<abathur>
I think I just learned some let's-just-pretend-that-didn't-happen knowledge
<gchristensen>
oooh do share
<drakonis>
oho
<drakonis>
that sounds like fun.
<abathur>
I am running an absurd little thing I am working on inside of a nix build (during checkPhase) in part to borrow the sandbox, so I'm writing some output to files and copying them to $out during install phase
<abathur>
so, I had some files turning up empty, and I've been chasing my tail for a while
<abathur>
but, they're getting copied before they're done getting written
<abathur>
if I sleep a few seconds, they have content...
<gchristensen>
oh god
<abathur>
hehe
<abathur>
ya
<gchristensen>
and did you figure it out?
<abathur>
well, the 5-second sleep before copying to $out/ solves my immediate problem
<abathur>
and absolves me of the impostor syndrome I was feeling as I tried to figure out how I was botching my FDs that badly and couldn't even figure out where
<abathur>
I think the reason it's not done writing is hopefully just intrinsic to the weird little thing I'm writing
<abathur>
I lowkey hope it proves useful enough to publicize, because I suspect publicizing it will cause shelludites to turn up at my house of horrors with their pitchforks
<abathur>
oops, shelluddites even
<abathur>
afk dinner
<gchristensen>
found a nice bug in osh on nixos stable, thankfully not in osh in unstable: exit traps not running when I expect :o
<V>
I was contemplating setting up yubikey for root
<V>
Possibly in addition to instead of replacing a password
<lovesegfault>
V: do you know how to do that?
<lovesegfault>
I've wanted to try the same, yubikey as root login 2FA
<lovesegfault>
but haven't been able to
endformationage has quit [Quit: WeeChat 2.9]
<energizer>
lovesegfault: did you try what the wiki says?
<lovesegfault>
energizer: last time I tried this I ran gentoo
<V>
lovesegfault: no, but presumably using PAM
<V>
With rules such that both modules have to pass
<V>
"You can also set security.pam.yubico.control to "required" in order to have multi-factor authentication."
<V>
I'd double check configuration.nix(5) however
<V>
IIRC the PAM subsystem is getting a major overhaul sometime soon also
<V>
risson has a PR
<lovesegfault>
considering how long just the unbound improvements are taking to land improving PAM will land 2024 :P
<V>
oh mood
<V>
I wonder what happens if you're using a github action from someone who github nukes the account of
<V>
Clearly PRs and issues disappear
<andi->
Can you vendor them?
<V>
Well, presumably, by copying the action to your own account
<andi->
Everything else without a git mirror on hardware you control is irresponsible
<energizer>
what's your threat model andi-
<andi->
Random dude on GH gets their account suspended. Now my project can't deploy because we don't have a copy of the action
<energizer>
you can fork for that, dont need to control the hardware
<andi->
Err those also go away
<V>
Given how many times I've misread the topic in here, I think we need #nixos-polar
<V>
NixOS but much less tropical
<andi->
energizer: see youtube-dl. All the forks were also gone that were done via the fork button.
<V>
Yeah, I explicitly said "copy" b/c of that
<V>
The current dependence on GitHub as a whole doesn't seem particularly good for us
<energizer>
did they delete non-fork-button forks (clone and reupload)
<V>
Only if reported I think
<V>
Would require manually finding those
<energizer>
you can just search for a commit hash github-wide
<andi->
No idea but I still think you want copies that are updated outside of that garden
<V>
There were reuploads as protest
<energizer>
maybe but pretty low risk imo
<V>
No that doesn't work if you just mangle the root commit
<energizer>
if you have to mangle all the commits then staying in sync is hard
<V>
You only need to mangle one commit
<V>
The hashes will cascade
<energizer>
yeah that means you can't pull
<V>
You can still work with patches
<energizer>
cuz they share no commits
<V>
Or rebase
<V>
That still works
<energizer>
too much work
<V>
Pull is just fetch + merge
<V>
No?
<energizer>
yes
<V>
Git is smart enough to realise which commits just had the committer or date or message or such changed
<V>
I rebase a *lot*
<V>
But the point is, none of this would be required if not depending on github as a platform
<energizer>
is nixos.org hosted in kazakhstan
<energizer>
or is it susceptible to dmca
<V>
This isn't about DMCA
<energizer>
why would github take down youtube-dl if not for dmca?
<energizer>
otherwise a simple fork is sufficient
<V>
This is about GitHub pulling users' accounts and this wiping out all their issues and PRs on the nixpkgs repo
<energizer>
in that case i misunderstood
<V>
I have tabs open that I've refreshed and gotten a 404 because the PR/issue just... disappeared
<colemickens>
I'm pretty annoyed about it
<colemickens>
I'm sure there's things I don't know, but it's very disruptive.
<andi->
With Nix we insist on having exact content hashes and a copy of the sources in the binary hash provided by hydra doing a good job. Yet we accept mutable things like tags in GH action and GH in general for nixpkgs. Looking back at at unison example from hours ago...
<samueldr>
sorry, cookie joke (oatmeal and raisin)
<ar>
heh
<hexa->
not sure if you noticed, but it has en ePrivacy filter
<colemickens>
^^ was just looking at the same thing
<ar>
yup
<ar>
nifty little feature, though probably won't be that useful now that i work full-time remote and don't commute
<ashkitten>
ar: is ar short for arachnist?
<ar>
ashkitten: it is
<ashkitten>
very good username
<hexa->
ar: I've only mentioned it because they mentioned it 4 times as a feature :P
ece has quit [Read error: Connection reset by peer]
ece has joined #nixos-chat
Guest88372 has quit [Read error: Connection reset by peer]
sorear has quit [Ping timeout: 256 seconds]
jared-w has quit [Read error: Connection reset by peer]
jared-w has joined #nixos-chat
sorear has joined #nixos-chat
Guest88372 has joined #nixos-chat
jared-w has quit [Ping timeout: 240 seconds]
cole-h has quit [Ping timeout: 264 seconds]
jared-w has joined #nixos-chat
__monty__ has joined #nixos-chat
lunc has joined #nixos-chat
<__monty__>
Just noticed something that's always annoyed me about firefox's "Save link as" dialog. It's focused on the last directory you saved to. More often than not I want to save files to sibling directories. This means I have to select the parent directory from a dropdown. Is there a way to have the parent directory visible in the first miller column? https://imgur.com/J2RG6fY
lunc has quit [Ping timeout: 246 seconds]
cosimone has joined #nixos-chat
j4m3s__ has quit [Quit: WeeChat 2.9]
lassulus_ has joined #nixos-chat
lassulus_ has quit [Client Quit]
<pie_>
since im nagging this poor linker author anyway, do yall have any wishes for linkers?
<colemickens>
I can't handle anything that doesn't use gitstatusd. nixpkgs is too tasking on any other prompt and once you taste gitstatusd you realize how slow every other git prompt is
<bbigras>
do you use it with Powerlevel10k?
<colemickens>
yes
<bbigras>
I should try to bribe the starship devs into looking into gitstatusd. according to github I'm a starship dev but I only contributed like 1 module.
<bbigras>
or at least make sure the new git implementation would be at least as fast as gitstatusd. otherwise it could be better to just use gitstatusd.
<siraben>
what's a good alternative to oh-my-zsh? i want something more lightweight but useful
<V>
(that aside, you can always shell out to bash; fish 3.0 is fairly similar in syntax though now)
<bbigras>
I get your point but it's unlikely I wont see the extra stuff when I paste it
<siraben>
V: i have a few other linux devices with bash on them and so on
<siraben>
is it a problem to use fish and switch back and forth?
<siraben>
or just install fish everywhere
<V>
while you're safe b/c you're using fish (which detects pastes and puts stuff on a multiline editor), for other people the snippet could just contain a newline
<V>
siraben: how do you mean?
<V>
are you asking if it'd be confusing to use two different shells at the same time?
<siraben>
V: yeah
<siraben>
So POSIX capability is fine losing if it means we're more productive personally?
<V>
I sometimes drop into bash or end up there due to nix-shell or w/e, I certainly haven't had any issues writing in it
<siraben>
some form of tradeoff
<bbigras>
V: ah I see. I didn't realize the paste would run
<V>
like, I've not forgotten how to write bash and it's fairly easy for me to tell which shell I'm in
<V>
that said, I used to specifically write POSIX-compliant shell scripts a lot of the time, so maybe I'm better practiced at it than the normal user
<V>
<siraben> So POSIX capability is fine losing if it means we're more productive personally? ← but by default? I'd say yes
<siraben>
Hm, yeah
<V>
fish is a lot more comfy to write in for interactive use, extra little functions, moderately complicated scripting
<siraben>
What about keybindings? Have to relearn those in fish or not?
<V>
ooh, yeah. there's a bunch of different stuff, but some very intuitive things
<V>
e.g. alt+left/right to navigate back and forth through your cwd history
<V>
there's a vi-mode, and also you can write your own key binds
<siraben>
will the usual emacs/readline keybindings still work?
<siraben>
I must say, I feel closer to switching than fish than before, losing posix compat was the only blocker for me
<siraben>
but from what i've heard from people who have switched it's often a non-issue
<bbigras>
ctrl-a and ctrl-e works with fish
<V>
Yeah, the emacs stuff will work by default. ^W, ^Y, ^A, ^E, ^K, ^U...
<V>
if you type stuff, ^F to complete the line, alt+right to complete single words
<V>
^N/^P works for history navigation if you don't want to use the arrow keys
<siraben>
i might try switching back to bash with some sensible defaults config first, see if that satisfies my needs
<siraben>
V: fish handles its environment variables differently right?
<V>
siraben: kind of. there's local variables (forced to be lexically scoped), normal variables (lexically scoped, but affect existing declared variables from upper scopes if such exist), exported variables, global variables, and "universal" variables. the lattermost propagate to all instances of fish and are persisted.
<V>
there's also "path" variables, which fish will automatically convert into arrays for processing
<V>
anything that ends in PATH will be a path variable, get split on colons, turned into an array, and joined w/ colons when exported
<V>
you can enable/disable this behaviour on any variable you choose
<V>
the fish manpages are really good
bkv has joined #nixos-chat
bqv has quit [Ping timeout: 258 seconds]
bkv is now known as bqv
neeasade has quit []
<abathur>
colemickens: wasn't aware of gitstatusd, or even to really do anything but whine about it; though I do note it's daemon is chewing up over 330 MB of ram after running gitstatusd_query in nixpkgs :o
<abathur>
I just used a bare gitstatus_start; haven't looked closely enough at the flags to know if specific options would reduce that
bqv has quit [Quit: WeeChat 3.0]
cirno-999 has quit [Ping timeout: 240 seconds]
cirno-999 has joined #nixos-chat
rajivr has quit [Quit: Connection closed for inactivity]
<siraben>
V: thanks, that's interesting
j4m3s__ has joined #nixos-chat
<__monty__>
Another vote for fish, it's incredibly productive in its default configuration. The small bits of syntax difference are hardly noticeable after a while imo, and fish's syntax is usually nicer anyway : )
<{^_^}>
serokell/deploy-rs#38 (by bbigras, 1 day ago, open): rollback even if I got "Success activating!"
aleph- has quit [Ping timeout: 256 seconds]
aleph- has joined #nixos-chat
cole-h has joined #nixos-chat
bqv has joined #nixos-chat
Raito_Bezarius has quit [Ping timeout: 260 seconds]
Raito_Bezarius has joined #nixos-chat
bqv has quit [Quit: WeeChat 3.0]
bqv has joined #nixos-chat
* infinisil
tries to get overwatch running on nixos
* supersandro2000
is very happy about his vgrep fzf alias
* infinisil
gives up
cosimone has quit [Quit: cosimone]
cosimone has joined #nixos-chat
Dotz0cat has quit [Ping timeout: 272 seconds]
<bbigras>
siraben: I have been playing the-powder-toy that I stole from your home.nix. Shame on you for making me waste all my Christmas Eve in my undies while my gf is waiting.
Guest88372 has quit [Ping timeout: 260 seconds]
sorear has quit [Ping timeout: 260 seconds]
jared-w has quit [Ping timeout: 260 seconds]
jared-w has joined #nixos-chat
sorear has joined #nixos-chat
Guest88372 has joined #nixos-chat
jared-w has quit [Ping timeout: 260 seconds]
jared-w has joined #nixos-chat
<supersandro2000>
GitHub finally got around making their review window consistent between the discussions and the file tab