<{^_^}>
[nixpkgs] @nixos-channel-bot pushed 6 commits to nixpkgs-19.09-darwin: https://git.io/JewpU
<werner291>
Kernel 5.2, seems anything lower doesn't like my GPU very much.
o1lo01ol1o has quit [Ping timeout: 268 seconds]
Acou_Bass has joined #nixos
boogiewoogie has joined #nixos
<clever>
werner291: everything says it should be working...
<NoctisLabs>
New issue, the application using Timidity isn't finding `soundfont-fluid`, even though it's installed. I should add that this application is running with the Steam Runtime, is it the case that applications running in this way don't have access to system resources like soundfonts directly?
emilsp has quit [Read error: Connection reset by peer]
<werner291>
I'll keep poking at it then. Might be an upgrading issue now that I think about it, it's been a while since I did that.
shibboleth has quit [Remote host closed the connection]
drakonis has joined #nixos
<{^_^}>
[nixpkgs] @peti pushed to haskell-updates « hackage-packages.nix: automatic Haskell package set update »: https://git.io/JewpK
<nahamu>
what's the accepted way to nudge to get a PR reviewed / merged?
<clever>
,pr nahamu
<{^_^}>
nahamu: 0. Sorry. 1. If you complain about PR being in limbo, always provide a number («I have an unmerged PR (#23924)…»), bot will print PR title and it might bring you an impulse-merge. 2. Please look up who commits similar work and ping these people (or request-review), keeping up with all the PRs is not really possible.
<lordcirth__>
oh neat. So any further improvement would seem to require improvements to the Hydra software itself?
<gchristensen>
right
<lordcirth__>
Hmm. It seems like testing improvements at that scale could be rather difficult.
<gchristensen>
the biggest improvement doesn't require scale to test
<gchristensen>
all the build results go back through the main machine for signing and then upload to s3, being able to offload the sign-and-upload to a second (or N) machine(s) would be good
<lordcirth__>
Ah, are there numbers on what % of load is caused by that?
<gchristensen>
probably, if you're familiar with prometheus all the data is public
<lordcirth__>
I am not yet, but I will look into it
<clever>
gchristensen: i found that hydra often gets bottlenecked with xz compression
<clever>
gchristensen: weirdly enough, its doing it so well, it doesnt make a single syscall, so its imposible to even know why its "hung" for 2mins with only strace, lol
<gchristensen>
...nice
h0m1 has joined #nixos
red[m] has joined #nixos
<clever>
gchristensen: its just compressing one std::string into a second pre-allocated std::string, so it needs zero fileio, and zero memory allocations
<gchristensen>
wow
<andi->
Yeah, the time I lost trying to figure out what it is doing and it boiled down to that...
<clever>
andi-: it also puts the drv name into the threads argv[0]
<andi->
clever: ah, nice
<clever>
so if you hit the right keys in top, you can see what drv each thread is compressing
<clever>
Hc i think
<andi->
I used `perf` in the end
<clever>
i had to attach gdb and grab a backtrace, to see what it was "hung" on
<lordcirth__>
Hmm, xz does take a while to compress
<lordcirth__>
clever, is this compression happening on the central server?
<andi->
yeah
<clever>
yep
<clever>
gchristensen: how many cores on hydra.nixos.org ?
silver has quit [Read error: Connection reset by peer]
<andi->
IIRC we need to change `hydra-queue-runner` to be able to run in multiple processes/servers/…
o1lo01ol1o has joined #nixos
<clever>
another factor that can slow some hydra's down (but not nixos's), is checking upstream caches for narinfo files
<gchristensen>
clever: ...several
<clever>
to see if something has been built before
<gchristensen>
12
<clever>
gchristensen: ive also only really noticed the xz bottlenecks when dealing with ghc derivations
<clever>
oh, it does log the cpu time spent...
<lordcirth__>
12 isn't a ton. But scale-out is definitely better.
<clever>
gchristensen: those might even be small enough that the L1 or L2 cache could hold the entire thing
<leo_>
clever: Can you give me your opinion on a little code?
<clever>
leo_: probably
<gchristensen>
lordcirth__: not sure at the moment. what're you thinking?
<leo_>
1 sec and I send to you.
<lordcirth__>
gchristensen, mostly curious, but also wondering if I could do anything about it.
<gchristensen>
got it, work in that direction would be good for sure
<lordcirth__>
clever, these journal entries for compression time, could that data be exported to Prometheus?
<clever>
lordcirth__: sort of tricky, if you just report total time spent compressing, you wind up just showing cpu usage for that task (time spent vs time elapsed)
dhess has quit [Ping timeout: 265 seconds]
felixfoertsch has quit [Ping timeout: 246 seconds]
<clever>
lordcirth__: what else could you do with that metric?
<leo_>
Instead of using home manager, for instance
<nh2>
clever++
<{^_^}>
clever's karma got increased to 239
<leo_>
I use systemd.tmpfiles.rules in conjunction with nixops deployment.keys
<aanderse>
as someone who is not a functional programmer i'll try to keep up :)
<aanderse>
please, continue
<leo_>
That's it. .-.
<aanderse>
ha ha ha ok
<leo_>
infinisil was the one to suggest it. I tried to abstract it but I dunno what I'm doing. xd
<leo_>
I'm the newb of the newbs.
<clever>
leo_: id need an hour to even figure ouy what your doing, lol
<aanderse>
ah ok
<leo_>
clever: Does it seem too bad? D:
<clever>
gchristensen: under /queue-runner-status, id like things like .jobsets.<foo>.seconds and .shareUsed to be in prometheus, using tags for <foo>
<leo_>
andersee: I also set some backup acls in systemd.tmpfiles.rules
<clever>
gchristensen: everything under .machineTypes is also of use, with tags again
<CMCDragonkai>
I've noticed that even if I use `boot.zfs.forceImportAll` and `boot.zfs.forceImportRoot`, that if I did not properly `zpool export rpool`, zfs isn't force importing the root. I'm using a separate OS to then boot up and import rpool and then export rpool, and then when Ir eboot, the first system is able to import rpool again.
<leo_>
They're abstracted as backupAclsHead,backupAclsFoot
<CMCDragonkai>
Ok here's the situation. I used QEMU to create 2 raw disks. Then I booted into QEMU using NixOS installation netboot to set them up. Which I did successfully. Then I shutdown (after importing the rpool). Then I booted into QEMU using a second NixOS image that I had built from scratch. And this system fails to import rpool. However if I go back to the first system and export the rpool, then the second system succeeds
<infinisil>
leo_: Lol yeah, checked it out when you posted it earlier
<CMCDragonkai>
The second system has all the force import enabled settings.
<CMCDragonkai>
So its as if those settings don't actually force import!?
<infinisil>
CMCDragonkai: Oh hm don't vm images not use any fileSystem.* values?
<infinisil>
I think that's what the forceImport thing relates to
<clever>
gchristensen: ah, why isnt that just in hydra itself? it already has a statsd, oh, thats not prometheus
<CMCDragonkai>
my vm image is custom built nixos that does setup fileSystem
<scoates>
potentially dumb question: is there a way for me to get nix to completely forget about everything it previously did for a particular directory's shell.nix ?
<CMCDragonkai>
And it everything works if I make sure to export the rpool prior to booting into my second system
<CMCDragonkai>
my second system boots up, zfs is imported, directories are mounted...
<infinisil>
CMCDragonkai: And what's the error you get if you don't do that?
<CMCDragonkai>
the stage-2 hangs
<CMCDragonkai>
it hangs on trying to import rpool
<CMCDragonkai>
i think its a systemd task
<CMCDragonkai>
its like [..*...]
<infinisil>
Can you check the logs from that task?
gagbo has quit [Ping timeout: 240 seconds]
<CMCDragonkai>
and moving that star back and forth like nightrider
<leo_>
infinisil: Is it bad?
<CMCDragonkai>
possible to do that, but i dont know, cause when the system hangs, eventually it fails, and then goes to some systemd recovery
<infinisil>
leo_: Bad is subjective. If this works great for you then it's not bad :). (but the code probably wouldn't be accepted in nixpkgs :P)
gagbo has joined #nixos
<infinisil>
CMCDragonkai: You can retroactively check out the logs
felixfoertsch has joined #nixos
<leo_>
T-T
<leo_>
It works, dunno if it works great. xd
<leo_>
It wouldn't be accepted into nixpkgs? Why?
<infinisil>
leo_: What exactly does that do?
<o1lo01ol1o>
I've got a nixos install on a desktop to which I've just added a new SSD. What's the fastest way to move the installation to the new ssd?
<leo_>
Copy some keys from /keys/by-hash/{key-id-hash} (where id is either name or passName) to directories specified by rCTarget (using systemd.tmpfiles.rules)
<leo_>
Or sets some backup acls (so you don't need to do it as root)
bvdw has quit [Read error: Connection reset by peer]
<infinisil>
o1lo01ol1o: Something like `cp -r /nix /ssd/nix`?
<leo_>
Or creates some directories (by systemd.tmpfiles.rules)
<leo_>
In the specified order. .-.
bvdw has joined #nixos
jboy has quit [Remote host closed the connection]
<o1lo01ol1o>
infinisil: will that preserve the needed links for users and store?
<infinisil>
leo_: Hm, one problem with this I can see is that it's essentially a way to set up stateful symlinks/copies from anywhere to anywhere, which could be a more general option
<infinisil>
leo_: Which could be used for dotfiles and such. Which would be nice for NixOS in general, but I'm not sure how well this works with atomicity and such
<leo_>
Yeah!
<leo_>
Now I got what you're saying. xd
<leo_>
My "solution" is very adhoc.
jboy has joined #nixos
<infinisil>
o1lo01ol1o: Yeah cp doesn't change anything. Just make sure to mount /ssd/nix as /nix in the end
<infinisil>
o1lo01ol1o: I'd usually do `cp ...` then change the /nix entry in fileSystems.* to point to the new storage, then reboot
<infinisil>
Though this might be a bit problematic if /nix is part of the filesystem mounted at /
<o1lo01ol1o>
infinisil: it is
<o1lo01ol1o>
where is fileSystems.*?
<infinisil>
o1lo01ol1o: Simple solution: Boot into a live stick after that to empty /nix from / (or move it to /nixold)
<infinisil>
o1lo01ol1o: In hardware-configuration.nix
<o1lo01ol1o>
ah, ok
<infinisil>
o1lo01ol1o: Oh wait, do you want to move the whole filessytem or just /nix?
<infinisil>
Well it works the same for the whole filessytem really.
<o1lo01ol1o>
infinisil: it's even simpler: currently i'm installed to one ssd and I'm out of space. I'd like to spread the install across a couple of drives, either putting mount points in different places or by some other magic
<infinisil>
o1lo01ol1o: You're not using zfs?
<o1lo01ol1o>
maybe?
<o1lo01ol1o>
it's 19.09
<infinisil>
You'd know if you were using zfs, so you're probably not using it hehe
<infinisil>
zfs is a filesystem
<o1lo01ol1o>
I know, but I didn't enable it manually
<infinisil>
and with it it's really easy to add more storage without having to mess with partitioning and such
<o1lo01ol1o>
ok, so maybe I'll just reinstall with zfs partitions then
<infinisil>
So you know how to use zfs already?
pie_ has quit [Ping timeout: 240 seconds]
<gchristensen>
(we should probably add docs to the manual)
<infinisil>
Not saying you need to use it, you could also use lvm volumes to do the splitting into multiple disks
<o1lo01ol1o>
I've never used it, no. (And if there aren't docs then maybe I'll hold off . . . )
<slack1256>
How do I change the value of /proc/sys/kernel/sysrq via configuration.nix?
<slack1256>
Some time ago it was changed from 1 to 16, but I like to trigger the OOM killer manually instead of waiting for it to fill the swap.
<o1lo01ol1o>
infinisil: that looks like it would drain several hours of time to get right. I think I might just copy my configuration.nix and hardware.nix somewhere and reinstall the whole thing.
<infinisil>
o1lo01ol1o: Do you want to use space on both disks uniformly or not?
<infinisil>
Without having to designate certain filesystems to certain disks that is
<o1lo01ol1o>
infinisil: that would be nice, yes
<infinisil>
Because if you do, you need something like lvm or zfs to handle that
<o1lo01ol1o>
how's the lvm support? I've used it before but not in nixos
felixfoertsch23 has joined #nixos
<infinisil>
I think it's pretty good, nixos-generate-config can handle it afaik
<{^_^}>
[nixpkgs] @nixos-channel-bot pushed 6 commits to nixos-19.09-small: https://git.io/JewpU
felixfoertsch has quit [Ping timeout: 252 seconds]
<justanotheruser>
this worked in 19.03 but fails with 19.09
drakonis has quit [Quit: WeeChat 2.6]
slack1256 has quit [Ping timeout: 265 seconds]
<lordcirth__>
justanotheruser, can you run "file /nix/store/j2mm5fyrghl1xv6jmh765z4x6z3xf2js-python-remove-bin-bytecode-hook/nix-support/setup-hook" and see what it says?
<justanotheruser>
/nix/store/j2mm5fyrghl1xv6jmh765z4x6z3xf2js-python-remove-bin-bytecode-hook/nix-support/setup-hook: data
<lordcirth__>
hmm, I'd expect ELF
stzsch has quit [Ping timeout: 276 seconds]
<lordcirth__>
If it's supposed to be executed
<justanotheruser>
is there a way I can just remove this operation? I tried ` removeBinBytecode = false;` to no avail
<{^_^}>
[nixpkgs] @bjornfor pushed commit from @aanderse to release-19.09 « moodle: 3.7.2 -> 3.7.3 »: https://git.io/Jervo
robogoat has joined #nixos
LysergicDreams has quit [Ping timeout: 240 seconds]
NoctisLa1 has joined #nixos
o1lo01ol1o has joined #nixos
LysergicDreams has joined #nixos
bahamas has joined #nixos
bahamas has quit [Changing host]
bahamas has joined #nixos
NoctisLabs has quit [Ping timeout: 246 seconds]
iqubic` has joined #nixos
iqubic` has quit [Remote host closed the connection]
iqubic has quit [Ping timeout: 250 seconds]
o1lo01ol1o has quit [Ping timeout: 240 seconds]
mexisme has joined #nixos
<{^_^}>
[nixpkgs] @bjornfor pushed commit from @ivan to master « chromium: 78.0.3904.87 -> 78.0.3904.97 »: https://git.io/Jervy
pie_ has joined #nixos
chreekat_ has joined #nixos
ThatDocsLady has joined #nixos
red[evilred] has quit [Quit: Idle timeout reached: 10800s]
<chreekat_>
When running the nix installer via ansible (yeah ...), I'm frequently getting errors along the line of "Fatal error: glibc detected an invalid stdio handle". I wonder what *that's* about
<alexarice[m]>
anyone getting an error that looks like `error: assertion failed at /nix/store/lxdjr1ahi11dhgn3j9i9hkip1b297qwq-nixos-19.09.1208.ef8c34c4721/nixos/pkgs/development/compilers/gcc/common/configure-flags.nix:24:1` when upgrading
bahamas has quit [Quit: leaving]
<{^_^}>
[nixpkgs] @magenbluten opened pull request #73328 → linux config: add support for xdp sockets and ebpf jit → https://git.io/JerUB
<oscarvarto>
Any guidance/suggestions will be appreciated
iyzsong has joined #nixos
<oscarvarto>
(I also tried using vscode + hie configuration as home-manager's options suggest. I actually asked for help to set up this, but still stuck). Thanks in advance.
o1lo01ol1o has joined #nixos
cyphase has joined #nixos
<oscarvarto>
(by the way, it's pretty early here, I will check (possible) answers asap)
<__monty__>
oscarvarto: What do you get when you run that command in a terminal?
<symphorien>
not if you have installed packages with nix-env
rardiol has quit [Ping timeout: 252 seconds]
shyim has joined #nixos
jeyj0 has joined #nixos
fendor has quit [Read error: Connection reset by peer]
<jeyj0>
Hello, I'm very new to NixOS and currently setting up my dev environment in a vm, to have an easy transition once I installed it install it on my Desktop.
<jeyj0>
However, I've run into a problem trying to get nvm (node version manager) to work. Has anybody gotten that to work previously and could help?
<exfalso>
Hi, I have a weird situation. We have a multi-user nixos setup, and I just added a user. We already had a couple of users using nix fine, however if this new user issues a nix-env -i, it seems *nix-daemon SIGKILLs their shell*
<exfalso>
Has anyone seen this before?
<exfalso>
nix-daemon logs say "nix-daemon[7350]: unexpected Nix daemon error: writing to file: Broken pipe"
<exfalso>
but i'm not sure whether this is the reason for or is a consequence of the kill
<lux2>
symphorien my problem is I cant install something
<lux2>
nix-shell -p nodejs-12_x throws an error
<lux2>
while nix-shell -p nodejs-10_x works
<etu>
jeyj0: nvm is kinda a hack to handle multiple versions of the same software and being able to switch between them
<infinisil>
exfalso: No idea, but do the nix daemon version and nix-env version match?
bahamas has joined #nixos
<etu>
jeyj0: You probably want to look into nix-shell and looking into creating shell.nix files to load (possibly with direnv) to get access to different software for different directories
<etu>
jeyj0: The default nodejs seems to be 10.x, we also have nodejs_latest, nodejs-10_x, nodejs-12_x, nodejs-13_x (at least on unstable) if you want to be specific.
<dminuoso>
In general, is it safe/sane to do something like `{ config, ... }: let f = config.some.stuff; g = import ./foo.nix f; in { ... }`?
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @r-ryantm to master « atlassian-cli: 8.5.0 -> 9.0.0 »: https://git.io/JerkC
<jeyj0>
etu: I actually need node v7, as well as 10, 12 and 13 in different projects (various incompatibilities that I'm hoping to fix asap, but still, it'll take a while...). So 10, 12 and 13 aren't any issues, but 7 will, I guess. I'm currently using nvm for that, which works wonderfully. So it'd really be a shame if I didn't get it to work on NixOS
<jeyj0>
etu: This (https://github.com/nvm-sh/nvm/issues/2065) is basically the same issue as I have. Maybe you can take a look and see if there's an obvious direction you can push me to?
<etu>
jeyj0: With the help of pinning you can pull packages from other versions of nixpkgs. I'm on unstable but use packages from 19.03 on a daily basis due to legacy reasons.
<{^_^}>
nvm-sh/nvm#2065 (by ORESoftware, 21 weeks ago, open): nvm on nixos and bash hashing
romildo has joined #nixos
<romildo>
How to increase the disk image size used by the VM built with 'nixos-rebuild build-vm'?
orivej has joined #nixos
<__monty__>
jeyj0: Have you tried setting a different "npm config prefix"? If nvm is trying to write to that path it won't work because the store is read-only.
xkapastel has joined #nixos
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<jeyj0>
__monty__: yeah, I've tried to set it to the correct path for the specific nvm version, which apparently fixed the setup for people trying to install nvm via homebrew. I basically got it to a point where nvm didn't complain, but calling `node` still used the nixpkgs node. Removing that one causes nvm to complain that /usr/bin/env node doesn't
<etu>
jeyj0: I pulled that from the channel for the 17.03 release :)
Chiliparrot has quit [Client Quit]
<__monty__>
jeyj0: Then prepend that correct path to your PATH.
<exfalso>
infinisil: it's very strange. The kill happens immediately after nix-env reading /nix/store/zg66y04g2bvmw41cgrywysr86s40g5cc-nix-2.2/share/nix/corepkgs/buildenv.nix
<etu>
jeyj0: If you put this in a file named "shell.nix" http://dpaste.com/1QMM8BQ and run "nix-shell" in that same directory you should have nodejs version 7 available
<__monty__>
I agree with etu that doing this with proper shell.nix files will be a better approach. I admit though that not everyone wants to spend time setting everything up right up front, so I'm just trying to show that nixOS can still handle dirty dirty software : )
noudle has joined #nixos
<lux2>
how do I install nodejs-12_x?
<exfalso>
it seems it happens during a derivation build of user-environment. The .lock file is still there in /nix/store
<exfalso>
infinisil: any idea on how to debug this?
<lux2>
nix-shell -p nodejs-12_x
<lux2>
error: undefined variable 'nodejs-12_x' at (string):1:94
<lux2>
(use '--show-trace' to show detailed location information)
<etu>
lux2: What channel are you on?
<{^_^}>
[nixpkgs] @nixos-channel-bot pushed 13 commits to nixos-19.09: https://git.io/JerIv
<etu>
Izorkin: Does anything in nixpkgs depend on 2.x?
<l33[m]>
Im changing my hdd for a sdd. Hoping to clone my nixos partition.... i presume i have to modify hardware-configuration.nix .... whats the chance of it working on a scale of 1 to 2.
<Izorkin>
etu: no
o1lo01ol1o has joined #nixos
<etu>
Izorkin: Is 2.x maintained and get point releases?
<Izorkin>
etu: latest update - 2018-09-07
cosimone has joined #nixos
<etu>
l33[m]: It depends on how your hardware-configuration looks, if it just point out the devices /dev/sdX, you shouldn't need to change anything. But if you use /dev/disk/by-uuid/ or so and don't clone the partition table (instead you create new partitions) you will need to change the uuids.
<etu>
Izorkin: nixpkgs is not a repository for unmaintained software, so if nothing depends on it and they don't support that major version. We shouldn't keep it around.
o1lo01ol1o has quit [Ping timeout: 240 seconds]
oscarvarto has joined #nixos
<Izorkin>
etu: then delete options phpOptions mysqliSupport ?
morgaine[evilred has joined #nixos
joelwallis has joined #nixos
<Izorkin>
"--with-pdo-mysql=${if mysqlndSupport then "mysqlnd" else libmysqlclient}" and "--with-mysqli=${if mysqlndSupport then "mysqlnd" else "${libmysqlclient}/bin/mysql_config"}"
<etu>
I believe that ubuntu is trying to get rid of it before 20.04 to not have to support it until 2025 (+ 5 years for ESR)
<lordcirth__>
adisbladis, so that PR didn't happen. Any ideas where one could continue, preferably in manageable chunks?
<lordcirth__>
I would expect so, though it will be a ton of work
<etu>
Izorkin: I'll check it this weekend.
<Izorkin>
ok
<adisbladis>
lordcirth__: I don't know.
<adisbladis>
Looks like RHEL 8 is supported until at least 2023 with extended support until 2025, so presumably they'll have python2 patches until then.
<adisbladis>
lordcirth__: I believe FRidh is working on something, but I haven't checked the progress
kenran_ has quit [Ping timeout: 276 seconds]
<etu>
Ubuntu 18.04 should be until 2023, and they have ESR for paying customers adding 5 years. So they should backport security patches until 2028 which is insane :p
<FRidh>
adisbladis: its a bit funny with that PEP actually. Originally, it said it could/would point to python3 eventually. Then suse did that, and then they changed views saying python should always point to python2
<FRidh>
in nixpkgs we actually do something similar to suse. python will point to 2 and 3, with 2 taking precedence if both are present
<adisbladis>
FRidh: Btw, this is not a short term change, but what do you think about completely removing pythonXPackages (and state that pythonX.pkgs is the blessed way)?
<FRidh>
adisbladis: I don't know. I would *really* prefer a common approach here in how we deal with package sets
<FRidh>
I introduced pythonX.pkgs some years ago, but it may actually make overriding messier
<FRidh>
a significant issue is also, how do we deal with functions in a package set
<adisbladis>
FRidh: Hmm, yeah.
<FRidh>
ideally they're not in the set, but where then, especially if they're interpreter-dependent?
Makaveli7 has quit [Quit: WeeChat 2.6]
<adisbladis>
I've wanted to use only lib functionality from pythonPackages without any actual packages some times.
<FRidh>
now some functions are part of the passthru of the interpreter, and others part of the package set
<{^_^}>
[nixpkgs] @nixos-channel-bot pushed 2 commits to nixos-19.09: https://git.io/JerLO
<joelwallis>
Hi. I've just installed Nix on my macOS and it shows my /nix folder having 226MB of size. Is it correct (a brand-new installation containing so much data on it)?
<adisbladis>
joelwallis: Doesn't sound abnormal.
<adisbladis>
joelwallis: Since /nix esentially contains a complete userland with nix + all it's dependencies.
waleee-cl has joined #nixos
<joelwallis>
So these are dependencies of Nix itself?
dsg has joined #nixos
boogiewoogie has joined #nixos
<etu>
joelwallis: Since it brings it's own libs and everything it sounds reasonable. Nix uses a lot of bash and perl which takes quite some space.
<joelwallis>
Interesting. Thanks for the explanation.
<joelwallis>
I'm new to Nix/NixOS world, so everything is still new to me. I'm currently reading the Nix Manual and exploring around
endformationage has joined #nixos
<etu>
joelwallis: And you should also have an untared copy of nixpkgs in there. Which also takes quite some space.
<adisbladis>
FRidh: I'm just really scared about passing the interpreter and package sets separately. It's so easy to get it wrong
<lordcirth__>
nixpkgs master, excluding .git, seems to be 170MB
<dsg>
I'm going through the CVE roundup issues on GitHub, to try to update the status and find out which ones can be closed (to make the list less scary). What are the currently supported NixOS channels? Is it enough for the fix to be in nixos-19.09, or should I be looking at whether it's been cherry-picked for 19.03 as well?
<FRidh>
adisbladis: yes whenever I see it in a PR I comment about it that they should not do that
<FRidh>
so indeed having only one would be preferred
joelwallis has quit [Remote host closed the connection]
<lordcirth__>
Why does nix why-depends have to download all the packages involved? Shouldn't the nixpkgs metadata be enough?
<FRidh>
how about a pythonX.lib and pythonX.pkgs. I suppose this could be done also for other package sets. Are there any cons to it?
<adisbladis>
lordcirth__: It's the runtime closure of the package, which is not known by nix metadata.
<adisbladis>
FRidh: How would that affect build hooks?
sigmundv has quit [Ping timeout: 245 seconds]
NoctisLa1 has quit [Read error: Connection reset by peer]
ddellacosta has joined #nixos
<FRidh>
adisbladis: do you mean, would they be part of libs or hooks?
<FRidh>
*libs or pkgs?
NoctisLa1 has joined #nixos
<FRidh>
**lib
<adisbladis>
FRidh: I'm not clear on all the implications of a change like that :)
<adisbladis>
But on the surface it sounds good
MmeQuignon has joined #nixos
rauno has quit [Ping timeout: 240 seconds]
boogiewoogie has quit [Remote host closed the connection]
<MmeQuignon>
Hi, I upgraded nixos from 19.03 to 19.09, should I update system.stateVersion ?
<gchristensen>
no
<FRidh>
buildPythonPackage and such would move to pythonX.lib, but the hooks would be derivations thus part of .pkgs. Unless we would want to have something like .buildSupport as well?
mariel__ has quit [Remote host closed the connection]
<joelwallis>
I would like to describe my own personal CLI dependencies through a text file. How could it be possible with Nix?
boogiewoogie has joined #nixos
<joelwallis>
I would like, for instance, to describe git, zsh and tmux as dependencies of my user. So every time I reinstall my OS I could just rebuild my own environment through this description file.
<symphorien>
,home-manager
<joelwallis>
I'm currently reading about Nix expressions, but they seem to be tailored for package publishers
<alexarice[m]>
joelwallis: home manager has a manual that might help.
<alexarice[m]>
nixos manual might also help
Ariakenom has joined #nixos
<joelwallis>
I'm reading about Home Manager now
<boogiewoogie>
if it's just installing packages you want to specify for your enviroment, not more general options, I think specifying them in ~/.nixpkgs/config.nix should do the trick, or am I missing something?
<{^_^}>
[nixpkgs] @ebkalderon opened pull request #73344 → makeFontsCache: Fix cross-compilation, use nativeBuildInputs → https://git.io/Jertn
<alexarice[m]>
boogiewoogie: does that install them per user?
<joelwallis>
So, there's no way to do it with the "vanilla Nix"?
<Ariakenom>
my store is 40GB and I did GC with --delete-old. can I make it smaller?
<boogiewoogie>
afaik yes, I'm not using that though since I'm on a single-user nixos system
<alexarice[m]>
joelwallis: There is
<alexarice[m]>
I'll try to find a link
<joelwallis>
alexarice[m] thanks in advance! :D
<boogiewoogie>
isn't the config.nix for vanilla nix?
<exfalso>
is there a way to enforce deployment order with nixops?
<exfalso>
i'd like a deployment of a specific machine to run last
<{^_^}>
[nixpkgs] @Mic92 opened pull request #73345 → clang-tools: teach about nix's include path → https://git.io/JertB
bahamas has quit [Quit: leaving]
<talyz>
exfalso: I'm not aware of any deployment ordering options, but you can use --include and --exclude explicitly to only deploy to certain hosts
o1lo01ol1o has joined #nixos
bvdw has quit [Read error: Connection reset by peer]
bvdw has joined #nixos
glittershark has joined #nixos
Fare has quit [Remote host closed the connection]
<romildo>
I am needing some help with managing a branch in my clone of my fork of nixpkgs. I have submitted https://github.com/NixOS/nixpkgs/pull/63813 which received contributions from others. In an attempt to update the branch locally I may left it in an inconsistent state. What should I do to fix it?
<joelwallis>
romildo is it a Nix question or a Git question?
<romildo>
joelwallis, it is a Git question.
Fare has joined #nixos
<joelwallis>
So, you want to get your branch/pull request containing only your last two commits?
<joelwallis>
I see you've got some commits from a user worldofpeace in your PR
<Ariakenom>
if users.users.USER.password = ""; is that the same as no password?
<Ariakenom>
if not, can I set no password?
<waleee-cl>
I think it's just that the initial password will be empty? You can set it later (preferably right after install then)
<romildo>
joelwallis, I want my local branch containing all the commits in the PR, including mine and worldofpeace's.
<Ariakenom>
Im making a bootable usb and would prefer no password on it
<__monty__>
,declarative joelwallis
<{^_^}>
joelwallis: There are multiple ways of managing declarative profiles. 1) Attrset, compatible with imperative use of nix-env https://git.io/fAQHW ; 2) buildEnv, providing more control over the paths that are linked into the profile https://git.io/fp0aU ; 3) home-manager, providing nixos-like config for your ~ https://github.com/rycee/home-manager
<alexarice[m]>
Ariakenom: You can probably set it to autologin
<alexarice[m]>
__monty__: think the bot is dead
<Ariakenom>
autologin still has password for the login screen and such?
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @r-ryantm to master « syslogng: 3.23.1 -> 3.24.1 »: https://git.io/JerqO
<exarkun>
Can I get rid of the nix dependency on aws-cpp-sdk? If I set `nix.package` to something like `nix.overrideAttrs { withAWS = false; }` or some such? Will this actually work?
<__monty__>
alexarice[m]: Maybe the bot got banned on the bridge? Good to know though, but joelwallis doesn't have an [m] so they should be seeing it : )
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
<mananamenos>
hi, `nix-shell --pure -p ghc cabal-install cabal2nix --run "cabal init"` before i used to get interactive initing of project asking me stuff, now it just finishes with .cabal file, no src folder..what did i do?
<{^_^}>
[nixpkgs] @zupo opened pull request #73349 → Add checkPhase for pipenv → https://git.io/Jerqr
<{^_^}>
[nixpkgs] @FRidh pushed commit from @dtzWill to master « microcodeIntel: 20190918 -> 20191112 »: https://git.io/JerqF
zebra has quit [Ping timeout: 265 seconds]
<{^_^}>
[nixpkgs] @FRidh merged pull request #73100 → documentation: show combining python35.withPackages and mkShell → https://git.io/JeV2D
<{^_^}>
[nixpkgs] @FRidh pushed commit from @raboof to master « documentation: show combining python35.withPackages and mkShell »: https://git.io/JerqN
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @risicle to release-19.03 « opencv4: add patches for CVE-2019-14491, CVE-2019-14492, CVE-2019-14493 & CVE-2019-15939 »: https://git.io/JermO
<scoates>
I noticed that my (new) local build of nix-pills has the chapter name in the <title> but the published version does not: https://nixos.org/nixos/nix-pills/garbage-collector.html ; is there just some lag here? (was going to try to implement this, but it seems it's already "done")
<asheshambasta>
Hi everyone, I'm a new nixos user. I'd like some help on learning how to install emacs 27 on my system via a nix derivation. Is there some place I can read up about this? I've tried googling this a little bit but I've only found some resources spread here and there.
<joelwallis>
ashes
<Twey>
Why does `nix eval` require brackets around strings? `nix eval '"hello"'` says ‘don't know what to do with argument '"hello"'’ but `nix eval '("hello")'` is fine
<joelwallis>
asheshambasta you can install it by running nix-env -i emacs
<dminuoso>
Mmm, it seems acme.security is still using ACMEv1.. and letsencrypt has shut down support for that 2 weeks ago. How is this not already reported?
<dminuoso>
Twey: Because otherwise the shell will interpret the double quotes
<dminuoso>
Twey: You could also do `nix eval \"hello\"`
<Twey>
dminuoso: That's why I put them in single quotes
<dminuoso>
Oh
<Twey>
dminuoso: '"hello"' is equivalent to \"hello\" in bash
<asheshambasta>
joelwallis: doesn't that depend on the nix-channel?
<joelwallis>
Twey asheshambasta oh, yeah. I didn't notice the version, sorry
<exarkun>
dminuoso: I tried security.acme and had to give it up and roll my own with certbot
<asheshambasta>
basically, I'd like to pin some 27.x version and install that.
<dminuoso>
exarkun: For the same reason?
<exarkun>
dminuoso: yes
cosimone has quit [Quit: Terminated!]
<dminuoso>
exarkun: Im just really surprised that its been 2 weeks and there's not even an issue.
<dminuoso>
Do people not use letsencrypt?
<dminuoso>
exarkun: Do you have a working nixos module already?
<justanotheruser>
Hi, I'm having an issue with building a derivation - "cannot execute binary file" - full traceback: https://paste.debian.net/plain/1116001 - derivation: https://paste.debian.net/plain/1116002 This worked in 19.03 but not 19.03. `file /nix/store/j2mm5fyrghl1xv6jmh765z4x6z3xf2js-python-remove-bin-bytecode-hook/nix-support/setup-hook` -> `data`
<exarkun>
dminuoso: I think what I found is that in principle security.acme is fixable but it requires upgrading simp_le which requires upgrading some other library which had some non-trivial issue and that's where I threw my hands up
<eyJhb>
I have been using TLP for some time, with charging thresholds etc. but after I have disabled it, it still doens't want to charge above 70%. Any way to reset it?
<Twey>
asheshambasta: (particularly the reference list at the top)
<Twey>
asheshambasta: In your case, you probably just want to pull the emacs derivation from nixpkgs and slightly modify that :)
<asheshambasta>
Twey: thanks, am I right in assuming that the modification will only entail modifying the: URL and SHA of the "package" being pulled to build locally?
<Twey>
asheshambasta: Maybe. Depending on what's changed, 27 might require the build to be done differently to 26, and then you'll have to fiddle with the build steps
joelwallis1[m] has joined #nixos
<joelwallis1[m]>
This Riot is by far the most well tailored option for IRC chatting I've ever seen
<asheshambasta>
Twey: okay, I see, thanks a lot for the pointers.
<tilpner>
joelwallis1[m]: It's a decent Matrix client, but it's a terrible IRC client
<tilpner>
,matrixbridge
<{^_^}>
The IRC<->Matrix bridge sends multi-line messages as links: `* Someone sent a long message: < https://matrix.org/_matrix/media/v1/download/matrix.org/... >`. This means that highlights don't work, people have to open your link, and it gets less visibility than multiple single-line messages
<catern>
is there a recent worked-out example/guide to configuring vim with Nix? I don't use vim myself so I can't judge which is best (I just pointed the user at the Nixpkgs manual section)
<aranea>
What measures does NixOS take to ascertain the integrity of downloaded packages? AFAICS, binary packages are (or at least can be) signed, but the nixpkgs.git commits aren't signed, and neither are the release ISOs, so when you're building from source or initially installing NixOS, you're only protected by TLS?
<tilpner>
aranea: Building from source: There are hashes of the sources in the nix expressions you build from
domogled has joined #nixos
<gchristensen>
aranea: how would you like to first establish the root of trust with the Nix ecosystem?
<tilpner>
aranea: Fetching from cache: There is a signature mechanism for substitutions
<aranea>
tilpner: Sure, but the expressions themselves could've been tampered with.
<tilpner>
aranea: Fetching nixpkgs: You've discovered the weak link
<aranea>
By github, or someone hacking github, for example.
<gchristensen>
or
<gchristensen>
somebody could commit a malicious expression
<aranea>
gchristensen: Sure, that's always tough.
<gchristensen>
imo I consider it more likely
<tilpner>
Much simpler than being GH: steal credentials from any of the many contributors
<aranea>
Of course, but that's the /inherent/ risk of every distro.
<tilpner>
And then push while spoofing authorship
ddellacosta has quit [Ping timeout: 268 seconds]
<aranea>
You can't avoid having to trust your own guys, but it's nice to at least not to have to trust any unrelated third parties.
<tilpner>
On my todo list: Send notifications when contributors with GPG keys registered push without signing (or a new key)
<gchristensen>
that is a cool idea
<exarkun>
So you imagine someone attacking your first download of nixpkgs and tampering with the expressions to point at something malicious?
<aranea>
Yeah, that sounds like a useful step.
<tilpner>
exarkun: You're re-downloading nixpkgs all the time, so there's a lot of first downloads to tamper with
<exarkun>
tilpner: just trying to understand the concern
<aranea>
exarkun: Yes, for example.
<tilpner>
(Where you is a typical nix user, who we have no statistics about)
ng0 has quit [Remote host closed the connection]
ng0 has joined #nixos
__red__ is now known as redredred
<tilpner>
aranea: You can't avoid having to trust the team of committers, but you can avoid having to trust any single 1 (or n) of them
<tilpner>
(Though that quickly gets expensive)
romildo has quit [Quit: Leaving]
<aranea>
Yeah, you can enforce reviews of every commit or sth like that.
<gchristensen>
as usual, any discussion around this is much more productive if you have a threat model described
<gchristensen>
then I suspect you have come prepared!
ambro718 has joined #nixos
<aranea>
Anyway, sth else entirely: How well suited is nixpkgs to a usecase where I practically use it as an entire distro, but provide the kernel, init system and service manager myself?
<aranea>
(I guess I shouldn't call it NixOS anymore at that point...)
morgaine[evilred has quit [Quit: Idle timeout reached: 10800s]
orivej has joined #nixos
boogiewoogie has quit [Quit: Leaving]
<Twey>
aranea: There are projects like notOS and SLNOS that do this kind of thing
<dminuoso>
exarkun: Mind my asking, how do you deal with renewal after 90 days with that?
<Twey>
aranea: It's doable I guess? A lot of the nice NixOS modules assume certain things, especially about systemd, but if you're willing to forego/rewrite those, I don't see any particular reason it should be too difficult
<exarkun>
dminuoso: well, I haven't been using it for 90 days yet, of course! :) But sometime before then I'll write a systemd timer service that re-starts that service, I guess.
<dminuoso>
exarkun: You mean both that service and the service needing that certificate, right?
<exarkun>
dminuoso: yea
<exarkun>
dminuoso: I think the mechanism for restarting the service consuming the certificate might be indirect rather than something the timer service does explicitly. maybe.
<aranea>
Thanks for the pointers! Do you happen to have a link for SLNOS? I found the notos repo, but searching for "nixos slnos" only turns out various mentions, nothing authoritative.
<exarkun>
dminuoso: like, systemd has file units that can trigger on a file changing. maybe having a file service for the certificate makes sense?
<aranea>
Is it intended that the nixos-generate-config tool mentioned in the install guide isn't present on the minimal ISO?
<infinisil>
aranea: it totally should be present there
<betaboon>
when booting from the minimal-installer-iso i am still required to have internet access for install ?
<aranea>
betaboon: Yes.
tilpner has quit [Quit: tilpner]
<slabity>
betaboon: Yes. Unless you can download an offline version of nixpkgs
<aranea>
infinisil: Well, it's definitively not anywhere in the path on the nixos 19.9 x86_64 minimal iso I just downloaded. :)
<infinisil>
aranea: Does /run/urrent-system/sw/bin contain it?
<betaboon>
merghhh. maybe someone has a suggestion: I'm just working on the installer tests. trying to boot into efi-mode right from the beginning (opposed to booting into bios-mode initially). i got that somewhat working right now. right now I'm booting from efi to installer iso. but as that needs network access I'm stuck. any suggestions ? XD
<aranea>
infinisil: No.
domogled has quit [Read error: Connection reset by peer]
<infinisil>
aranea: Does it contain nixos-install?
<aranea>
Yup.
<infinisil>
Weird
xkapastel has joined #nixos
<gchristensen>
strange, I just fetched one yesterday and it did have it
<aranea>
Yes. The binary even appears to be in the store, it's just not linked into the PATH.
<gchristensen>
what is your PATH?
<aranea>
Uhh, it's pretty long, and I don't have working copy&paste with that VM, but it's whatever was the default.
mariel__ has quit [Ping timeout: 252 seconds]
<gchristensen>
aye
<aranea>
Including the /run/current-system/sw/bin that infinisil mentioned.
tilpner has joined #nixos
fendor_ is now known as fendor
<aranea>
Uh, huh. Sorry for the noise.
<gchristensen>
what's up
<gchristensen>
?
domogled has joined #nixos
<aranea>
It's there now. I guess the PATH /was/ wrong in the shell I was using initially; it's there now. Or I was incredibly blind for a minute, that's not completely out of question either.
<gchristensen>
hm :) okay
<gchristensen>
glad to hear it is there
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<gchristensen>
because you probably want packages not inculded in that store
<{^_^}>
[nixpkgs] @Ma27 pushed to structured-attrs « make-derivation: assert that configureFlags is a flat list »: https://git.io/JerCs
<samueldr>
betaboon: ugh, didn't think about that
<samueldr>
the way it works in the current test is that the "host's" store is mounted at /nix/store with the non-full-boot VM
<betaboon>
samueldr: yeah thats what i thought.
kuznero has joined #nixos
<kuznero>
Hi All!
<betaboon>
I'm close to giving up here :/
<samueldr>
betaboon: that would be understandable
<samueldr>
it's not an easy thing
psyanticy has joined #nixos
<betaboon>
but it feels like being so close. the iso is booting in efi-mode directly after the efi-bootloader XD
<betaboon>
(still all very hacky tho)
<samueldr>
I guess that if we figure out the right mount with virtfs fs for /nix/store it could work
<betaboon>
samueldr: so like having the host nix-store as a "fake" partition in the installation system ?
<samueldr>
I was thinking mostly like the current test already does
<samueldr>
plainly mounted as /nix/store
<kirelagin>
Is cross-compiling for Darwin a thing? I think I remember someone saying that it works, but I am getting “unsupported on this platform” pretty early during bootstrap :(.
<tilpner>
leo_: The example seems to set both at once?
<leo_>
It's either mkSetuid or mkSetcap
<betaboon>
samueldr: meh, this keeps getting more and more complex
<samueldr>
yeah
<tilpner>
Oh, and mkSetcap uses root only, and you don't want root?
<samueldr>
might need to rather look into not booting the actual iso, boot what is being booted, but in full EFI
<samueldr>
I'm not sure though that qemu can
<leo_>
Yes.
<tilpner>
(No, it merges with s, which would overwrite root)
<tilpner>
leo_: That should still work
<leo_>
For instance, I want cap_sys_tty_config+ep
<tilpner>
Please try it
<leo_>
As a capability
<leo_>
And I want `setgid = "bla";`
<tilpner>
Yes, please try that
<leo_>
setgid = true; group = "bla";*
<leo_>
It doesn't work. I tried. xd
<__monty__>
Am I imagining stuff? Can't find a commit that removed it so I guess it's never existed? What license should I give an unspecific unfree project?
<leo_>
And in the code you can see that you can only set one
<tilpner>
leo_: You switched into that config, and the permissions weren't right?
<leo_>
Yep.
<betaboon>
samueldr: i havent been able to boot "disk.img" through efi
<leo_>
It branched to mkSetcap
<tilpner>
leo_: On line 81 it allows you to overwrite root:root
fiddlewood has joined #nixos
<tilpner>
And the ping example suggests it's possible
<leo_>
Yeah, the group and the user.
<leo_>
But I also want setgid. .-.
asheshambasta has quit [Remote host closed the connection]
<tilpner>
Oh, now I see
<__monty__>
I guess I was looking for licenses.unfree ...
<{^_^}>
[nixops-aws] @tewfik-ghariani opened pull request #16 → Displaying EBS size in nixops-info → https://git.io/JerCX
o1lo01ol1o has joined #nixos
<tilpner>
I don't see a good way to get both without patching nixpkgs :/
<betaboon>
samueldr: but that might have just been due to bootorder stuff which i havent figured out
<ArdaXi>
I've just run into a weird issue where recently (a day ago or so) any 32-bit builds fail because meson does not recognise the wrapped gcc as a valid compiler, does anyone know where I should be looking?
fiddlewood has quit [Quit: WeeChat 2.6]
gentauro has quit [Read error: Connection reset by peer]
gentauro has joined #nixos
joelwallis has quit [Remote host closed the connection]
reallymemorable has joined #nixos
glittershark has joined #nixos
thc202 has quit [Read error: Connection reset by peer]
<__monty__>
Hmm, so `nix-build -E '(import <nixpkgs> {}).pass.withExtensions (es: [ es.pass-otp ])` fails on some tests. But `nix-shell -E <the-same>` and then manually running through the phases is completely successful, no test failures. How do I debug this? Is the nix-shell just the environment for pass without the extension, cause that's when the problems with build show up?
<dminuoso>
Mmm. So Ive been thinking for a while how to do letsencrypt properly via nixops and Im really not sure.
<dminuoso>
At best one might cause a cert activation in activationScripts and then just keep a timer for renewal, but that causes somewhat slow boot time.
pamplemousse has joined #nixos
<dminuoso>
I mean if Im willing to live with some racy code, one might perhaps just *fork* some getNewCertificate.sh in system.activationScripts and then maintain a systemd timer to periodically renew.
LysergicDreams has quit [Remote host closed the connection]
<tilpner>
dminuoso: I'm not that familiar with the problem. Could you explain why services for renewal with proper dependencies aren't an option? activation scripts sound very wrong here
<tilpner>
dminuoso: Activation scripts also run at times when there is no network, so they shouldn't do anything that depends on it
<dminuoso>
tilpner: You have a chicken and egg problem.
<dminuoso>
tilpner: In order to renew a certificate it needs to exist in the first place.
<{^_^}>
[nixpkgs] @etu opened pull request #73359 → [19.09] profiles/graphical.nix: Drop systemWide pulseaudio in iso → https://git.io/JerWw
<tilpner>
dminuoso: Is that what the current self-signed services are trying to solve?
<tilpner>
(acme-selfsigned-*.service)
jgt1 has quit [Ping timeout: 268 seconds]
<dminuoso>
tilpner: It's what acme is for yeah, the nixos module no longer works and Im thinking about how to fix it properly.
<dminuoso>
tilpner: In simple terms, in order to obtain a certificate "foo.bar" you run a certbot that runs a web server, and then letsencrypt verifies whether that web server is reachable via "foo.bar". On the initial set up a new certificate is created, and you would then subsequently renew that certificate every month or so.
<tilpner>
Do you renew the same keys, or do you generate new ones to then have signed by them?
<tilpner>
(I thought it was the latter, but you say it's the former)
chloekek_ has joined #nixos
<dminuoso>
tilpner: You can do both, but new certificates are limited in terms of frequency.
<dminuoso>
So letsencrypt strongly encourages to renew existing certificates, rather than recreate them
<{^_^}>
#73235 (by dhess, 2 days ago, open): Hydra fails to start on recent nixpkgs
<infinisil>
Where `gwch: aliased to git whatchanged -p --abbrev-commit --pretty=medium`
<dhess>
it seems like Hydra isn't seeing PERL5LIB anymore.
<dhess>
or at least, not the transitive dependencies of its Perl libs. I can add, e.g., DBIxClass to perlDeps, but then it breaks on something else, which I add... ad infinitum
<dhess>
Hydra's derivation has a strange way of setting PERL5LIB, but I've also tried doing it the way other mkWrapper apps do it, and that also doesn't seem to work.
<zfnmxt>
Are there any other NixOS mobile efforts aside from what samueldr is doing? Any one have some grasp of what the prognosis looks like for a reasonably-usable NixOS phone in the next year or so?
<dhess>
Volth is always breaking my nixpkgs shit :)
fendor has joined #nixos
<shyim>
I am getting random such errors like error: NAR info file '65qgr1w285k5qx9ccbh0fdc0ph8i5f75.narinfo' is corrupt. I did run already nix-store check contents. It didnt helped
<dminuoso>
"Exception: unable to activate new configuration (exit code 3)"
<dhess>
FRidh: thanks! I have had so much trouble trying to figure this one out.
<dminuoso>
Not bad. This is really helpful.
<shyim>
like mine nar info error :D
<dhess>
FRidh: beat you to it ;)
Chiliparrot has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
<rnhmjoj>
what is this "Oh no!" thing in the new python tests? it's driving me insane
kuznero has joined #nixos
<kuznero>
Hi All!
<dhess>
FRidh++
<{^_^}>
FRidh's karma got increased to 12
<dhess>
:*
<dminuoso>
Mmm, is there a list of what exit codes there can be when switching to another configuration?
<kuznero>
I am starting to do some python. Are there any recommendations or tips or guidelines on what is the optimal development flow with Python in NixOS?
reallymemorable has joined #nixos
<dminuoso>
Concretely, I want to figure out what exit code 3 is.
NoctisLabs has joined #nixos
<dhess>
kuznero: it's not a great story IMO. It appears that pypi2nix is probably the best way to do it currently.
hasanzia has joined #nixos
<kuznero>
dhess, is it because of the vastness of the libraries available in python?
<kuznero>
Is there anything like pypi store on nixos side? Like there is for haskell's hackage (if I'm not mixing things up)?
<dhess>
kuznero: no, it's more because Python doesn't have a strictly adhered-to way of building/installing Python packages.
<dhess>
It's the old perennial Python issue, not specific to Nix, really.
NoctisLa1 has quit [Ping timeout: 268 seconds]
<kuznero>
dhess, but can I use same virtualenv, pip tools that I get to use in Ubuntu for example?
<dhess>
kuznero: sure, you can if you want. It's not very Nix-y, though.
<dhess>
but sometimes that's the best you can do without massive changes upstream.
<dhess>
in my experience, anyway. I'm not a regular Python dev, I just run into occasional Python apps that aren't very easy to package in Nix.
<kuznero>
I've tried pgAdmin4 before, but that was uphill at the time.
* kuznero
will be back in ~15 minutes
<dhess>
kuznero: many Python packages are already handled in nixpkgs and you might not need to do this.
<dhess>
it's applications more than libs that I run into problems with.
<acowley>
I need to have a parameter to a function that creates an overlay. This works fine if the user uses `overlays.nix`, as they can import a nix file and apply it to that parameter to produce a proper overlay function. But this does not work if the user wants to use an `overlays` directory. Is there a way I should go about using some field of `nixpkgs.config` to obtain a value for this parameter?
mexisme has joined #nixos
xkapastel has quit [Quit: Connection closed for inactivity]
<acowley>
Should I just ask users to set a field in their config.nix, and then I look for that field in my overlay?
<FRidh>
acowley: no. Maybe you can add your changes in a sub package set so it can be a function?
<acowley>
FRidh: You're suggesting to not use an overlay at all?
kalipso has quit [Remote host closed the connection]
mexisme has quit [Read error: Connection reset by peer]
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/Jerls
<acowley>
I just tried out having a setting nixpkgs.config, and it seems to work fine.
mexisme has joined #nixos
kenran has joined #nixos
<acowley>
Sorry, meant to say, "...having a setting in nixpkgs.config..."
Chiliparrot has joined #nixos
kleisli has quit [Remote host closed the connection]
<tilpner>
dminuoso: I can't view nixos.org currently, my ISP is having problems. How did you resolve nixos-19.09 to this commit?
danshick has joined #nixos
<tilpner>
(And also no search engines except for bing, so...)
<infinisil>
tilpner: A valid use-case for a VPN I'd say!
<infinisil>
Although would be a bother to set up if you know it's resolved soon anyways
<tilpner>
Yes, I guess I could set one up
<__monty__>
tilpner: Would a resolved ddg/google ip help?
<tilpner>
No, DNS works fine
mexisme_ has joined #nixos
<__monty__>
Such a weird issue...
<tilpner>
I am hoping they will resolve it till tomorrow. Otherwise I'll go with the VPN
<infinisil>
It's surprising that our internet works at all lol
<infinisil>
Most of the time anyways
<__monty__>
So, I just updated my channels, nixpkgs-unstable, and now mpv crashes with this error http://ix.io/1WUt Should I report the package broken? How do I do that? Is this even a problem with the package or could it be a bug in nix? (I'm on macOS if that matters.)
<tilpner>
__monty__: I don't know any Darwin, but I doubt it's Nixs fault
mounty has joined #nixos
mexisme has quit [Ping timeout: 268 seconds]
kenran has quit [Ping timeout: 265 seconds]
<slabity>
builder for '/nix/store/ABCDEF-gtk-doc-1.32.drv' failed with exit code 1
<slabity>
I get the above error when I try to rebuild my system, but nix-shell seems to build it fine
jb55 has quit [Remote host closed the connection]
<ArdaXi>
slabity: I had the same issue, could it be you're building a 32-bit version of something?
jb55 has joined #nixos
<slabity>
Actually it looks like the error is only on nixos-unstable-small, not nixos-unstable
<danshick>
Does anyone have any advice for generating images for bare metal deployment? I am in an environment where I cannot control the PXE boot image on the network. Is it possible to generate an image for bare metal install, or should I install manually and then switch to something like https://github.com/andrewchambers/terraform-provider-nix for updates/changes after it is up and running?
<slabity>
danshick: You can easily generate isos from a NixOS configuration
<ma27[m]>
I usually do `nix run ` or use a `nix-shell`
<aanderse>
my users don't know or want to know what nix or nixos is, and i need to keep it that way
<aanderse>
running nixos stable as the core os, over nixops, and will want to access php versions from old versions of nixos as well
<aanderse>
so anyone have anything "better" (easier) than shell.nix + direnv? if lori isn't in stable yet that isn't an option, unfortunately
<__monty__>
You have them running nixos even though they don't want to deal with nixos?
<{^_^}>
[nixpkgs] @nixos-channel-bot pushed 3 commits to nixpkgs-19.09-darwin: https://git.io/Jer8Q
<__monty__>
Lorri isn't a necessity I guess, just makes both shell.nix and direnv so much nicer though.
<lordcirth_>
The users probably don't want to "deal" with any distro or package manager
<aanderse>
__monty__: you bet. i'm a sysadmin and i *hate* sysadmining debian as a php web development platform
<gchristensen>
aanderse: you could make a wrapper around each version of php with a `php7.x` `php7.y` etc. executable name
<__monty__>
Sure but that's not really an option, lordcirth_?
<aanderse>
i've converted probably ~20 debian web servers to nixops+nixos web servers over the past 12-18 months and my developers don't really know or care. epic.
<gchristensen>
:O yay! :D
<aanderse>
gchristensen: yeah that is one option, though direnv maybe even easier as they can just cd to the directory and use php without considering version... though if they don't cd to directory... hmm
<aanderse>
gchristensen: yeah, i think i'm going to win a bunch more servers over to nixos soon because of the ability to mix and match software so easily :)
<dminuoso>
Or rather, because apparently https://github.com/systemd/systemd/pull/10413 hasn't made its way into nixos yet, I cannot see what results some of the systemd units have.
<{^_^}>
systemd/systemd#10413 (by keszybz, 1 year ago, merged): catalog: fix name of variable
<dminuoso>
How can I go about debugging this?
<tilpner>
dminuoso: Post the changes you made to the activation script
<dminuoso>
The underlying systemd unit starts perfectly fine by hand
<tilpner>
dminuoso: That's all? No fiddling with activation units?
<dminuoso>
tilpner: That's all.
<rihardsk[m]>
heya, anyone run into this error before? "error: moving build output '/nix/store/pxx1qkxgn1wq7qkpx6sq5gcqisl2qxjk-documentation-highlighter' from the sandbox to the Nix store: Permission denied"?
<rihardsk[m]>
i tried to crosscompile nixos for aarch64 (for the rockpro64 single board computer). seems like a nix bug to me
<dminuoso>
tilpner: I've fiddled with it for a while, but gotten rid of it..
<samueldr>
rihardsk[m]: cross-compiling or compiling through qemu-user?
_deepfire has quit [Remote host closed the connection]
<deni>
I have nix installed on Debian locally (laptop) and a NixOS server. I want to use the remote server as a builder. nix ping store works fine. I've added `builder = ssh://builder` (and appropriate .ssh/config) to ~/.config/nix/nix.conf` but when I try to run a build locally with `nix-build -j0` to make sure that everything gets built on the server it fails with lack of valid signature. How do I set up
<deni>
the remote builder to properly sign the packages?
<samueldr>
(asking since it changes the stack in quite radical ways)
<rihardsk[m]>
lordcirth_: oooh, hadn't come upon that before. thanks, will check it out
qubasa has joined #nixos
<samueldr>
though right now on unstable (or master) it won't work
<rihardsk[m]>
samueldr: cross-compiling. i couldn't figure out how to do it through qemu, so i thought i'd try my luck with cross-compiling first
<samueldr>
alright
<lordcirth_>
samueldr, btw, I was planning on trying to remove the broken package; but I can't figure out how to run a nix why-depends on the image build.
<lordcirth_>
Neither did I manage to find where sdImage.nix specifies the list of installed packages
<tilpner>
dminuoso: system("/nix/store/y7qs6g4qyss8f4a4sx8an4hqwx2v4093-systemd-243/bin/systemctl", "daemon-reload") == 0 or $res = 3;
<samueldr>
regressions on nixpkgs-side, where dependencies low in the tree are doing things that I don't quote know how to fix right now
<tilpner>
dminuoso: That's the line responsible for the exit of 3
kleisli has quit [Ping timeout: 265 seconds]
<tilpner>
dminuoso: You should now check in which cases systemctl daemon-reload returns 3
<{^_^}>
[nixpkgs] @nixos-channel-bot pushed commit from @risicle to nixpkgs-19.03-darwin « opencv4: add patches for CVE-2019-14491, CVE-2019-14492, CVE-2019-14493 & CVE-2019-15939 »: https://git.io/JermO
<dminuoso>
tilpner: Hold on, where did you get that from?
<tilpner>
dminuoso: That's from bin/switch-to-configuration, which nixops calls
<aminechikhaoui>
tilpner I don't think daemon-reload would return 3, but for any error code switch-to-configuration would return 3
<dminuoso>
tilpner: systemctl daemon-reload does not return 3, the whole thing returns 3 if daemon-reload gives me a non-0 exit code.
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @lopsided98 to master « libzip: 1.3.0 -> 1.5.2 »: https://git.io/Jer4Y
qubasa has quit [Ping timeout: 240 seconds]
<deni>
I don't think I understand the difference between a builder and binary cache and substiters. :(
<tilpner>
deni: They can all be the same machine
<etu>
aanderse: If you really need several php versions available under different names I would make a derivation that basically makes a symlink from $out/bin/php7.3 -> /nix/store/...php-7.3/bin/php
qubasa has joined #nixos
<tilpner>
deni: I don't know if the distinction between binary cache and substituter is useful
<deni>
tilpner: sure but the docs are unclear.... I don't want to start serving a binary cache over http. I just want to use ssh to build stuff remotely on the server cause they're faster than my laptop.
<aanderse>
etu: not as big of a fan of direnv? (i've never used direnv, but it seems like a good solution as well)
<deni>
tilpner: I thought that was what we were talking about...that is what I'm trying to do
<tilpner>
(See nix.buildMachines or builders in nix.conf)
<dminuoso>
tilpner: Gah. Thanks, turns out there's another spot - and the error appeared in journal but it was so innocent, it looked like a warning rather..
<dminuoso>
The ownership on /var/www was not correct..
kleisli has joined #nixos
<etu>
aanderse: Direnv is nice
<deni>
tilpner: I did read that in the docs but that syntax doesn't seem to work on non nixos machines (the laptop is running Debian)
<etu>
aanderse: I use that a whole lot
<tilpner>
deni: nix.buildMachines is NixOS-specific, but builders is not
<tilpner>
deni: By default, builders refers to /etc/nix/machines, where you can add new build machines
<tilpner>
Someone please fetch the docs for deni?
<rihardsk[m]>
lordcirth_: my setup looks pretty similar to that in the github link. is that expected to be working currently (i'm on the unstable channel)? if so, i could try that and see if anything changes
<deni>
tilpner: I don't know what syntax `builders` has...the distinction between nixos and non nixos is very bad in the docs (and sometimes mixed)
<clever>
deni: /etc/nix/machines has the same format on all platforms
<lordcirth_>
rihardsk[m], no, there are at least 2 packages that don't build; one can be fixed by reverting a commit, the other is hard
<{^_^}>
[nixpkgs] @jonringer pushed commit from @filalex77 to master « cargo-deb: init at 1.21.1 »: https://git.io/Jer4Z
<deni>
tilpner: I don't really want to have stuff in the global /etc/ folder since that was discouraged the last time I did it and I was told to use ~/.config/nix
<{^_^}>
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. See: tofu-vim
<deni>
sure thing.... once I figure out the exact syntax...cause when I copied the nix expression from here https://nixos.wiki/wiki/Distributed_build in ~/.config/nix/nix.conf it didn't work
<tilpner>
gchristensen: sha256sum /dev/null is what I use when I need a wrong hash
<clever>
tilpner: in vim, you can just do: 52i0<escape>
<deni>
clever: I don't think I understand the direction this is going.... I'm not copying over pre-built things from my laptop to the server .... am I? I'm trying to copy unbuilt things, have the server build it, and then copy the result down to the laptop...Or is this not how it works?
<clever>
deni: when you use a build machine, you first download all dependencies to the current machine, then you copy them to the remote machine, and ask it to do the build, then copy the product backwards
<lordcirth_>
rihardsk[m], I'm looking into removing the offending package, but I'm still trying to figure out how to use 'nix why-depends' on the sdImage.nix package list. Or where that list is defined, for that matter.
<tilpner>
deni: The Nix expression in the wiki is for NixOS
<clever>
slabity: nixops will override the default private key location, to use its own keys
<clever>
slabity: so it never loads ~/.ssh/id_rsa directly
<rihardsk[m]>
lordcirth_: any suggestions on how to proceed then? should i try qemu instead?
<rihardsk[m]>
for a bit more context on how my setup looks like - i found a prebuilt image that came from a nixpkgs fork with patches for the rockpro64 sbc https://github.com/Thra11/nixpkgs/tree/rockpro64 and flashed it onto the device. works great. now i'd like to be able to push configuration changes and deploy some software/services from the comfort of my x86 machine, using nix-build and nix-copy-closure. so that's why i'm atempting
<rihardsk[m]>
crosscompilation
<slabity>
Tried deleting root's .ssh directory, but no luck
<clever>
slabity: but if an ssh agent is running, those keys will leak into nixops and give it more power then it "should have"
<slabity>
Oh, that makes sense. What would I need to do then?
<lordcirth_>
rihardsk[m], I dunno. Hopefully I will have time soonish to try shrinking the image.
<clever>
deni: when true, the remote machine will obey its own `substituters` in its own `nix.conf`, to fetch what it can first
<deni>
clever: that flag is supposed to be set on the laptop?
<clever>
deni: its set on the local machine, the one starting the build
<WhatisRT>
Hi! Having a wierd issue with ethernet on my RPi: in 19.03 everything works just fine, but after updating to 19.09 it won't connect to ethernet with my router, but it will connect to ethernet shared from my laptop. Does anyone know something about this?
<deni>
clever: ok cuase I found that and set it but wasn't sure honestly where it was supposed to be set.
<clever>
WhatisRT: i think the default for dhcp may have changed
spacetato has quit [Quit: leaving]
<deni>
clever: another question.... if something is built already locally but isn't on the build machine, and I have -j0 will it rebuild everything on the remote machine or will it transfer the already built things?
<slabity>
clever: That did not seem to work... You mean login as root, start ssh-agent, and then add my user's id_rsa? Or something else?
<WhatisRT>
clever: I read something like that on the manual, but it isn't clear to me how to configure it properly
<tilpner>
infinisil: One sshuttle later, I can access nixos.org again :)
<lordcirth_>
tilpner, do you use tinc a lot? You like it?
<tilpner>
tinc is good
<lordcirth_>
Oh, tinc does mesh routing. Didn't know that. Definitely need to try it
cosimone has quit [Remote host closed the connection]
<tilpner>
I can recommend tinc unless you need Gbit speeds, or all your nodes are private
<deni>
tilpner: clever anyway thanks for the help...it's building now and I'll wait and see how it goes. I think it's actually uploading the stuff I already have built locally
<deni>
clever: I just realized I'm asking "does it copy over prebuilt stuff" and you wrote above needing to be a trusted user to be able to copy prebuilt stuff.... :D
<exfalso>
Hi, I'm using builtins.filterSource in a derivation, however I use a filter function that simply returns true for directories. This means the source tree ends up with empty directories (that originally contained files that don't match the filter). This in turn invalidates nix caching when a new directory is added.
zupo_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<exfalso>
Is there a way to avoid inclusion of empty folders, and folders that contain empty folders etc?
zupo has joined #nixos
<{^_^}>
[nixpkgs] @nixos-channel-bot pushed 3 commits to nixos-19.09-small: https://git.io/Jer8Q
cosimone has quit [Remote host closed the connection]
<deni>
clever: tilpner this appears to be the syntax for /etc/nix/machines right http://dpaste.com/0TPQNZ8?
<clever>
slabity: as the user that runs nixops, start ssh-agent, and do an ssh-add
<clever>
deni: the features at the end must be , seperated
<clever>
deni: enless your trying to set the mandatory features
fendor_ has quit [Ping timeout: 250 seconds]
<mica[m]>
If an application gets an update, can I get that update into the latest release? So darktable was at 2.6.2 when 19.09 came out, but dark table has since updated to 2.6.3. How do I get that into 19.09?
orivej has quit [Ping timeout: 265 seconds]
<infinisil>
mica[m]: stable only gets bug-fixes and security updates
<slabity>
clever: It did not seem to work. The key was added just fine, but I still get the same error
<clever>
slabity: does the key you added actually exist in /root/.ssh/authorized_keys ?
<infinisil>
mica[m]: Use the unstable channel (or pick the packages you want from it) if you want more up-to-date packages
<deni>
clever: by features you mean "kvm,benchmark"? I should update that in the docs then
<slabity>
Oh, I generated a new one. Nevermind I'll try adding it
<deni>
clever: I really just want to set system, maxJobs and speedfactor but there doesn't seem to be a way to do that in this file
<clever>
deni: some of the columns can be , seperated, space always seperates columns from eachother
<slabity>
clever: Did not seem to work. I can ssh root@localhost without a password now though
<slabity>
So I believe it took effect
<clever>
slabity: are you running nixops in the same terminal?
<exfalso>
is there a way to access the list of matching paths of builtins.filterSource for further processing in nix?
<slabity>
No
<clever>
slabity: nixops can only access the ssh agent if the $SSH_AUTH_SOCK var is set right
<slabity>
Ah, that seems to have fixed it. Thanks
<slabity>
clever++
<{^_^}>
clever's karma got increased to 240
<tilpner>
exfalso: Perhaps you can call readDir during filterSource, but that sounds like it would be really expensive
<slabity>
So another question I had, is it possible to deploy single logical machines in a network instead of the entire network? Like pick-and-choose?
<{^_^}>
[nixpkgs] @jonringer pushed commit from @Pamplemousse to master « pythonPackages.itanium_demangler: init at 1.0 »: https://git.io/JerB0
Mrmaxmeier7 has quit [Remote host closed the connection]
<deni>
clever: I can't really get that syntax in the `machines` file to work. It only works if I specify `builders = ssh://builder` in nix.conf directly. I suspect it's because I can't omit the ssh key in the middle ... but I don't need it since I specify the stuff in .ssh/config as the docs suggest. :/
danshick has quit [Quit: Leaving]
ng0 has quit [Ping timeout: 260 seconds]
alexherbo2 has joined #nixos
avn has quit [Remote host closed the connection]
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #73083 → llvm_9: disable failing tests on 32-bit ARM → https://git.io/JeVBX
<{^_^}>
[nixpkgs] @matthewbauer pushed 3 commits to master: https://git.io/JerBr
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #73084 → llvm_7: disable failing test on armv6l → https://git.io/JeVRv
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/JerBo
<{^_^}>
[nixpkgs] @colemickens opened pull request #73365 → firefox-wrapper: use libglvnd → https://git.io/JerBi
civodul has quit [Quit: ERC (IRC client for Emacs 26.3)]
justanotheruser has quit [Ping timeout: 268 seconds]
selfsymmetric-mu has quit [Remote host closed the connection]
bvdw has quit [Quit: bvdw]
romildo has joined #nixos
bvdw has joined #nixos
<slabity>
Strange issue: If I deploy with NixOps, I am unable to open up any Xorg windows after it finishes. However, if I use `nixos-rebuild` it goes back to working just fine...
<slabity>
Same exact config
<infinisil>
slabity: Same nixpkgs version?
ng0 has joined #nixos
<slabity>
Specifically, applications fail with "Cannot parse arguments: Cannot open display:"
<slabity>
infinisil: Yes. I even explicitly set my NIX_PATH on both commands
<slabity>
Let me see if the LD_LIBRARY_PATHs contents match after each one...
<infinisil>
slabity: Can you `nix-diff` the system from nixos-rebuild and nixops?
<infinisil>
Like the /nix/store path /run/current-system points at after nixos-rebuild/nixops deploy
<slabity>
This is the first time I've heard of `nix-diff`. Let me see
reallymemorable has quit [Quit: reallymemorable]
<{^_^}>
[nixpkgs] @adisbladis merged pull request #73359 → [19.09] profiles/graphical.nix: Drop systemWide pulseaudio in iso → https://git.io/JerWw
<{^_^}>
[nixpkgs] @adisbladis pushed 2 commits to release-19.09: https://git.io/JerRJ
<{^_^}>
[nixpkgs] @nixos-channel-bot pushed commit from @risicle to nixos-19.03 « opencv4: add patches for CVE-2019-14491, CVE-2019-14492, CVE-2019-14493 & CVE-2019-15939 »: https://git.io/JermO
justanotheruser has joined #nixos
fusion809 has quit [Remote host closed the connection]
<infinisil>
slabity: Do the /nix/store files directly not work?
<infinisil>
slabity: But if not, you can find out the .drv file for a non-.drv file with `nix-store -q --deriver /nix/store/..`
cosimone has joined #nixos
leo__ has joined #nixos
<slabity>
So /run/current-system is a directory linked to a system-profile in the /nix/store. But it does not specify which .drv it came from. The command you just gave me does not seem to work.
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @r-ryantm to master « postgresql11Packages.pg_partman: 4.1.0 -> 4.2.2 »: https://git.io/JerRZ
<nh2>
clever: hmm, I'm having some trouble with my ghc override. It prints `Build with [the ghc I want]`, but then the actual build is being done with a different GHC, namely the `nativeGhc = buildHaskellPackages.ghc;` (from the generic haskell builder). Do you know how I can override that one as well?
NoctisLabs has quit [Quit: WeeChat 2.6]
reallymemorable has joined #nixos
NoctisLabs has joined #nixos
glittershark has quit [Ping timeout: 252 seconds]
Fare has quit [Ping timeout: 240 seconds]
mexisme_ has quit [Ping timeout: 268 seconds]
acarrico has joined #nixos
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
sondr3 has joined #nixos
Soo_Slow has quit [Read error: Connection reset by peer]
Soo_Slow has joined #nixos
<clever>
nh2: are you overriding the ghc of haskellPackages?
<sondr3>
anyone here use JetBrains IDE for Rust with the Mozilla overlay? I can't get it to figure out the src of the rust lib, it wants to import things like `std::String` and such
<nh2>
clever: yes
<clever>
nh2: is it a normal or a cross compile?
ambro718 has quit [Ping timeout: 240 seconds]
<{^_^}>
[nixpkgs] @JohnAZoidberg opened pull request #73369 → argp-standalone: Find ar when cross compiling → https://git.io/JerRX
cosimone has quit [Remote host closed the connection]
cosimone has joined #nixos
jgt1 has joined #nixos
ayerhart has joined #nixos
<{^_^}>
[nixpkgs] @risicle opened pull request #73370 → [r19.03] samba: add patches for CVE-2019-10218, CVE-2019-14833 & CVE-2019-14847 → https://git.io/JerRb
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @puffnfresh to master « unison-ucm: 1.0.M1d-alpha -> 1.0.M1e-alpha »: https://git.io/JerRh
romildo has quit [Quit: Leaving]
<{^_^}>
[nixpkgs] @worldofpeace pushed commit from @W95Psp to master « ppsspp: use qt5's mkDerivation »: https://git.io/Jer0e
<{^_^}>
[nixpkgs] @worldofpeace pushed commit from @W95Psp to release-19.09 « ppsspp: use qt5's mkDerivation »: https://git.io/Jer0v
<elvishjerricco>
Hm. I've got a proprietary .deb I want to install, but I don't want to check it in to my system config repo, and I do want it to update with my system's nixpkgs. Having trouble meeting those requirements together...