the dreaded @No such device@, which i havent found a solution to yet
<rawtaz>
and this other way is the one that the actual author of the vmware shared folders code in vmware fusion is using, so heck, it should be the right way to do it
<Dezgeg>
I think mount.vmhgfs would try to mount using the kernel driver which you probably don't have loaded
<rawtaz>
yeah
tzemanovic has joined #nixos
<rawtaz>
he said something about that, they essentially started using FUSE instead of kernel modules with kernel version 4, i think
wpcarro has quit [Remote host closed the connection]
wpcarro has joined #nixos
<rawtaz>
considering that the thread was start with "this stopped working in fusion v10, worked in fusion v8", im interpreting his reply as "we changed this in v10"
<Dezgeg>
maybe you could make a symlink mount.vmhgfs-fuse -> vmhgfs-fuse if that magically helps
<Dezgeg>
at least sshfs comes with one
<Dezgeg>
ntfs-3g does the same
matthewbauer has quit [Ping timeout: 240 seconds]
<rawtaz>
i guess i would have to do that inside the unit-s path
<rawtaz>
but if that works, why doesnt the binary that is already there and is the original one work_
<rawtaz>
?
mizu_no_oto has quit [Quit: Computer has gone to sleep.]
<rawtaz>
err, wait
<rawtaz>
well. i cant spend more time on this tonight, ive been at this shit for four hours now
<rawtaz>
time to make dinner at 1am instead.
<rawtaz>
starting to get annoyed by it not working as it should be
wpcarro has quit [Ping timeout: 244 seconds]
<rawtaz>
Dezgeg: thanks for trying to help for now! :)
<infinisil>
And to make it work: Every time the build does a read() call to a file which is not yet built, pause the process until the path is ready
haslersn has quit [Ping timeout: 252 seconds]
wpcarro has joined #nixos
sl3dge11 has joined #nixos
<sl3dge11>
Аⅼⅼaһ іs dⲟing
sl3dge11 has quit [Killed (Sigyn (Spam is off topic on freenode.))]
<catern>
infinisil: why would you want to do that?
<infinisil>
catern: Umm... Yet to be thought of lol
<infinisil>
Maybe it can increase parallelism or so
<catern>
sounds kind of like some build systems that magically intercept system calls to detect dependncies
<infinisil>
Oh!
<catern>
and rebuild deps in response
<infinisil>
The whole nix-store could be implemented like this
<infinisil>
Yeah that too
<catern>
hah, neat idea
<infinisil>
This is then a lazy build system
<catern>
well, essentially you get to a FUSE filesystem
<catern>
which builds on demand
<infinisil>
Yeah! Now that sounds much better
mkoenig has quit [Quit: Lost terminal]
<catern>
it needs to have the derivations though
<catern>
also impure builds should probably be done strictly
mkoenig has joined #nixos
<catern>
in case they fail
<catern>
actually, what happens in general on build failure?
<catern>
EOOPS?
<infinisil>
No idea what that is, but some signal to terminate the calling build
chessai has quit [Ping timeout: 252 seconds]
<catern>
it's an error code I just made up, which signals that the implementer of the system call made a mistake :)
<infinisil>
Although, it might be a bit hard to make things strict actually
<infinisil>
E.g. when I want to have a program that doesn't need to build anything when it runs, I would have to go through all possible program paths to get to all possible derivation paths
<infinisil>
Oh, but you can still just use the buildInputs like today
<infinisil>
Never mind
<clever>
infinisil: i just remembered, there is a package that will search $PATH for a given string to auto-detect support for something
<clever>
and due to the hashes in store paths, it gets false positives
fragamus has joined #nixos
alex`` has joined #nixos
<infinisil>
clever: For example? What things does it search for
<clever>
infinisil: i think it searched for the name of a dep to decide if the cflags are already setup or not
<clever>
and if the dep's name existed in a certain env var, it didnt bother adding it
<Dezgeg>
I think all of the laziness would go away immediately when stdenv goes to look in every thing in buildInputs to see if $foo/include, $foo/lib, $foo/bin exists