<matthewbauer>
nh2: yeah, was pkgsMusl.coreutils working before that commit? it would surprise me if that pr broke that test, but i guess it's possible
<genesis>
we need a nix mentorship program :-)
<genesis>
i do coffee :-°
acarrico has quit [Ping timeout: 258 seconds]
<aanderse>
genesis: feel free to ask away in here any questions you have :)
<nh2>
matthewbauer: ah, looks like before the commit the test printed: `date-debug.sh: skipped test: Timezones database not found`
kisonecat has quit [Ping timeout: 255 seconds]
<genesis>
aanderse : i try to do some reviewing, i would like to be reviewed on that, that's not a part i do generally.
<aanderse>
are you bignaux?
<genesis>
yes
silver has quit [Quit: rakede]
<matthewbauer>
nh2: yeah, which is annoying. maybe ping dtz about this, i suspect coreutils devs aren't doing any testing on musl, so it's not too surprising if it doesn't work
ayerhart has quit [Read error: Connection reset by peer]
<nh2>
matthewbauer: or maybe this city `Edmonton`'s time zone changed? On Ubuntu it certainly reads `/usr/share/zoneinfo/America/Edmonton`
<nh2>
matthewbauer: if I compile the latest coretuils on Ubuntu, it does show me what the test expects, so probably something is odd with nixpkgs or musl
<lightbulbjim>
Any feedback is welcome as I'm pretty new to contributing
vk3wtf has joined #nixos
<genesis>
hi lightbulbjim : i'm bignaux ;)
InFlames has joined #nixos
<genesis>
i think you still forget the hicolor icon theme btw.
<genesis>
but it could be you suppose the icon has to go there
orivej has quit [Ping timeout: 246 seconds]
<genesis>
and it hasn't. why not copy the icon in $out/share/icons/
<lightbulbjim>
I was just tinkering with that actually. When I build hicolor-icon-theme by itself I get the share/icons/hicolor/* tree created, but adding it to my derivation (I tried both nativeBuildInputs and buildInputs) seems to do nothing.
<lightbulbjim>
Is it just used to create the directory tree? If so is it really necessary? Creating the icon directories in postInstall seems to be working (and I've done that on some other packages too).
<genesis>
my issue was you're supposing people use hicolor theme
<genesis>
and the icon has to go there.
<drakonis>
libgl keeps giving me a hard time :|
<genesis>
and i don't see such requirement in the repository.
<lightbulbjim>
Happy to add hicolor-icon-theme to buildInputs, although I don't see any difference on my system when I tried it.
<genesis>
that's a bit too much when you just have to copy the icon in a directory $out/share/icons/
<lightbulbjim>
Oh, the graphicsmagick thing is because the upstream source only includes a single .ico icon, but XDG wants png, svg or xpm. So I use graphicsmagick to extract the imbedded images from the .ico into pngs.
<genesis>
yes that's a good idea, i get it, but please PR the authors with it for next time
jackdk_ has quit [Ping timeout: 245 seconds]
<genesis>
and tell him to do a release in the same time :)
kisonecat has joined #nixos
<genesis>
you could have both in less time than your PR will be merge
<lightbulbjim>
I can send a patch upstream and pull it into my derivation if you like. Only thing is it will be a bit annoying if the author rejects it, but I suppose we can deal with that if it happens.
<genesis>
don't forget to mention the PR in the other side, but yes it's a good way to do
<drakonis>
is patchelf available in declarative form?
<lightbulbjim>
Ok, I'll do that and update the PR when ready. Thanks!
<genesis>
but i fear the version you packaged is so old, that this pb has perhaps been already fixed
<lightbulbjim>
Oh I see
<genesis>
when you see project is active but not do release, ask :)
<lightbulbjim>
It's the latest stable release; the only options are the stable releases (every ~1 year) or master
<genesis>
yeap, one release a year is not so bad :D
<drakonis>
to be fair, if you poke folks here, they can merge it
<drakonis>
and its not one per year
<drakonis>
its twice a year, 6 months each cycle
<genesis>
we speak about the software lightbulbjim packaged
<drakonis>
ohhh
<drakonis>
okay
<drakonis>
fair enough
<genesis>
i know nix devel prefers stable release, but i often they are less stable than tip :)
<dmj`>
is it possible to forward traffic through an SSH connection established by nixops
Supersonic has joined #nixos
<angerman>
matthewbauer: I've looked at the emulator stuff (specifically qemuArch) in nixpkgs, any reason not to use aarch64 qemu but always defaulting to arm?
<matthewbauer>
angerman: isArm is just the 32-bit cpu. It should be isAarch32 to make that more clear though
<angerman>
matthewbauer: right. isAarch32 -> arm and isAarch64 -> aarch64, I guess.
<teto>
do we have a meta issue for the haskell ecosystem ?
<{^_^}>
[nixpkgs] @volth opened pull request #61423 → nixos/desktop-managers: let them work 'services.xserver.enable = false' → https://git.io/fjW8u
<angerman>
matthewbauer: `${hostPlatform.emulator pkgs}` is also a nice trick, but there is no feature lookup, is there? E.g. do I know that that emulator will support filedescriptors?
ricekrispie2 has joined #nixos
bsima1 has joined #nixos
<ar1a>
anyone know if its possible to mute {^_^} for notices but not messages?
fusion809 has quit [Remote host closed the connection]
Ralith_ has quit [Remote host closed the connection]
<angerman>
matthewbauer: the issue with ghc (e.g. what I might be interested in) is that I can only use `iserv` directly if I have filedescriptors, otherwise I need to fall back to the proxied network bridge.
<matthewbauer>
angerman: yeah does wine work with iserv?
<angerman>
iserv-proxy + iserv-remote, yes. I couldn't get bare iserv to work.
<angerman>
matthewbauer: so you have ghc <-(fds)-> iserv-proxy <-(network)-> wine remote-iserv
<clever>
angerman: i'm curious as to why stdin/stdout of `wine iserve.exe` didnt work?
<angerman>
stdin/stdout does, but ghc chooses random fds for it's communication channel outside of stdin/stdout/stderr
<angerman>
you usually get 5 and 7 iirc.
<clever>
ahh
<clever>
also of note, wine doesnt do anything to block the linux api
endformationage has quit [Ping timeout: 245 seconds]
<clever>
one sec
<clever>
hnd := LoadLibrary('ntdll.dll');
<clever>
 Assigned(GetProcAddress(hnd, 'wine_get_version')) or Assigned(GetProcAddress(hnd, 'wine_nt_to_unix_file_name'));
<clever>
angerman: you can query the addresses of internal wine symbols, and easily detect wine
<clever>
and if you can convince the compiler, you can just staticly link in some linux muslc, and call the linux open() or read() and bypass wine
<clever>
angerman: i think this is the section we want
<clever>
the first 80% of that page, is more about taking "windows" source, linking against wine, and creating a "native linux" binary
<angerman>
clever: but wouldn't that mean that we would tie ourselves down to wine, just to prevent the local network bridge we do right now? And as it is right now we can swap out wine for a real windows machine if we wanted to.
<clever>
yeah, you would probably want to make it optional, so you can skip the network if wine and an FD is available
<angerman>
That sounds like a project for the backburner.
<clever>
yeah, if its already working with network, you could keep it like that for now
<clever>
only thing i can see you gaining from winelib, would be shm
<angerman>
That qemu-arm blows up with Signal (11) on iserv when built with a performance ghc built is something I'm tempted to look into before that :D
<clever>
moving large chunks of data to the child proc via shared memory
<angerman>
clever: right. It's working with network right now (for our not even trivial code base).
<angerman>
clever: and it's mostly askign the child process to load objects/dlls/ for which the communication channel just hands over the paths on the file system.
Rusty1 has quit [Quit: Konversation terminated!]
<clever>
what about the blob of TH code that it compiled and now has to run?
<angerman>
The only "memory" part are small ByteCodeObjects we ship. They are in the sub 10kb range for even really large ones.
<clever>
though thats likely very small
<clever>
and all the import statements are covered by the .dll files the deps provide
<angerman>
yes. The staging restriction basically means that the function you call in your splice has to come from an object file.
<clever>
this is why i can see value in the TH code having its own set of imports, that are clearly isolated from the target deps
<clever>
then you could compile the TH for a different platform
<clever>
oh, i just remembered, there was some talk before about the compiler leaving the entire parsed AST in the .hi files?
<clever>
ignoring #ifdef mess's, could you take that AST, and then rebuild the closure of a given splice, for x86-linux?
<angerman>
clever: we have lots of the ast in the .hi files. obsidian system is using something like that for cross TH. It works for most, just breaks at the boundary where you host compiler has different backing types than your target. (e.g. Chars backed by Int or Words)
wfranzini has quit [Remote host closed the connection]
<clever>
ah
<clever>
but that still has IO in the TH splice?
<angerman>
That appraoch (not fully automated) is what evilSplicer/zeroTH do. By going native with the TH splice, you are fully bound by the target, it's results and features. Has pros and cons as wlel.
<angerman>
clever: well, you just take the fully evaluated splice from the host and put it into the target.
<clever>
i'm more thinking about things like makeLenses, its a pure function and doesnt really care about IO or platform
<angerman>
that means process and file logic is run on the host.
<clever>
would it even care what type a Char is? its just generating AST's
<angerman>
clever: right, that trivially works if you just swap in the splice result from the host when you build for the target.
<angerman>
that TH doesn't differentiate between some pure macro-expansion like sub language and the IO/file/process side is quite annoying.
<angerman>
Otherwise you could build a complete solution for the pure macro-expansion part.
<angerman>
Types only help you as much as you trust your code not to call unsafePerformIO or some other escape hatchery.
<clever>
flip on safe haskell and the compiler wont let you unsafePerformIO!
<clever>
just beware of the io that must not be named!
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
dansho has quit [Remote host closed the connection]
noam_ has quit [Ping timeout: 244 seconds]
dansho has joined #nixos
dne has joined #nixos
wfranzini has joined #nixos
python476 has joined #nixos
wfranzini has quit [Client Quit]
wfranzini has joined #nixos
esmerelda has quit [Ping timeout: 257 seconds]
esmerelda has joined #nixos
Tucky has joined #nixos
<MasseR>
Could someone explain what's happening here. I have two derivations, both evaluating to the same hash, but they shouldn't (two lines coming up)
<{^_^}>
cannot import '/nix/store/rn7av2h6j6ipl9a2ax13xmac8y96lgz1-cabal2nix-scalpel-core-0.7.0', since path '/nix/store/49ivyjw6bpiq49j4yyaxggxqadvvyic8-cabal2nix-scalpel-core-0.7.0.drv' is not valid, at /va...
<clever>
callHackage just combines callCabal2nix and hackage2nix
<clever>
so it should work
<MasseR>
(0.7.0 should fail, but I want 0.6.0 which exists in hackage but not in 19.03)
<clever>
you may also need callHackageDirect which has a slightly different api
aszlig has joined #nixos
<clever>
MasseR: what happens if you try callHackage?
<MasseR>
... I might have gotten my nix store all messed up
<angerman>
clever: ok. So stupid example: the url returns `$<date:8byte>$<data>` and the date is mutated every second. Now if I just write 8 zero bytes into the fetched data I would still get the same output hash right?
<angerman>
clever: alright. I'll look into trimming the hackage index then...
<{^_^}>
[nixpkgs] @Mic92 pushed commit from @Izorkin to release-19.03 « zfsUnstable: 0.8.0-rc4 -> 0.8.0-rc5 »: https://git.io/fjW0n
Jetien_ has joined #nixos
<clever>
angerman: basically, if a derivation has the following attributes: outputHash, outputHashAlgo, outputHashMode
<clever>
angerman: then it is a fixed-output derivation, and gains network access
<clever>
outputHashAlgo is usually sha256, but something like yarn2nix can just copy the hash&algo from yarn.lock, and not even fetch things when translating
<clever>
outputHashMode = "flat" means that you are giving the raw hash for the $out file
<clever>
outputHashMode = "recursive" means you are giving the hash of the nar archive of $out (it can be a file, dir, or symlink)
<clever>
nix-hash, without --flat, gives you the recursive hash
<clever>
angerman: this will recursively fetch repos, parse the DEPS file in the root of each, and generate an attrset of fixed-output derivations, then generate a runCommand that copies them all into one dir
<clever>
angerman: you will need at least the hash of the hackage index at that given time, and that hackage index would need to contain the hashes of the cabal files and source tars
___laika has joined #nixos
<angerman>
clever: I have hackage.nix for that.
<angerman>
I need the indexState only to make cabal produce a deterministic plan.json file
<clever>
ah
<angerman>
the constraint is that your hackageIndexState < the hackage.nix checkout you have. That's not perfect but should be managable as we auto-update hackage and stackage.nix periodically.
pie_ has joined #nixos
mexisme has quit [Ping timeout: 246 seconds]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<{^_^}>
[nixpkgs] @Lassulus pushed commit from @r-ryantm to master « ttyplot: 1.2 -> 1.4 »: https://git.io/fjWEh
FedyaPryanichkov has quit [Quit: Quit]
cfricke has joined #nixos
chaoflow has quit [Ping timeout: 246 seconds]
Makaveli7 has joined #nixos
<mog>
is there a way to temporarily disable a binary cache?
<qyliss^work>
mog: --option substitute false
toppler`` has quit [Ping timeout: 246 seconds]
dansho_ has joined #nixos
<mog>
qyliss^work, perfect
<mog>
thanks
drager has joined #nixos
<drager>
Hi, what happened to androidenv.platformTools in 19.03?
toppler`` has joined #nixos
dansho has quit [Ping timeout: 258 seconds]
kozova1 has joined #nixos
<kozova1>
Hello, I'm trying to run this command: nix run nixpkgs.hplipWithPlugin -c sudo hp-setup,
<kozova1>
however that doesn't work due to this error: error: Package ‘hplip-3.19.1’ in /nix/store/via1aa5d96v2qz4yyg978x51hnampjaz-nixos-19.09pre179307.bc94dcf5002/nixos/pkgs/misc/drivers/hplip/default.nix:217 has an unfree license (‘unfree’), refusing to evaluate.
<kozova1>
{ nixpkgs.config.allowUnfree = true; }
<kozova1>
a) For `nixos-rebuild` you can set
<kozova1>
in configuration.nix to override this.
<kozova1>
b) For `nix-env`, `nix-build`, `nix-shell` or any other Nix command you can add
<kozova1>
{ allowUnfree = true; }
<kozova1>
to ~/.config/nixpkgs/config.nix.
<kozova1>
(use '--show-trace' to show detailed location information)
<{^_^}>
[nixpkgs] @peti pushed to master « xautolock: bump to latest git version »: https://git.io/fjWu4
<kozova1>
Hello, I'm trying to run this command: nix run nixpkgs.hplipWithPlugin -c sudo hp-setup, however that doesn't work due to this error: https://pastebin.com/QCgBmdZd
<xorAxAx>
kozova1, have you tried to understand the error message?
<kozova1>
yeah, and I think I do
<kozova1>
however, my configuration.nix has nixpkgs.config.allowUnfree = true; inside it
<kozova1>
and I have no idea how to apply the 2nd fix
<xorAxAx>
you need to put it into tyour local conf
gal_bolle has joined #nixos
<xorAxAx>
because the configuration.nix is only used for the system profile
<kozova1>
ok, is that local conf ~/.config/nixpkgs/config.nix ?
<xorAxAx>
yes
<kozova1>
how do I init that file?
<kozova1>
like, is it supposed to have something in it?
<ottidmes>
I like to the idea of managing my SSH files through my NixOS configuration as well, I saw this mentioned here on the IRC once, but I cannot quite remember the approach nor find it by googling about it. So what I want to achieve is instead of having my id_rsa and such in ~/.ssh/ per user, to have something like /etc/ssh/per-user/name/
<xorAxAx>
i think you can copy and paste the contents from the error message
__monty__ has joined #nixos
<kozova1>
dammit now I get another error
<kozova1>
how can I pass an environment variable into that command?
<kozova1>
xorAxAx I get this: ImportError: No module named dbus
<kozova1>
in my experience I can fix that by adding this env variable: 'PYTHONPATH=/nix/store/dil3g0x4vx07bca34mg5w6cqzl7qq9jr-python2.7-dbus-python-1.2.4/lib/python2.7/site-packages
<kozova1>
however I have no idea how to pass that
sigmundv has joined #nixos
<{^_^}>
[nixpkgs] @FRidh merged pull request #51641 → Allow skipping the use of the PYTHONNOUSERSITE variable → https://git.io/fpDXx
<___laika>
im having another issue with running vscode inside of a nix-shell :/ i installed hie-ghc864 with all-hies, and vs-code is starting pointing at a ghc864 compiler, but for some reason hie in vscode fails with "<command line>: cannot satisfy -package-id r-0.1.9.2-b277754a:r-0.1.9.2-b277754a is unusable due to missing dependencies:exceptions..."
_e has quit [Remote host closed the connection]
_e has joined #nixos
<___laika>
(hie blames it on ghc-mod, for which i have seen lots of issues with similar error messages, but almost all of them seemed to be based on .stack-work issues
<python476>
for a newcomer it feels more like a deep sea to drown in, but one can also get lost in a forest too
<python476>
you use nixos as a daily driver ?
<python476>
I wonder if companies use it ..
<gchristensen>
companies do use it
<python476>
pretty cool
<python476>
it's worth being known a bit more than nix conferences :)
<python476>
gchristensen: the hydra part only or the user level os too ?
<gchristensen>
my full time job is around Nix and NixOS, and I run it on 3 laptops, and a half dozen other systems (personally. I have more running NixOS but they're not my own)
<gchristensen>
sure
<python476>
cool, super cool
<{^_^}>
[nixpkgs] @joachifm merged pull request #61436 → [Backport] Tor browser related stuff for 19.03 → https://git.io/fjWBT
<{^_^}>
[nixpkgs] @joachifm pushed 7 commits to release-19.03: https://git.io/fjWgq
<{^_^}>
[nixpkgs] @joachifm pushed 8 commits to release-19.03: https://git.io/fjWgm
<gchristensen>
there are companies using NixOS on servers and laptops
silver has joined #nixos
<{^_^}>
[nixpkgs] @alexeymuranov opened pull request #61450 → [19.03] Backport GDM PAM fix (cherry-picked from master) → https://git.io/fjWgY
<Synthetica>
python476: NixOS is easily mature enough for a daily driver, I honestly can't imagine switching to anything else anytime soon
srid has quit [Ping timeout: 268 seconds]
Anton-Latukha has joined #nixos
<python476>
Synthetica: I'm still on arch, mostly because I have zero neurons left to learn the details of nix/guix
<python476>
now how skilled are you ? I'm mostly curious about nix/guix being mainstream OSes
<python476>
for slightly tech saavy users (not kids or grandpas)
<python476>
(brb)
kisonecat has joined #nixos
srid has joined #nixos
<goibhniu1>
python476: you need to be comfortable using the command line, editing text files and recognising things like syntax errors
rprije has quit [Ping timeout: 246 seconds]
<goibhniu1>
it's also a particularly nice distro to maintain for friends & family
riottest000[m] has joined #nixos
noam_ has joined #nixos
alexherbo21 has joined #nixos
alexherbo2 has quit [Ping timeout: 246 seconds]
noam has quit [Ping timeout: 258 seconds]
acarrico has joined #nixos
kisonecat has quit [Ping timeout: 245 seconds]
hamishmack has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
bsima1_ has joined #nixos
<typetetris>
are there general guidelines wether to use builtins.fetchGit or builtins.fetchTarball? Which one will be faster?
civodul has quit [Ping timeout: 252 seconds]
<bsima1_>
fetchTarball will be faster
<bsima1_>
it doesn't have to clone the entire git repo
<__monty__>
I'm trying to get cabal-install installed from the github repository but I'm getting this error and I can't find any documentation on callcabal2nix and how to pass this sourceCabalDir which I think I need to change because cabal-install's in a subdir of the repo. Anyone familiar with this? http://ix.io/1IV0
<immae>
bsima1_: whell, neither is fetchGit actually (maybe he does, but he doesn’t have to)
Tucky has quit [Quit: WeeChat 2.4]
jackdk_ has joined #nixos
Ariakenom has quit [Quit: Leaving]
<immae>
if you have a specified revision you’re not obliged (again: per git protocol, I don’t know what fetchgit does) to get the whole repo, and then it’s more or less equivalent to fetching a tarball
<bsima1_>
does anyone know how I look at logs from NixOS auto-upgrades?
Ariakenom has joined #nixos
Guest99706 has quit [Ping timeout: 258 seconds]
Guest99706 has joined #nixos
<etu>
bsima1_: journalctl -u nixos-upgrade
noam__ has joined #nixos
<bsima1_>
etu: where are those logs being written to?
<etu>
bsima1_: The systemd journal
<bsima1_>
i mean can i read them just from the filesystem? i'm in a live image right now
<clever>
bsima1_: one second
<clever>
--root=ROOT
<clever>
Takes a directory path as an argument. If specified, journalctl will operate on journal directories and catalog file hierarchy underneath the specified directory instead of the root directory (e.g. --update-catalog will create
<clever>
so: journalctl --root=/mnt/ -u nixos-upgrade
<jackdk_>
Hi, I can't get reflex-0.6.1 to build using because it's depending on two different versions of haskell-src-exts ( https://pastebin.com/BnHagELV ). Looks like the culprit might be from haskell-src-meta. Have I diagnosed this correctly, and how can I fix it?
kreisys has quit [Client Quit]
schjetne has joined #nixos
<bsima1_>
xorAxAx: ah i was thrown off by the systemd message
<jomik>
Can someone help me fix a package? It fails due to not finding some libaries that my linker needs, I just have no clue what.. I think... The error is `*-binutils-2.31.1/bin/ld: cannot find -lm`. The expression is here : https://git.io/fjWaf
<jackdk_>
Am I misreading this? package reflex (reflex-0.6.1) requires haskell-src-exts-1.20.3-2fryh11Yuz7lCtLeeIQn8; package haskell-src-exts-util (haskell-src-exts-util-0.2.5-2B7aQaMMTPTKJAXL1gl4C8) requires haskell-src-exts-1.21.0-1MqMUIJU7y29AVS9NHxlR4
kreisys has joined #nixos
Tritlo has quit [Ping timeout: 248 seconds]
sppky has quit [Ping timeout: 248 seconds]
marcinkuzminski has quit [Ping timeout: 248 seconds]
noam has joined #nixos
marcinkuzminski has joined #nixos
blackriversoftwa has quit [Ping timeout: 252 seconds]
oharvey has quit [Ping timeout: 248 seconds]
jw358 has quit [Ping timeout: 248 seconds]
Neo-- has joined #nixos
samrose has quit [Ping timeout: 248 seconds]
aveltras has quit [Ping timeout: 252 seconds]
dmj` has quit [Ping timeout: 252 seconds]
siel has quit [Ping timeout: 252 seconds]
samrose has joined #nixos
bgamari has quit [Ping timeout: 252 seconds]
englishm has quit [Ping timeout: 252 seconds]
grenade has quit [Ping timeout: 252 seconds]
duoi has quit [Ping timeout: 252 seconds]
{^_^} has quit [Ping timeout: 252 seconds]
dukedave has quit [Ping timeout: 252 seconds]
Elon_Satoshi has quit [Ping timeout: 252 seconds]
kini has quit [Ping timeout: 248 seconds]
Tritlo has joined #nixos
Musk_Nakamoto has joined #nixos
bgamari has joined #nixos
noam__ has quit [Ping timeout: 268 seconds]
aveltras has joined #nixos
jw358 has joined #nixos
blackriversoftwa has joined #nixos
dycan has quit [Remote host closed the connection]
siel has joined #nixos
<__monty__>
jackdk_: Oh, you're right. Didn't read past the first two. Have you tried jailbreaking hlint and haskell-src-exts-util?
oharvey has joined #nixos
englishm has joined #nixos
MilkmanDan has joined #nixos
<MilkmanDan>
NixOS still uses systemd and the Suckless respin project went essentially nowhere, correct?
dukedave has joined #nixos
<gchristensen>
NixOS uses systemd, and I think the people who started SLNOS use SLNOS still, but I don't know where it is on the internet
rauno has quit [Ping timeout: 245 seconds]
englishm has quit [Excess Flood]
<MilkmanDan>
I guess for the purposes of this discussion, "hidden somewhere on the Internet" is functionally equivalent to "nowhere". :)
englishm has joined #nixos
<MilkmanDan>
And I also guess I should have said "only uses systemd", since I suppose it's possible to support others. But I didn't find mention of any, so....
kini has joined #nixos
<gchristensen>
NixOS and all its modules has systemd "baked in"
<gchristensen>
but of course you could use Nix, just like NixOS does, to create a Linux which doesn't use systemd
<qyliss^work>
I also have a non-systemd NixOS variant running on a seldom-used computer
<qyliss^work>
(nothing to do with SLNOS)
grenade has joined #nixos
<qyliss^work>
it's not _hugely_ difficult to modify NixOS to use a different init, but you have to port all the services.
<xorAxAx>
cant you statically generate init scripts from the unit attrsets?
iyzsong has joined #nixos
<xorAxAx>
couldnt*
<gchristensen>
"couldn't you just implement all the semantics of systemd with bash?"
<lewo>
(such as generated configuration files for instance)
<qyliss^work>
lewo: is that your project?
<jackdk_>
__monty__: Hm, jailbreaking everything does not work. Looks like reflex has a dep hidden behind a flag, so that might be why jailbreaking isn't doing much. Perhaps the thing to do is chase reflex devs for a metadata bump
<jackdk_>
(I tried with jailbreaks on: reflex haskell-src-exts-util hlint haskell-src-meta)
vidbina has joined #nixos
<lewo>
qyliss^work: more or less
<MilkmanDan>
qyliss^work: I'm sure you're right. I just don't have the time and talent to make that happen myself, although I'd certainly jump on the bandwagon once it gets rolling.
<jomik>
aszlig: Damn. I don't even get this error that we are getting from the firestorm.nix you tossed to me.. :( `*-stdenv-linux/setup: line 1311: ./b2: No such file or directory`
<qyliss^work>
lewo: awesome. I'm going to come back to looking at s6 in NixOS once I get through all the Ruby stuff I'm working on. Might be nice to compare notes at that point...
sppky has joined #nixos
linarcx has quit [Quit: WeeChat 2.4]
bsima1_ has quit [Quit: Page closed]
<apeyroux>
Hello everyone, there is a page that describes how to become a "committer" to help validate the pr on gh?
jtojnar has quit [Quit: jtojnar]
<__monty__>
jackdk_: Yes, unless you can figure out a way to pass --allow-newer to the cabal invocation for reflex.
<lewo>
qyliss^work: yep, of course!
knupfer has quit [Remote host closed the connection]
knupfer has joined #nixos
<MilkmanDan>
Sounds like I should check back in a few months, then. There's hope. :)
<MilkmanDan>
Cheers folks.
MilkmanDan has left #nixos [#nixos]
<gchristensen>
over/under on major progress in a few mo? :P
<qyliss^work>
I was pleasantly surprised by how easy it was to get a proof of concept, tbh
<gchristensen>
yeah
<qyliss^work>
For me, I'd say it mostly comes down to how much free time I have, and whether things I judge to be more important like the Ruby stuff end up coming up
<qyliss^work>
Fix the existing code I feel responsible for before adding more ;)
<gchristensen>
I used s6 once for an "embedded" VM
dmj` has joined #nixos
<qyliss^work>
I really like it
<gchristensen>
and another time I used systemd for an "embedded" VM and was struck by how it was a smaller disk image and booted faster
<qyliss^work>
oh wow, really?
<gchristensen>
yep.
<qyliss^work>
interesting
<sphalerite>
s6 was smaller and booted faster?
<sphalerite>
Or systemd?
<gchristensen>
systemd
<sphalerite>
huh.
<qyliss^work>
s6-linux-init has been rewritten and released 1.0 in the past week or so
<qyliss^work>
wonder if it's faster now
<gchristensen>
"booting quickly" is systemd's raison d'être
<gchristensen>
it is why it does the fancy socket API stuff, and has a huge tree mapping dependencies
<qyliss^work>
stage 1 is a binary rather than an execline script now
<gchristensen>
and, systemd was a dependency of the image anyway -- so it was going to be in the image whether or not I used it
<PyroLagus>
systemd is a software suite and not one huge binary like many people seem to believe
<jackdk_>
__monty__: OK, thanks. That's tomorrow's problem, I think.
<gchristensen>
anyway, iirc I was able to go from off to userspace shell in 4.5s-5s with systemd and qemu
<PyroLagus>
sure, but that's not a problem. that just means there's many pieces of software that are well integrated
<PyroLagus>
as opposed to many that are glued together with sh and regex
<PyroLagus>
the unix philosophy is nice in theory, but in practice it hasn't been done too well, i think
bsima1_ has joined #nixos
civodul has joined #nixos
jomik has quit [Quit: WeeChat 2.4]
<pie__>
does anyone here have any idea how to write gui tests for eclipse?
<pie__>
im trying to have an integration test for my upstreaming of ghidra and i am utterly lost in how to do this, so im going to give up on it till i can find some help
<pie__>
(anyone know anyone that does java development on nix(os)?)
<gchristensen>
sounds unpleasant
<samrose>
pie__: how would you run these tests?
<samrose>
pie__: what testing framework are you using specifically?
<pie__>
samrose, right now i just have a .nix file that i run with nix-shell something-00.nix and manually check if the console output / gui looks sane
<pie__>
samrose, really i just want some automation for starters
<pie__>
make sense?
<pie__>
im not actually using any specific java testing framework because i barely know java
<pie__>
and im integration testing *other people's* stuff
<pie__>
gchristensen, its at least mostly unpleasant because i really dont give a damn about it :( but i care about the end result being good
<samrose>
pie__: I am not sure how deep you are into it already, but if you are using nix-shell, it's usually good from my experience to just build up the pinned dependencies that you need, and then run the existing tests/scripts/etc from the source code you have inherited
<pie__>
and working tests are good for my code
<samrose>
if you want automation beyond the just running nix-shell, I'd look at either running it in circleci with the nix/nixos project docker container, or in soemthing like jenkins.
<pie__>
i dont see any tests :/
<samrose>
pie__: am I stating the obvious here? if so, what more do you need to know
jtojnar has joined #nixos
<samrose>
pie__: aha I get it
<pie__>
samrose, what youre saying is a level beyond what im doing i think
<pie__>
samrose, youre talking about orchestrating the whole thign
<samrose>
pie__: you are looking for maybe some eclipse gui testing tools
<pie__>
samrose, im talking about getting ecl- yeah that
<pie__>
i saw this jubula thing but i have no idea how to use it, let alone on nixos
<jtojnar>
now to find how to fix "Resource limits disable core dumping for process 14775 (qemu-system-x86"
jackdk_ has quit [Ping timeout: 244 seconds]
<samrose>
pie__: if you have target OS for testing, that is a consideration too.
<pie__>
had to run off for a second
<pie__>
basically "all i want to do" is start the debugger and display stopping at a breakpoint
Makaveli7 has quit [Quit: WeeChat 2.4]
<pie__>
samrose, yeah ghidra provides an eclipse plugin that does..some stuff, idk :P makes it easier to attach to a debuggee i guess? IDK what this java stuff normally looks liek
jtojnar has quit [Quit: jtojnar]
<pie__>
it also adds some project types to eclispe
<samrose>
pie__: for a while back in the bad old days, making a UI via eclipse was a good option. Much of the world has moved on, but people in Java land sometimes still do this, and it works great for them :-D
jtojnar has joined #nixos
<clever>
samrose: i found the android ui editor in eclipse to be good 90% of the time
shibboleth has quit [Quit: shibboleth]
<clever>
i even wrote my own custom eclipse plugin to fetch backtraces from my server, display them in the editor, and let me click on any stack frame to jump to that line in the code
<samrose>
clever: I here you, it is a good IDE imho, but maybe not my personal first choice for a general desktop application UI
<clever>
yeah, ive only used it for java based projects
josiah_s_ has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<samrose>
actually I am sure it works really well, but it just becomes a whole other world of stuff to absorb
<samrose>
pie__: LOL this jubula site has everything except the "getting started" info you are looking for!
<__monty__>
How do I make sure an overridden package picks up an overridden dependency without overriding that dependency "globally?"
toppler`` has quit [Ping timeout: 258 seconds]
<clever>
__monty__: call .override on each thing you want to affect
nikivi has joined #nixos
<clever>
samrose: looks like i never put that eclipse plugin on github, its likely in one of my backups of backups of backups, lol
<pie__>
i mean i probably dont even need a proper test, just something to call the eclipse ui functions i need
<pie__>
fancy <clever> i even wrote my own custom eclipse plugin to fetch backtraces from my server, display them in the editor, and let me click on any stack frame to jump to that line in the code
<sphalerite>
I've got a bluetooth headset, and I've managed to connect it just fine, playing audio works, but the microphone doesn't work (show up at all in pavucontrol). Anybody know things about this?
<clever>
sphalerite: there are seperate profiles for music (stereo, but no mic) and phone usage (mono, both directions), the profile also changes the samplerates for the playback channel
<sphalerite>
clever: yeah, there's only a2dp sink and HSP/HFP. The HSP/HFP thing also has awful quality
<sphalerite>
(and doesn't have the mic either)
<clever>
sphalerite: but the HSP/HFP should have a mic
<sphalerite>
nope
<__monty__>
clever: Gonna need a little more explanation. I'm using callcabal2nix to override cabal-install but it needs a newer Cabal. I can't override Cabal with callcabal2nix because the latter depends on the former. So I want to override Cabal but only for cabal-install.
<samrose>
this guy had a video on jubula testing "getting started" too https://www.youtube.com/watch?v=H3DD_5xGHnI seems like it'd be easier to use something that will let you run it headless if at all possible
<clever>
sphalerite: nothing on the input devices tab for both modes?
<sphalerite>
clever: nothing but the laptop's mic and monitors of the output devices
<pie__>
samrose, i probably want a headed thing - or at least the application under test to be headed
<clever>
__monty__: Cabal cant be overridden, because its one of the builtin packages
<pie__>
or to have visual output in some manner (screenshotS? video?)
<clever>
__monty__: any attempt to force it will lead to eratic failures, due to 2 Cabals being present
<clever>
sphalerite: thats strange
<pie__>
samrose, anyway, thanks, looks like a starting point at least
<__monty__>
clever: So there's no way for me to install the HEAD version of cabal-install?
<pie__>
samrose, the hardest part for me is probably getting the environment set up
<clever>
__monty__: you would either need to somehow force the version of Cabal earlier in the overlay setup (causing ghc to rebuild), or use angerman's haskell.nix stuff to allow overriding anything that was builtin with the old system
<clever>
__monty__: more into in #haskell.nix
<clever>
info*
<pie__>
samrose, alternatively, if you have too much time on your hands, id be happy to walk you through the ghidra side of things and you can code this up ;D ;PPP
<samrose>
pie__: I don't have that problem I am sorry to say :-D
<pie__>
onoes
<samrose>
pie__: alternatively there may be just some generic "desktop application testing" tools that are open source in a language that you know, that can do the interaction tests you need
<pie__>
i might even be able to code up some garbage with xephy and xdotool
<pie__>
samrose, did you know that or do you have some search queries
<samrose>
pie__: I used to work in a business that tested things like this, after we coded them up, and I recall airtest being a toold
<sphalerite>
clever: so no ideas beyond that? :/
<clever>
sphalerite: let me fetch some of my BT devices and see what happens
<pie__>
sphalerite, maybe look at what hardware lists see? sorry not very helpful
MmeQuignon has joined #nixos
<sphalerite>
pie__: what do you mean by hardware lists?
{^_^} has joined #nixos
<clever>
sphalerite: i ran blueman-manager, connected it as a hands free device, and it instantly appeared in pavucontrol, in the HSP/PFP profile, and is an input device
<{^_^}>
[nixos-org-configurations] @edolstra pushed 5 commits to master: https://git.io/fjWwS
<pie__>
but maybe thats too much effort for little gain
i1nfusion has quit [Remote host closed the connection]
<sphalerite>
pie__: there's no USB involved
i1nfusion has joined #nixos
Ralith_ has joined #nixos
<pie__>
samrose, ok colleage convinced me that maybe i should look at rcptt anyway
cgm[m] has joined #nixos
noam_ has joined #nixos
<sphalerite>
clever: hm, it works with my android phone
anks has joined #nixos
<clever>
sphalerite: ive also noticed fun things a laptop can do to bluetooth on phones
<python476>
goibhniu1: good point about maintenance, immutable/rollbackable might be a breeze
<clever>
sphalerite: the laptop can connect, claiming to be a headset, and then it routes all audio into pulse!
<sphalerite>
clever: yeah, that's not what I want to achieve though :)
<sphalerite>
(though I will want to do that eventually for other purposes)
<anks>
I'm trying to start with development of packages with Nix. But seriously. I need to be able to go to definition of some random functions defined who-knows-where, such as "project" for example. How to do it?
<sphalerite>
clever: in fact, with HSP/HFP it doesn't even appear as an output device :D
josiah_sama has joined #nixos
<clever>
sphalerite: weird, are you using pulse full?
<sphalerite>
clever: yep
<sphalerite>
iirc think non-full pulseaudio doesn't have bluetooth support at all
<clever>
sphalerite: journal?
<sphalerite>
clever: mentions it being a keyboard (presumably for the volume and playback control keys), but nothing else
MmeQuignon has quit [Quit: WeeChat 2.4]
<clever>
sphalerite: bluetooth is still a bit of a mystery to me, from what ive seen, the dongle in the computer needs to support every profile you want to use
petrkr has quit [Remote host closed the connection]
hoijui has quit [Remote host closed the connection]
hoijui has joined #nixos
<pie__>
samrose, rcptt seems to be a bit easier but now im stuck adding a application to test :P it asks for a location and doesnt accept anything i can think of giving it
josiah_sama has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
Guest30029 is now known as sphalerit
sphalerit is now known as Guest51647
<samrose>
pie__: check these guys out they can likely help you from here as we are getting offtopic for nixos https://wiki.eclipse.org/IRC
<pie__>
samrose, fair enough, i kind of tried that, but maybe someone will know about rcptt
josiah_sama has joined #nixos
<samrose>
pie__: I've never used rcptt so I can't help sorry to say
<Izorkin>
yorick: with --with-mypkgs=${src} used source of current pakages, need use source others pakages
sigmundv has quit [Ping timeout: 246 seconds]
fendor has quit [Ping timeout: 258 seconds]
<gchristensen>
Izorkin: what did ${mypkgs.src} do?
<{^_^}>
[nixpkgs] @bjornfor merged pull request #61393 → openscad: Darwin support and tidy → https://git.io/fjWZX
<{^_^}>
[nixpkgs] @bjornfor pushed commit from @thefloweringash to master « openscad: Darwin support and tidy »: https://git.io/fjWo3
<Izorkin>
with ${mypkgs.src} used archive /nix/store/nhdr6z78j7zxbp588wb2nafsk8cdyq5a-mariadb-10.2.17.tar.gz
lucus16 has quit [Ping timeout: 252 seconds]
<Synthetica>
Izorkin: You could always explictly unpack that, right?
delroth has quit [Ping timeout: 245 seconds]
mexisme has joined #nixos
Lisanna has joined #nixos
<Izorkin>
Synthetica: used sanbox mode, gives an error message - not found
<sphalerite>
Izorkin: you may also be able to use the srcs attribute rather than src
<sphalerite>
Izorkin: I'm not sure how exactly unpackPhase behaves with srcs, and it doesn't seem to be documented in the nixpkgs manual, but if you use srcs = [ (fetchurl foo) mariadb.src ]; it will probably unpack both
<gchristensen>
ah, yeah, that one is a bit annoying since it combines the documents, then pipes, then validates
<gchristensen>
one sec
<Denommus>
lopsided98: thanks
<LnL>
gchristensen: oh, I removed some links and it's gone
<gchristensen>
did you close your xlink?
<gchristensen>
xref*
pshevtsov1 has joined #nixos
nschoe has quit [Ping timeout: 258 seconds]
lejonet has quit [Ping timeout: 258 seconds]
mupf has quit [Quit: WeeChat 2.3]
___laika has quit [Ping timeout: 246 seconds]
<LnL>
dunno what I did, builds fine now
knupfer has joined #nixos
<gchristensen>
I kept hurting myself on <xref linkend="foo"> vs. <xref linkend="foo" />
<lemonpie>
what's the 'best practice' split on installing things via home-manager's package system vs your user's nix-env ? nix-env seems to be far more convienient in the cases where you might install/uninstall frequently-ish, and home-manager for things that are definitely 'part of your everyday workflow', is this on the right track ?
mupf has joined #nixos
<LnL>
I'm using link
<pshevtsov1>
Hi! I use nix on Debian. I've just noticed that mouse cursors over applications installed from nixpkgs (e.g. nix-env -iA nixpkgs.rxvt_unicode) are different then over the apps from apt repos. That's just an aesthetical glitch but I'm really wondering how I could make X apps from nixpkgs use mouse cursors from the base system?
<cizra>
Heya, I'm a semi-newbie. How do I install gnome-keyring using environment.systemPackages in nixos/configuration.nix? Whatever I write, it claims gnome-keyring doesn't exist -- yet nix search and nix-env -qaP show it.
<gchristensen>
cizra: instead of installing the package, add services.gnome3.gnome-keyring.enable = true to your config
<cizra>
gchristensen: Thanks! For the next time, how do I know I have to do this?
<cizra>
Also, so far so good - I've gotten most everything working that I need.
pshevtsov has left #nixos [#nixos]
pshevtsov1 has left #nixos ["WeeChat 2.4"]
pshevtsov has joined #nixos
<{^_^}>
[nixpkgs] @fgaz opened pull request #61464 → sfxr: init at 1.2.1 → https://git.io/fjW1f
josiah_sama has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
josiah_sama has joined #nixos
fusion809 has joined #nixos
acarrico has quit [Ping timeout: 255 seconds]
<sphalerite>
lemonpie: sounds right, although for one-off things you can also use nix-shell
srid has joined #nixos
lejonet has joined #nixos
zeta has quit [Quit: WeeChat 2.4]
kisonecat has joined #nixos
i1nfusion has quit [Remote host closed the connection]
i1nfusion has joined #nixos
<{^_^}>
[nixpkgs] @risicle closed pull request #60553 → pythonPackages.asyncssh: 1.15.1 -> 1.16.1, fixing build with some caveats around ec25519 support → https://git.io/fjZOD
freeman42x has joined #nixos
<cizra>
Is there a way to avoid conflicts spam when nixos-rebuild switching? Namely, kill is provided by procps and coreutils, neither which I installed explicitly.
<gchristensen>
cizra: there is a fix for that to be released
i1nfusion has quit [Remote host closed the connection]
i1nfusion has joined #nixos
<cizra>
gchristensen: Thanks!
thomashoneyman has joined #nixos
josiah_sama has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<__monty__>
pshevtsov: I *think* that's because the nix-installed stuff isn't picking up on the theme your DE uses. If you can find out which one that is I'm sure there's a way to get the nix software to recognize it.
<astal>
it's very strange, coz i have 25G free memory in my root partition. but thx you, i'll this way, may be nix need free space in another parts
<{^_^}>
[nixpkgs] @c0bw3b merged pull request #61251 → py-radix: init at version 0.10.0 → https://git.io/fjC1e
<{^_^}>
[nixpkgs] @c0bw3b pushed commit from @mkgvt to master « py-radix: init at version 0.10.0 (#61251) »: https://git.io/fjWyx
rfold has joined #nixos
<rfold>
Is it possible to prevent callPackage from adding override and overrideDerivation keys when the package returns not a derivation but an attrset?
<symphorien>
rfold: use callPackages (plural) instead
<Thra11>
With a 'complex' output path such as /nix/store/...-armv7l-unknown-linux-gnueabihf-stage-final-gcc-7.4.0-lib, is it possible to refer to that path in a derivation (Along the lines of ${package.someoutput}?
<symphorien>
> pkgsCross.raspberryPi.gcc.lib
<{^_^}>
attribute 'lib' missing, at (string):253:1
<Thra11>
`nix eval nixpkgs.pkgsCross.armv7l-hf-multiplatform.gcc.cc.lib.outPath` gives "/nix/store/0ppzx6wgrq0vxamq7iprq6r6ff6zd52d-gcc-7.4.0-armv7l-unknown-linux-gnueabihf-lib", not the path I want, "/nix/store/28nb6zzh6mjj3706s0w4jyvvh5hb4dnb-armv7l-unknown-linux-gnueabihf-stage-final-gcc-7.4.0-lib"
Izorkin has quit [Ping timeout: 255 seconds]
Izorkin has joined #nixos
mexisme has joined #nixos
<rfold>
symphorien: this expression returns an attrset that has override and overrideDerivation, even though it uses callPackages: (callPackages ({callPackages}: {images = callPackages ({}: {}) {};}) {}).images
azdle has quit [Read error: Connection reset by peer]
<matthewbauer>
Thra11: You need to use buildPackage I think. So `nix eval nixpkgs.pkgsCross.armv7l-hf-multiplatform.buildPackages.gcc.cc.lib.outPath` or alternatively `nix eval nixpkgs.pkgsCross.armv7l-hf-multiplatform.stdenv.cc.cc.lib.outPath`
myskran has joined #nixos
shibboleth has quit [Quit: shibboleth]
abathur has quit [Ping timeout: 258 seconds]
i1nfusion has quit [Remote host closed the connection]
<rfold>
symphorien: from the source code it seems that callPackages adds override and overrideDerivation to each value of the returned attrset
i1nfusion has joined #nixos
<rfold>
and there doesn't seem to be a function that only does the dependency injection, so I'll write my own.
<lostman>
@sphalerite that's the wrong way around. I want a derivation that I can build from my local dev directory which doesn't package any random stuff I have there. Would have to add everything to .gitignore. So I just want to grab everything tracked by git
<srhb>
It's a bit of a mess right now. Please do get involved :)
<q3k>
lovely, will see what I can do
mkoenig has quit [Remote host closed the connection]
<sphalerite>
lostman: I think you're misunderstanding the purpose of that project :)
<sphalerite>
lostman: see the "motivation" section of the readme
freeman42x has quit [Ping timeout: 258 seconds]
<srhb>
fwiw my "fluffy" opinion is that we should decouple more, despite the feeling that more explicit systemd dependencies feels and looks nice in tests.
<srhb>
q3k: eg. do what k8s does, let everything eventually connect and run.
<q3k>
you mean remove the kube-control-plane-online 'barrier'?
<srhb>
Accounting for all cases of certs/no certs, local/remote etc. is really difficult and complex, something I think we should not try to encode in units
<srhb>
q3k: Essentially yes.
<q3k>
yeah, i agree
<q3k>
k8s should absolutely handle such 'unsychronized' start procedures
<srhb>
Indeed.
jtojnar has quit [Remote host closed the connection]
<q3k>
it might take time to settle as all convoluted microservices do
<srhb>
It's theoretically simple to go back, we were more decoupled earlier.
<q3k>
but if it doesn't work at all that's a k8s bug
<srhb>
So it's a biased cherry-pick revert.
jtojnar has joined #nixos
<q3k>
if anything, i think the users should be able to enable/disable/configure this barrier
<q3k>
eg. default to with-barrier, and allow to fully uncouple of override with your own health check (for atypical setups)
<q3k>
not sure :)
<lostman>
@sphalerite I don't think I'm misunderstanding the purpose. I would just like to achieve similar thing using a different mechanism. If I temporarily create a file foobar in my dev directory then I don't want it to be included in `src` of the derivation and I don't want to add it to `.gitignore` either. OTOH files, even those that I'm currently working on, are tracked by git. Or I can always `git add` if I want it to be included
<srhb>
I would much rather they do it manually then. The k8s infrastructure is too complex right now. At its core, I believe we should have a very simple interface for each component that essentially just exposes a neat way to generically specify kubeconfig and command line flags.
zeta has joined #nixos
Izorkin has quit [Ping timeout: 245 seconds]
<q3k>
i feel like there should be two layers to this
<srhb>
q3k: From there, we can overlay a more complex ezmode k8s module if need be (even out of tree) -- but this would make changes much more easy to do
halfbit has quit [Ping timeout: 252 seconds]
<q3k>
one is the low-level k8s service configuration
<q3k>
and yeah, overlay eznixk8s on top of that
<q3k>
(that's a catchy name)
<srhb>
Yes, exactly.
<srhb>
It is :-)
zeta has quit [Client Quit]
<pie__>
q3k, oh huh. hi
<srhb>
q3k: At least johanot and I have discussed this a lot. We should pool our efforts and do the rewrite, but this issue should probably be fixed ad-hoc right now
<q3k>
pie__: hiya
<srhb>
q3k: That is, by removing our synchronizing efforts.
<q3k>
srhb: agreed
<pie__>
q3k, how long have you been in here :V
<q3k>
pie__: i just joined
<q3k>
pie__: but i've idled here before
<q3k>
pie__: mostly after bugging various people to look at my PRs :P
Izorkin has joined #nixos
<srhb>
Anyway, I'll poke johanot to look at the logs from this discussion (since he doesn't have an irc bouncer) and head to bed now :-P
mexisme has quit [Ping timeout: 246 seconds]
<q3k>
sure, g'nite
<srhb>
o/
<q3k>
i'll write a quick paragraph of my opinions on that second thread
jtojnar has quit [Quit: jtojnar]
<srhb>
Great!
lucus16 has joined #nixos
<lucus16>
Which texlive package provides \usepackage{tikz}?
o1lo01ol1o has quit [Remote host closed the connection]
Izorkin has quit [Ping timeout: 246 seconds]
o1lo01ol1o has joined #nixos
Izorkin has joined #nixos
linarcx has quit [Quit: WeeChat 2.4]
o1lo01ol1o has quit [Ping timeout: 246 seconds]
<sphalerite>
lostman: ooooh I see.
lovesegfault has quit [Ping timeout: 258 seconds]
<sphalerite>
lostman: builtins.fetchGit does almost that, but I think it'll take HEAD, not the current state of the working tree
<lostman>
@sphalerite this seems to do what I want: `src = builtins.fetchGit { url = builtins.toPath ./.; };`
<lostman>
@sphalerite just tested it and it looks like it takes the state of the working tree.