<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fhLQ2
<ldlework>
gchristensen: this should totally be a thing, shouldn't it
<samueldr>
step 1. write a script step 2. exec the script no manual intervention :) (and hopefully the script amounts to partition/format/mount/nixos-install
<ldlework>
samueldr: turnkey nix-challenge is god-tier
<gchristensen>
I followed my blog post on it which was close to TAS, but I replaced the partitions with ZFS, so some exploration
<gchristensen>
taking notes ...
<ldlework>
it is understandable if there are some things the person needs to do to get the overall process finished - not setting up tools, but just performing the reinstall and stuff
<ldlework>
turn-key though...
thc202 has quit [Quit: thc202]
<ldlework>
like the ability to boot into a usb key that does the whole thing without interaction
<samueldr>
I think I might do this when I finally migrate my debian-running server to nixos (legacy!)
<ldlework>
that'd be basically as good as it gets I think
<mog>
fixed-output derivation produced path '/nix/store/bmh2s8hlc3jmllajarl1f9f3y38mvc9x-0a3755c1799d3a4dc1875d4c59c7c568a64c8456.patch' with sha256 hash '1flnxgjjrrlzdk5klx47l8y8rg1vmzdd9vhzxxq6wiwyvd4bd92r' instead of the expected hash '0bizaf2yf93hwkrrjcl3fhawyhmw9dzq9pc283dxmmpxqvvif5xg'
<mog>
its fixed in github at least as its moved on to newer version
<mog>
but whats out is busted
equivrel has joined #nixos
<samueldr>
:/ 0a8f35c998f55f744674ed23b327162df39fb6ae here built and I'm using network-manager
<samueldr>
oh wait I was a bit too eager I thought it had finished building
* samueldr
waits again
<mog>
ya it builds a lot and then dies for me
fusion809 has joined #nixos
<jasongrossman>
That's funny. I built 18.09 a few hours ago.
<samueldr>
I say unlikely as gitlab probably shouldn't allow changing a commit *that* way
bpye has left #nixos ["Kicked by @appservice-irc:matrix.org : issued !quit command"]
<mog>
ill try again
bpye has joined #nixos
<mog>
i fetched the file by hand and am trying now
<mog>
same error i believe fixed-output derivation produced path '/nix/store/bmh2s8hlc3jmllajarl1f9f3y38mvc9x-0a3755c1799d3a4dc1875d4c59c7c568a64c8456.patch' with sha256 hash '1flnxgjjrrlzdk5klx47l8y8rg1vmzdd9vhzxxq6wiwyvd4bd92r' instead of the expected hash '0bizaf2yf93hwkrrjcl3fhawyhmw9dzq9pc283dxmmpxqvvif5xg'
orivej has quit [Ping timeout: 250 seconds]
bpye has left #nixos [#nixos]
bpye has joined #nixos
moriacat has joined #nixos
LysergicDreams has quit [Ping timeout: 250 seconds]
bpye has left #nixos ["User left"]
bpye has joined #nixos
<samueldr>
okay, now it finished building
<samueldr>
mog: how did you "fetch the file by hand"? fetchpatch does some dark wizardry to patch files to make them more stable
<samueldr>
so simply prefetching the URL isn't enough
<dtz>
errr or join "#Nixos:matrix.org" probably ;)
<iqubic>
What even is matrix?
tzemanovic has joined #nixos
infinisil[m] has joined #nixos
<infinisil[m]>
Hello from Fractal!
<dtz>
\o/
lawlesseel has quit [Ping timeout: 256 seconds]
tzemanov_ has joined #nixos
<infinisil[m]>
iqubic Use google to find out..
<dtz>
hope didn't make it sound too great, but I've been dying for a remotely usable matrix client on linux... fractal delivers! At least as of 4.0.0 I can use it for my various communications
<infinisil[m]>
It is a bit wobbly for me, but definitely usable
<infinisil[m]>
Like, the room directory thing is slow and bit buggy, no progress indication
<infinisil[m]>
Sending messages makes the view shift twice (once delayed)
<dtz>
oh, haha yeah. first sync for me was.... both slower than expected and a bit.. surprising in terms of "flow"
<iqubic>
Is it an upgrade over standard IRC?
<dtz>
oh no! I had that earlier but it stopped happening so I hoped they fixed it
<dtz>
that was... basically deal-breaker for me re:scrolling the wrong way--meant any conversation would lose sight of itself lol
tzemanovic has quit [Ping timeout: 252 seconds]
<dtz>
iqubic: no, at least that's not the comparison anyone would suggest.
<dtz>
iqubic: for me, at least, it's the web generation's (:P) replacement for what I used to use pidgin or bitlbee for: unify all the chat silos so I can get stuff done and no one has to care about my client and other way 'round
tzemanov_ has quit [Ping timeout: 252 seconds]
<infinisil[m]>
dtz: Do you also need to re-login everytime you quit the client?
<dtz>
it's also decentralized, which is rather nice. But I'm not sure i'd use it without the various matrix<-->foo bridges-- for example, right now this channel is bridged so we can use it
<dtz>
Infinisil: lol no! D:
<dtz>
to be honest I mostly, maybe shamefully, use it as an irc client of sorts lol >.<
<siraben>
Is it possible to make a nix build file that will use the latest commit of a Git repo?
<siraben>
And how do I get the SHA checksum and "rev" fields?
<siraben>
iqubic: I use matrix :)
<dtz>
first even looked into it since newer folks on my team didn't seem to find our IRC channel a natural place/method to communicate.... well anyway. I'm probably not the best advocate. Check out matrix.org and google it. ^_^.
<siraben>
Matrix has very good bridges, and there's also mobile apps
<siraben>
I think I've had one or two occasions where I was disconnected from all my IRC channels
<siraben>
In the past year, though.
<iqubic>
I see.
<samueldr>
siraben: rev, any commit-ish valid to git
<samueldr>
so putting master in there would seem to work...
<siraben>
iqubic, what's your current setup for staying connected to IRC?
<samueldr>
... until you realise the sha256sum will cause it to not re-download
<samueldr>
and siraben, as for getting sha256, it's always wiht a fresh use of TOFU
<samueldr>
,tofu
<{^_^}>
To get a sha256 hash of a new source, you can use the Trust On First Use model: use probably-wrong hash (for example: 0000000000000000000000000000000000000000000000000000) then replace it with the correct hash Nix expected.
<samueldr>
(not the only way, but one of the most convenient and universal)
<Church_>
samueldr: I'm gonna need to edit this post on packaging rails apps
<siraben>
clever: Btw, know of a GUI testing tool? that TI calculator emulator you made the Nix build for a while back doesn't have a CLI and it reports the values of registers on a GUI
<Church_>
It already looks blegh.
<siraben>
trying to make it so that I can run tests on my ROM after compiling it
<clever>
siraben: edit it to include CLI args?
<siraben>
clever: What do you mean?
<siraben>
Doesn't have a CLI display avaliable it seems
<clever>
siraben: maybe add a custom IO port, so you can write to a given addr in the emu, and it then immediately exits with the value you wrote?
<clever>
so the rom can test itself, and then return a status code
<siraben>
iqubic: You a fan of urxvt, by any chance?
<siraben>
Trying to switch over from terminator but want to use a declarative config
<siraben>
clever: Hm. let me see
<iqubic>
I have never used urxvt.
<siraben>
What terminal emulator do you use then?
<iqubic>
I'm still not sure why I'd want to matrix.
<iqubic>
I use termite.
<siraben>
it's mostly the always-on nature of it
<siraben>
Also cross platform, no issues with keeping track of log files etc.
<siraben>
Also means you're more reliant on it not failing, so it's a tradeoff.
<clever>
siraben: my main exposure to matrix, is when the servers fail, and 200 people mass-quit in irc
<clever>
which floods the channel and makes it unusable :P
<iqubic>
But how hard is it to set up?
MinceR has quit [Ping timeout: 240 seconds]
<siraben>
Not at all
<siraben>
you just make an account, activate the bridge
<siraben>
Join your desired channels and it's done.
<siraben>
I didn't use the Emacs Matrix Client, note.
<siraben>
I currently use Riot
ambro718 has quit [Ping timeout: 244 seconds]
<iqubic>
Where do I go to make an account?
<siraben>
clever: that was my initial exposure as well, haha
<iqubic>
I'm not sure if I want to use riot or not.
<iqubic>
"User ID already taken."???
<iqubic>
I can't make an account with the username `iqubic`
<siraben>
huh
<siraben>
Well you can change it to whatever
<siraben>
To IRC users you'll appear the same as normal
<siraben>
Because you'll be bridged
<iqubic>
I see.
<iqubic>
Is it giving me this error because I'm logged in already?
<siraben>
Logged in where?
<siraben>
IRC?
<siraben>
No.
<siraben>
Maybe we should move this off channel to PM or something
<iqubic>
I'm logged in on IRC, yeah.
<infinisil[m]>
#nixos-chat for offtopic discussions
<siraben>
clever: Hm. Doesn't seem to have CLI args that allow me to print the state or something
<siraben>
Thanks infinisil
<clever>
siraben: you would need to modify the emulator to add them
<siraben>
Time to crack open the soucre!
<siraben>
source*
<siraben>
iqubic: move over to #nixos-chat
<wedens>
for some reason I have links for "light" package in both `/run/wrappers` and `/nix/store`. light is enabled via `programs.light.enable = true`. how do I reference a wrapped binary and not binary from the store?
<wedens>
when I use `${light}/bin/light` from my home-manager configuration it referencec unwrapped binary that doesn't work
<samueldr>
/run/wrappers are (only?) setuid wrappers; as the store cannot hold setuid binaries
<samueldr>
so this is why it exists
* samueldr
checks for a solution to your problem now
<samueldr>
it seems the nixos stuff mostly assumes it can refer to /run/wrappers
<wedens>
so, the problem is referencing nixos binary from user configuration?
<samueldr>
even some non-nixos things e.g. `at` from nixpkgs defaults to sendmailPath being in `/run/wrappers`
<samueldr>
it's more that setuid wrappers are built outside of the nix build building your system
<wedens>
should I just use binary by name without full path?
<samueldr>
I don't know for sure, it must depend on what you're doing
<wedens>
I'm binding `light` program to a key in i3 configuration
<wedens>
not using full path should be safe enough here, I guess
<samueldr>
I personally would either use an unqualified name and rely on PATH or use /run/wrappers/bin/light directly
<samueldr>
yeah
<wedens>
the later won't work. it's /run/wrappers/wrappers.9YBUGK2B4Y/light and it changes after switching configuration
<samueldr>
I'm talking about the /run/wrappers/bin symlink which will resolve to the right wrappers dir
<samueldr>
if it doesn't you're in hot water as it's where sudo lives :)
<wedens>
ah, there is /run/wrappers/bin indeed
<samueldr>
it's also part of your PATH
<wedens>
I was just doing `realpath $(where light)` :)
<samueldr>
ah right
<samueldr>
if you don't want to rely on PATH, and feel icky at hardcoding /run/wrappers/bin/${tool} you could always contain the "mess" of it in a function, (tool: "/run/wrappers/bin/${tool}")
<samueldr>
so if in the future that implementation detail changes you have one location to fix
<wedens>
I'll go the easy road :) thanks for help
<wedens>
it's just a bit confusing that you don't know when it's better to reference the store and when it won't work
<samueldr>
I can see, at one point it all makes sense and... you become desensitized to the issue :/
init_6 has joined #nixos
iqubic has left #nixos ["ERC (IRC client for Emacs 26.1)"]
paulvictor has joined #nixos
Dedalo has joined #nixos
<Dedalo>
How to I add a user to the list of sudoers in NixOS?
<Dedalo>
I want to be able to modify configuration.nix using the user that I have created myself, because `root` doesn't have a Window Manager configured and I can't do copy and paste, like I do in i3.
<makefu>
Dedalo: users.groups.wheel.members = [ <your-username> ]; or users.users.<username>.extraGroups = [ "wheel" ]; ( out of my head, untested )
<clever>
sudoedit will copy the file to /tmp, run the configured editor, as non-root, then copy it back when the editor exits
<Dedalo>
makefu I have tried the second solution `users.users` but it doesn't work. I will try the first one. Do I just need to perform a `nixos-rebuild switch`, or do I need to reboot?
<makefu>
Dedalo: you need to log in and log out of your session
<makefu>
check "groups"
<Dedalo>
makefu thx, I try
<clever>
`id` also works, and shows more info then `groups
<clever>
`
<Dedalo>
makefu thanks, it worked
<{^_^}>
[nixpkgs] @worldofpeace pushed 2 commits to master: https://git.io/fhLNv
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
brodul has quit [Ping timeout: 240 seconds]
pepesza has quit [Ping timeout: 244 seconds]
pepesza has joined #nixos
kisik21 has quit [Quit: WeeChat 2.4-dev]
eyJhb has joined #nixos
<eyJhb>
Hey, I am currently trying to mount a external SSD on boot which is encrypted using luks. But I keep getting "failed to open key file" at each boot, at the service will eventually timeout...
<{^_^}>
#53067 (by Izorkin, 1 day ago, open): haskel: error build package insert-ordered-containers and uri-bytestring
Notkea has quit [Read error: Connection reset by peer]
<ottidmes>
I have a question about how to best organize the local sources uses by my configurations. The way I have currently organized my configurations requires sources to exists at locations outside of my configuration. With the consequence that you cannot just take my configuration and install it, it would fail due to those missing sources. I like my configuration to be self-contained, so I was wondering how others
<ottidmes>
do this. I saw someone using a vendor directory containing the sources, I like that idea, but how to go about your personal projects that you want to include, it seems weird to have all of it be contained in your NixOS configuration. My current idea is to default to just pointing to their git origin's for the sources, but overwrite them in an overlay if local checkouts exist.
Notkea has joined #nixos
random_yanek has quit [Quit: random_yanek]
jomik has joined #nixos
jomik has quit [Client Quit]
<Twey>
Why doesn't the `pipenv` derivation have `virtualenv` in its propagatedBuildInputs? It seems to be necessary if python3Packages = python36Packages but not if python3Packages = python37Packages, and I can't figure out why
<ottidmes>
Izorkin: seems unrelated to Nix, so you might have more luck on the Haskell channel, but if those packages are not well maintained you might try an older version of GHC or it might actually be fixed in a newer version of ghc. What is the version returned by: nix-instantiate --eval --strict --expr 'with import <nixpkgs> { }; haskellPackages.ghc.version'
<Twey>
ottidmes: You can also use fetchurl and friends
<Twey>
ottidmes: And you can store local changes as patches
<Izorkin>
ottidmes: "8.4.4"
goibhniu has joined #nixos
<ottidmes>
Twey: your second point, what would that look like? would you need to make patches for every change made?
<ottidmes>
Izorkin: then you could try haskell.compiler.ghc822 instead of haskellPackages to use GHC 8.2.2 instead of 8.4.4 for example
periklis has joined #nixos
FRidh has joined #nixos
<ToxicFrog>
What needs to happen to get #51980 merged?
<Twey>
ottidmes: I mean, it's your config, so you could just have one big patch
jomik has joined #nixos
<Twey>
ottidmes: But it might be better just to fork the source repo somewhere and refer to that in the config maybe?
<jomik>
How do people using Nix handle their emacs configuration? Do you install emacs-packages with nix and then just have an init.el? Or do you also handle your init.el with nix?
agander has joined #nixos
<ottidmes>
Twey: yeah, I think I will be doing the fork approach, but I have a script that takes a snapshot of just the changes made in the Git repo, so I will probably also do what you suggest with sort of patches, so that I do not always have to commit my repos while I might not want to at that particular moment
<Izorkin>
ottidmes: how to need add to configuration.nix to replace gcc
<ottidmes>
Izorkin: could you rephrase that, I am not sure what you mean? replace gcc everywhere?
<Izorkin>
replace gcc only haskellPackages
<FRidh>
instead of having a fixed location to a local patch in the tree `patches = [ ./my/patch ]` I would like to dynamically determine the patch `[ patches = [ "./${version}/patch" ]` is a string so not helpful. toPath can't be used either
<ottidmes>
Izorkin: How to replace haskell packages that require gcc? Not sure, if they used gccStdenv explicitly it might be possible by inspecting their function arguments, but I am not familiar enough with the details of how haskell packages are packaged in Nix to determine that
<ottidmes>
FRidh: ./. + "${version}/patch" might work
<ottidmes>
FRidh: ehm this: ./. + "/${version}/patch" might work
ddellacosta has joined #nixos
<FRidh>
yep indeed thanks
Dedalo has joined #nixos
<Twey>
"${./.}/${version}/patch" too I guess
<ottidmes>
Twey: yes, but you definitely do not want that, because it would copy ./. to the nix store
lawlesseel has quit [Ping timeout: 256 seconds]
ris has joined #nixos
<Twey>
ottidmes: I think that's equivalent
periklis has quit [Remote host closed the connection]
<Twey>
./. + "/foo" will also copy ./. to the store
<ris>
hmmm..... is there any way to get `nix-shell` to obey `meta.priority` in the way that `nix-env` does?
<Myrl-saki>
Twey: The only thing that forces /nix/store is string interpolation, I believe.
<Myrl-saki>
It just so happens that string interpolation is so ez to accident.
<symphorien>
ris: I think the mechanism behind meta.priority is in buildEnv
<ris>
orly
<ris>
interesting
<jomik>
Twey, ottidmes, Myrl-saki - I had this exact issue yesterday, ottidmes' does not copy the path to nix-store, whereas passing just the string, would :D
<jomik>
Just to confirm.
<ris>
oh actually i guess what i want really...
<ottidmes>
Twey: ./. + "/foo" results in just a path value, if you were to then do "${./. + "/foo"}" it would still not be equivalent, because it would copy only foo to the nix store, not the whole of ./. like your first suggestion would
<ris>
is to make one version of `git` appear in the $PATH before the other
<ris>
and that's not quite the same thing as `meta.priority` is it?
<Twey>
ottidmes: Weird, I would have thought that any path literal would be copied to the store (before conversion to a string and concatenation with the string). But I guess indeed path + string produces a path
<ottidmes>
jomik: passing a string would not copy the path to nix store
<symphorien>
ris: a buildEnv with a hiPrio git and a lowPrio git will contain only one $out/bin/git
<jomik>
"${./.}/${version}/patch" I mean? IIUC that is a string, ottidmes ?
<ottidmes>
jomik: the end result of "${./.}/${version}/patch" is indeed a string, compared to ./. + "/${version}/patch", which would end up as a path, but it is "${./.}" that is the interesting bit, like I showed in: "${toString (./. + "/foo")}" a path can always be converted to a string without copying over the path to the nix store
<ris>
symphorien: ah i guess that's an argument for actually using a `buildEnv` rather than just a derivation with a bunch of `buildInputs` which is what i'm currently doing
<ris>
symphorien: perfect answer thanks
<ris>
(assuming it works :)
<{^_^}>
[nixpkgs] @timokau opened pull request #53121 → vimUtils: derive pname from the plugin string → https://git.io/fhtJk
__monty__ has joined #nixos
<Myrl-saki>
clever: Did you check it out yourself while I was gone?
<m15k>
Someone already installed lxd on nixos? I'm a little bit lost. Seems that I can't get the service running.
<countingsort>
whats the difference between nativeBuildInputs an buildInputs?
<countingsort>
just compile time vs run time?
earldouglas has left #nixos [#nixos]
periklis has quit [Ping timeout: 250 seconds]
<Twey>
countingsort: nativeBuildInputs are things available to the build process itself, i.e. the thing doing the build, but not to the built program/in the built environment. buildInputs are things that can be used alongside the build product. The difference is chiefly different when cross-compiling: nativeBuildInputs might be something like a cross-compiler or source transformation tool (that runs on the build
<Twey>
host) while buildInputs can be things like dynamic libraries for the built application
<Twey>
countingsort: Compile-time vs runtime is done automatically in Nix, by looking for build inputs that are referenced in the final build product
<Twey>
countingsort: So that distinction doesn't exist when you're writing derivations
<countingsort>
Twey: i see, thanks
obadz has joined #nixos
<betaboon>
is there a binary cache for headless(xserver-less) configurations ?
<Twey>
Is there a way to see what a ‘dumping very large path’ warning is referring to?
<clever>
first, my SSD's have a firmware bug, that results in the drive going unresponsive randomly during writes
<clever>
second, one day, it failed, immediatelly after a `git commit && git push`
<clever>
after a forced reboot, the commit ceased to exist locally, yet it still existed remotely!
<p_l>
it's very hard to truly corrupt ZFS beyond automatic self-repair
<infinisil>
Haha
<clever>
the git app thought the data was commited to disk
<clever>
yet it undid itself!
<clever>
the corruption likely triggered an automatic rollback, which repaired it perfectly
<p_l>
one time I've seen it it involved running `master` branch of zfsonlinux when a bug got into it (never reached release, as it was caught), and the bug changed the on-disk format
<infinisil>
p_l: I wish I was more motivated to find the cause of that
<p_l>
the fix involved rewinding transactions to one from before the bug :D
<schopp0r>
hm. can somebody please help me: I would like to install *one* unstable package (which I will probably also patch) as a user, but keep the rest stable. how hard is this?
<{^_^}>
#53067 (by Izorkin, 1 day ago, open): haskel: error build package insert-ordered-containers and uri-bytestring
<{^_^}>
[nixpkgs] @ElXreno opened pull request #53126 → megasync: init at 3.7.1.0 → https://git.io/fhttD
tzemanovic has joined #nixos
<wedens>
is there a good example with instructions on how to organize nixos with home-manager and install existing configuration from scratch? let's say I have my nixos config, dotfiles, etc in a git repo and I want to install it on a new system. where do I clone it? should I somehow split system and per-user configuration so that I can install the later after installing the system?
eadwu has quit [Ping timeout: 250 seconds]
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
eadwu has joined #nixos
tzemanovic has quit [Remote host closed the connection]
<schopp0r>
infinisil: can I do the same in my per-user nix configuration?
tzemanovic has joined #nixos
<infinisil>
schopp0r: Yeah no problem
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
<schopp0r>
infinisil: ok I will try later. (currently in the train, I cannot checkout the repo here ^^)
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
<infinisil>
wedens: I personally use home-manager#97 to have NixOS + home-manager combined in a single config
tzemanovic has quit [Remote host closed the connection]
tzemanovic has joined #nixos
<infinisil>
(But it is a bit slow because it always needs to build the full system even if you just want to change a home-manager setting)
eadwu has quit [Read error: Connection reset by peer]
<wedens>
infinisil: so, you have all your configuration in /etc/nixos?
tzemanovic has quit [Remote host closed the connection]
<infinisil>
wedens: I have it all in ~/cfg, and (you could say) that I set `nixos-config=/home/infinisil/cfg` for doing system rebuilds
__monty__ has quit [Quit: leaving]
<infinisil>
You can do it any way you like really, Nix doesn't constrain you in that way
arjen-jonathan has quit [Ping timeout: 264 seconds]
<wedens>
infinisil: but you don't have $HOME when you don't have a system yet ;)
<wedens>
yeah, that's the problem of not being opinionated. too many options :))
<infinisil>
Heh yeah, I guess then it would be somewhere else
arjen-jonathan has joined #nixos
eadwu has joined #nixos
<wedens>
I thought of seeding the system from some accessible location and then moving configuration to ~ and re-linking config files. but i'm too spoiled by nix being declarative :)
<infinisil>
wedens: I'd recommend a wrapper script for whatever you need for the rebuild
<infinisil>
Like I have ~/cfg/deploy/rb for doing a rebuild
<infinisil>
It sets up everything relative to itself
<infinisil>
So it doesn't matter where I have the config (assuming I didn't mess up somewhere)
tzemanovic has joined #nixos
random_yanek has joined #nixos
random_yanek has quit [Max SendQ exceeded]
tzemanov_ has joined #nixos
random_yanek has joined #nixos
<schopp0r>
hm. how can I search for packages containing certain filenames? (I think I can remember there was something like that)
<symphorien>
nix-locate
<infinisil>
,locate
<{^_^}>
Use ,locate <filename> to find packages containing such a file. Powered by nix-index (local installation recommended).
<infinisil>
(nix-index is the package containing the nix-locate binary)
<wedens>
with `nixos-rebuild dry-build` it's possible to see what'll be built, but how to know what caused the build?
m15k has quit [Ping timeout: 256 seconds]
<{^_^}>
[nixpkgs] @timokau merged pull request #53121 → vimUtils: derive pname from the plugin string → https://git.io/fhtJk
<{^_^}>
[nixpkgs] @timokau pushed to master « vimUtils: derive pname from the plugin string (#53121) »: https://git.io/fhtmc
<infinisil>
wedens: Generally that's hard to say, there's no mapping from derivations to nix expressions
<wedens>
a list of inputs that caused changes downstream or something like that
<infinisil>
I recommend using git to track the changes of your config
<wedens>
for example, I did `nix-channel --update` and `nixos-rebuild switch --upgrade` and it caused recompilation of my custom emacs derivation. I'd like to know why it was needed to recompile it
<wedens>
recompiling emacs on every upgrade doesn't sound good :)
agander_ is now known as agander
<gchristensen>
what is custom about your emacs?
<wedens>
specific revision in emacs repo
<gchristensen>
aye
<gchristensen>
it is going to be very common that a channel update causes it to recompile, since emacs has lots of dependencies and it is common for them to change
<wedens>
does it make sense to use pinned nixpkgs for emacs derivation?
<wedens>
or will it duplicate half of my system?
<gchristensen>
how long does the compilation take?
<{^_^}>
#52995 (by 8573, 3 days ago, closed): Pandoc fails to build, citing missing dependency `tasty`
<jonreeve>
But it doesn't seem to be fixed for me
<jonreeve>
(And I'm on unstable)
<jonreeve>
I've tried nix-channel --update, but it still fails because of "tasty"
<jonreeve>
Anyone know a workaround?
<{^_^}>
[nixpkgs] @ikervagyok opened pull request #53127 → [Staging] xorg: updates and cleanup → https://git.io/fhtY0
<ottidmes>
jonreeve: you could use an overlay, you could maybe point nixpkgs to the specific commit it has been fixed (with the change that some things will not have been build yet, so it could trigger some builds)
kisonecat has quit [Ping timeout: 246 seconds]
<ottidmes>
s/change/chance/ (somehow often make that mistake)
kisonecat has joined #nixos
<{^_^}>
[nixos-hardware] @nocent opened pull request #91 → apple: Add support for MacBook 3,X → https://git.io/fhtYi
tilpner has quit [Quit: WeeChat 2.3]
kisonecat has quit [Ping timeout: 250 seconds]
tilpner has joined #nixos
_lawlesseel has quit [Remote host closed the connection]
_lawlesseel has joined #nixos
_lawlesseel has quit [Remote host closed the connection]
<schopp0r>
hm. ok. can I also set the nix-channel to a local repo? because I want to have a nix-channel with my patches.
tilpner has quit [Quit: WeeChat 2.3]
<infinisil>
schopp0r: No need for a channel, you can use the repo directly, just set `nixpkgs=/path/to/nixpkgs` in your NIX_PATH
tilpner has joined #nixos
<schopp0r>
infinisil: but I want to use stable packages otherwise
<schopp0r>
ah btw: I used to maintain bluez-tools, but I no longer do that. should I remove the package, or is there some mark for "orphanned" packages?
_lawlesseel has joined #nixos
<schopp0r>
(I mean, it is not good to leave packages like bluez-tools unmaintained, but I do not use it anymore.)
agander has quit [Ping timeout: 245 seconds]
kisonecat has joined #nixos
kisonecat has quit [Ping timeout: 246 seconds]
<{^_^}>
[nix] @dtzWill opened pull request #2608 → EvalState::resetFileCache: clear parse cache as well as eval cache → https://git.io/fht3F
<infinisil>
schopp0r: if you want to you can just remove you as a maintainer, but i mostly don't even do that. Even when I don't maintain stuff, i could still be the only person having a slight idea about the package and could help with it
darkf has joined #nixos
<{^_^}>
[nixpkgs] @matthewbauer merged pull request #53089 → release: add more unstable release-critical jobs → https://git.io/fhLDu
<{^_^}>
[nixpkgs] @matthewbauer pushed 2 commits to master: https://git.io/fhtsM
Thra11 has joined #nixos
shabius has quit [Remote host closed the connection]
shabius has joined #nixos
shabius has quit [Max SendQ exceeded]
shabius has joined #nixos
shabius has quit [Remote host closed the connection]
pbb has joined #nixos
shabius has joined #nixos
shabius has quit [Remote host closed the connection]
eadwu has quit [Ping timeout: 250 seconds]
shabius has joined #nixos
[rg] has quit [Read error: No route to host]
kisonecat has joined #nixos
GiGa has quit [Quit: Leaving]
shabius has quit [Remote host closed the connection]
<{^_^}>
#53067 (by Izorkin, 1 day ago, open): haskel: error build package insert-ordered-containers and uri-bytestring
tzemanov_ has quit [Remote host closed the connection]
astrofog has joined #nixos
boredom101 has joined #nixos
orivej has quit [Ping timeout: 272 seconds]
astrofog has quit [Remote host closed the connection]
sanscoeur has joined #nixos
clefru has joined #nixos
agander_ is now known as agander
shabius has quit [Quit: Leaving]
<clefru>
swrast_dri.so wants glibc-2.27 because /run/opengl-driver/lib/dri/swrast_dri.so: /usr/lib/libm.so.6: version `GLIBC_2.27' not found (required by /run/opengl-driver/lib/dri/swrast_dri.so). However my FHS env puts /nix/store/9gfa9yn1r9nxjxcx6ppzg5k4hsxg9gx9-glibc-multi-2.26-131/lib/libc.so.6 into /usr/lib. Can someone point me how to kick that to glibc-2.27 (which is also installed in the /nix/store). I can't seem to find what controls glibc selection,
<clefru>
when building the environment. I don't want glibc-multi-2.26 but glibc-2.27. Any hints?
eadwu has joined #nixos
fusion809 has quit [Remote host closed the connection]
sanscoeur has quit [Remote host closed the connection]
cnidario has joined #nixos
<infinisil>
clefru: I think updating your system should do the trick
<infinisil>
So you have 2.27
<clefru>
ignore that question. glibc_multi is hardcoded in build-fhs-userenv.nix
<clefru>
infinisil: I do think that my system is fully up2date wrt to the nixos-18.09 channel
ddellacosta has joined #nixos
<clefru>
actually it's not. sorry for the noise.
cyounkins has joined #nixos
* clefru
needs to find a way to make nix-deploy push a new nixpkgs to the target system.
<infinisil>
I'm setting `nix.nixPath = [ "nixpkgs=${/path/to/nixpkgs}" ]` in my config
<pbb>
I wondered: is it somehow possible to use the results of the nixpkgs:trunk hydra builds? it is not really a channel, right?
Dedalo has quit [Quit: My MacBook has gone to sleep. ZZZzzz…]
<pbb>
(quassel was updated to 0.13 in nixpkgs, and I want to upgrade to that new version, but it is not yet included in the nixos-unstable channel)
<infinisil>
pbb: Yeah you can, just `nix-store -r /nix/store/<path-that-hydra-built>` to make that path exist
<pbb>
okay, but I would have to retrive the result manually? is there some way to make nix fetch it by adding something to the configuration?
<clefru>
infinisil++
<{^_^}>
infinisil's karma got increased to 49
<infinisil>
:D
<infinisil>
pbb: You mean automatically?
<pbb>
yes
<infinisil>
You need to get the exact nixpkgs version of that build
<infinisil>
It's in the inputs tab somewhere
<infinisil>
And then build the nix attribute path for that nixpkgs
boredom101 has quit [Quit: Page closed]
<pbb>
eh, but how would I import that? I can import that nixpkgs version from source basically, but then it would compile the new quassel system, instead of using the builds from hydra, right?
<infinisil>
pbb: Nah, if you use the exact same version it will get the builds from hydra without building anything
<pbb>
okay, thanks. I'll try that then :)
<infinisil>
That's true for every build in Nix: It first checks all your caches whether it was built by them. If yes it fetches those. If not it recurses on the dependencies and builds it itself.
eadwu has quit [Ping timeout: 250 seconds]
<pbb>
okay, I thought the binary distribution was somehow linked to the channels. are all results from hydra added to the cache, or only those from channels?
<gchristensen>
all the things
<pbb>
great
Dedalo has joined #nixos
cyounkins has quit [Remote host closed the connection]
cyounkins has joined #nixos
<pbb>
another question: why should I prefer fetchfromgithub over fetchgit?
<infinisil>
It uses github's automatically generated tarballs for every revision
<infinisil>
Which can be *much* smaller than the full repository
<obadz>
what's the state of the art of open source remote desktop? something hopefully faster than vnc and where I can just remote into an existing X session…
shabius has quit [Remote host closed the connection]
shabius has joined #nixos
shabius has quit [Max SendQ exceeded]
Xal has quit [Ping timeout: 245 seconds]
obfusk has joined #nixos
shabius has joined #nixos
Xal has joined #nixos
random_yanek has quit [Quit: random_yanek]
<hodapp>
well, SPICE from Red Hat has been a thing for awhile
<hodapp>
if it's specifically X you want, NoMachine NX has also been around awhile and is excellent for speed
<hodapp>
I have comfortably used remote X over slow DSL with NX
<simpson>
Vanilla Xorg is not that bad. I've used all kinds of things on Xorg over the network.
<hodapp>
if "the network" refers to a wired LAN, sure
nDuff has joined #nixos
<hodapp>
as soon as any amount of latency is involved, speed goes to crap because of how chatty X11 is and how many operations require round-trips across the network
ryjm has quit [Ping timeout: 245 seconds]
ryjm has joined #nixos
orivej has quit [Ping timeout: 250 seconds]
orivej has joined #nixos
pie_ has joined #nixos
jmeredith has joined #nixos
drakonis has joined #nixos
kisonecat has joined #nixos
eadwu has quit [Ping timeout: 252 seconds]
agander has quit [Quit: Leaving]
random_yanek has joined #nixos
random_yanek has quit [Max SendQ exceeded]
eadwu has joined #nixos
random_yanek has joined #nixos
<pie_>
Mic92, any luck with the bsd bootstrap
Church_ has joined #nixos
kisonecat has quit [Ping timeout: 250 seconds]
<jasom>
obadz: why faster than VNC? I find VNC to be reasonably efficient
<jasom>
IIRC wayland lacks a network-transparent protocol due to the assumption that VNC-like protocols would be sufficient.
cyounkins has joined #nixos
<jasom>
obadz: though note that I find xfreerdp to be more stable than x11nvc (both share your existing x session, the former using the remote desktop protocol and the latter using vnc)
<jasom>
obadz: I also find xpra to be reasonably good, but you need to start your session as xpra from the get-go. Unlike the other two solutions it allows you to have remote windows appear as distinct windows on the local machaine (rather than a virtual desktop of the remote machine in a single local window)