gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
<lovesegfault> danderson: are you around?
<danderson> lovesegfault: ohai, I'm from the internet and I'm here to help
<danderson> sup?
<danderson> (and just as I say that I step away for a minute because the cats are VERY adamant that it is dinner time and I am committing a crime by not serving it right now)
h0m1 has quit [Ping timeout: 250 seconds]
<lovesegfault> danderson: sorry, had to step away myself :D
<lovesegfault> Just curious behavior I'd like to report, I can't ssh into any of my hosts anymore using their tailscale IPs
h0m1 has joined #nixos-chat
<danderson> That is weird. This on nixos? What tailscale version? Got a tailscale IP for one of the machines so I can peek?
<danderson> (can /query if you want to keep it private)
<lovesegfault> On it
<lovesegfault> sorry, delayed messages I'm on a war with some neighbors
rajivr has joined #nixos-chat
<samueldr> we are SPOILED that all of the NixOS build instructions are neatly all self-contained in Nixpkgs
<gchristensen> +1
<lovesegfault> danderson++
<{^_^}> danderson was put on Santa's "nice" list
<gchristensen> lol
<danderson> lovesegfault++
<{^_^}> lovesegfault's karma got increased to 49
<danderson> for finding a tailscale bug :)
<lovesegfault> :D
<gchristensen> nice!
<danderson> (I mean also ++ for other things probably, but in this specific instance :)
<gchristensen> lovesegfault: do you erase on boot?
<gchristensen> I have problems with tailscale breaking my boot because of it, something about my mounts or something not setup right
<gchristensen> it means one of my machines gets a new identity on every boot :P
<lovesegfault> gchristensen: yes
<lovesegfault> My systems nuke on boot
<gchristensen> how do you deal with tailscale state?
<lovesegfault> retain, bind mount
<gchristensen> hmm
<lovesegfault> (with impermanence)
<gchristensen> can you shar...
<gchristensen> hot dog
<lovesegfault> lol
<gchristensen> this impermanence thing seems good
<lovesegfault> talyz, etu, and myself collab'd on it :D
<lovesegfault> it's in nix-community
<gchristensen> nice
<gchristensen> this is why I should learn to use flakes :)
<lovesegfault> I learned and it's a PITA
<lovesegfault> happy to hop on a call and share what I know whenever
<lovesegfault> ROI is dubious TBH
<gchristensen> I think the ROI will be high eventually
<samueldr> you don't need to use flakes to use that right now if you don't need to
<cole-h> lovesegfault: A little heartbroken you didn't mention me
<samueldr> could be a git submodule
<cole-h> I *did* come up with the name, after all :P
<cole-h> (/s)
<lovesegfault> it's true, without cole-h we'd never have gotten to making it public or actually collaborating on anything :D
<lovesegfault> cole-h++
<{^_^}> cole-h's karma got increased to 143
<cole-h> As we all know, the hardest thing is naming things
<gchristensen> samueldr: or builtins.fetchGit
<samueldr> yeah
<lovesegfault> gchristensen: if you think it's worthwhile I'd love to do a "converting my NixOS config to flakes" workshop
<gchristensen> but I think there are real advantages to flakes that submodules and builtins.fetchgit can't provide, once they've become really usable
* elvishjerricco doesn't know much about what flakes are and at this point is too afraid to ask
<samueldr> AFAIUI it's multiple concepts built together, but one of it is strict declaration of dependencies
<samueldr> and another is a strict interface
<samueldr> so you should know how to get the nixos module from a flake, as it will always be named the same way
<samueldr> I don't really _know_ flakes, so I might be a bit off
<gchristensen> standards around interop + support for sharing and external dependencies
<lovesegfault> my main gripe with flakes is it's basically eelco's stream of consciousness code at this point
<lovesegfault> after the RFC debacle
<gchristensen> yes that is regrettable
<elvishjerricco> `nixos-rebuild --flake .#mymachine` wtf is this syntax
<lovesegfault> using # is the most insane portion of the design
<lovesegfault> like, _why_
<lovesegfault> you really
<lovesegfault> *gotta pick the comment char?
<elvishjerricco> So if A depends on nixpkgs and B, and B depends on nixpkgs, will each of them be given a different nixpkgs version if they've pinned different versions in flake.lock?
<samueldr> fragment part of an URL
<elvishjerricco> Hm?
<lovesegfault> elvishjerricco: yes
<samueldr> # is the fragment part of an url
<lovesegfault> you can override with .follows
tomberek has joined #nixos-chat
<lovesegfault> e.g.
<elvishjerricco> I don't like that... I much prefer trying to get everything on the same versions of dependencies when possible.
<gchristensen> I'm pretty sure you can do that elvishjerricco
<gchristensen> with "follows"
<lovesegfault> you can with follows
<lovesegfault> (see the last link I sent)
<elvishjerricco> gchristensen: yea, but that's going to require a lot of follows boilerplate :P
<gchristensen> I'm not sure that is true in practice but I dunno
<elvishjerricco> Plus it requires investigating the inputs of all your dependencies, which is just annoying
<lovesegfault> sort of thing that could be solved by an RFC-driven design
<cole-h> the only real problem (which IIRC was solved recently?) was when follows wouldn't let you override a dependency to a different name or something
<cole-h> e.g. someflake.inputs.nixpkgs.follows = "asdf" wouldn't cascade to the nixpkgs in flakes *it* depends on
<cole-h> (and would error out)
<lovesegfault> gchristensen: every person I've recommended that mouse to, by the way, has thanked me and never gone back
<lovesegfault> highly recommend it :D
<lovesegfault> I own 3
<gchristensen> good to know :)
<gchristensen> my mice are all, uh, 11 years old now
<gchristensen> kensington slimblade and rollermouse red
<lovesegfault> the MX Ergo is phenomenal
<elvishjerricco> For something like nixpkgs, I feel like it'd be better for a flake to output an overlay rather than declare nixpkgs as an input.
<cole-h> flakes can do that :P
<samueldr> I don't think I'll ever stop using logitech mice
<samueldr> their warranty service is wonderful, and the warranty being long (but shrinking) is (was) great for someone like me who... really uses... computer hardware
<samueldr> I probably could buy that one switch that gave up the ghost, and service those mice... but then again I don't have the need for more than two
<gchristensen> good to know :) off to bed ... g'night y'all
<cole-h> o/ good night
waleee-cl has quit [Quit: Connection closed for inactivity]
<aaronjanse> Aaah I just spend an hour tracking down the cause of a function being called that should never be called in a special scenario
<aaronjanse> Turns out I was calling that function in a debugging statement :(
<aleph-> Heheh
endformationage has quit [Quit: WeeChat 2.9]
<elvishjerricco> oh good. patching systemd requires rebuilding git and cargo and who knows what else just to build a nixos image. Why??
<infinisil> elvishjerricco: Might be related to udev
<infinisil> (since udev = systemd)
<elvishjerricco> infinisil: Why would git depend on udev?
<infinisil> Hm, no idea
<infinisil> But I wouldn't be surprised for udev to be a deep dep in general
<elvishjerricco> Is there a why-depends for .drv?
<lovesegfault> git depends on the whole universe IIRC
<lovesegfault> because it depends on something for docs
<infinisil> elvishjerricco: nix-store -q --tree $(nix-instantiate '<nixpkgs>' -A git)
<lovesegfault> Maybe I'm misremembering
<infinisil> So yeah, git does transitively depend on systemd
<elvishjerricco> openssh depends on libfido which depends on systemd-minimal
<elvishjerricco> huh
* lovesegfault longs for Nix use flags
<lovesegfault> environment.use = [ "!docs" ]; :P
<elvishjerricco> this is why other distros don't like us :P
<infinisil> lovesegfault: What would that do?
<lovesegfault> infinisil: I'm just horsing around, there's a lot more stuff that would need to exist for something like that to work
<lovesegfault> Have you ever used Gentoo / do you know about USE flags?
<infinisil> Nope!
<elvishjerricco> I guess content addressable builds will be pretty good at preventing these mass rebuilds for stuff like this. My systemd package will be changed, so libfido will be rebuilt, but as long as its reproducible then it'll produce the same output and everything else stays the same
<elvishjerricco> AFAIU use flags are basically configuration options for telling gentoo how to build and install stuff. So !docs would mean don't bother building docs
<lovesegfault> TL;DR: Gentoo package "derivations" can be passed USE flags that toggle build features, and these flags are unified across the ecosystem. So for any package, if I pass !X I know for sure it won't build any GUI stuff
<lovesegfault> (assuming it was written properly)
<infinisil> Ah so like nixpkgs' config
<lovesegfault> because of that you can set _global_ USE flags, like !X systemwide to never have any GUI anything on your system
<infinisil> Well, but more thoroughly applied
<lovesegfault> there are a whole bunch
<infinisil> Oh yeah that's a bunch indeed
<lovesegfault> so it's like you'll see Nix drvs that have enablePulseaudio ? true
<lovesegfault> in Gentoo this is all standardized, so you can do this globally with ease
<cole-h> lovesegfault: So, when's the RFC? :D
<lovesegfault> I ain't doing it
<cole-h> I kid. But, as-described, sounds like a good deal to me.
<lovesegfault> it's pretty sweet
<elvishjerricco> loadavg 31. noice
<lovesegfault> FWIW: the flakes situation discouraged me from participating in the RFC process
<lovesegfault> what's the point if it's not normative
<elvishjerricco> did flakes have no kind of rfc?
<cole-h> it did, but it was closed so eelco could just develop it as "experimental" and then open another rfc when it's ready for mainstream
<cole-h> iirc
<lovesegfault> Kind of, it felt more like a "I don't like all this disagreement, imma withdraw and do it anyway"
<infinisil> +1
<cole-h> it would be cool to have a fully-featured flakes, then have an rfc for it, then hammer down all the nails y'know
<elvishjerricco> I guess *some* implementation is better than a never-ending argument over implementation.
<lovesegfault> it was kind of sad, especially coming from the Rust community...
<elvishjerricco> how would the rust community handle an inability to agree on an rfc?
<lovesegfault> AIUI eventually the responsible team will just call it
cole-h has quit [Ping timeout: 252 seconds]
<eyJhb> Finally able to sign in to my GitLab account. Haven't been able to access it for 4 weeks...
<eyJhb> Ohh. When you sign in from a new location, and it gives your IPv6 address.. Oh that's life <3
tomberek has quit [Ping timeout: 240 seconds]
<talyz> gchristensen: impermanence is good ;)
<talyz> (conflicts of interest apply, etc)
<eyJhb> talyz: Conflict of interest? Noooo... Never :D
<eyJhb> I use the newest version of impermanence on my server ;) My laptop however still uses the older symlink one
<eyJhb> I wonder if enough time has passed for me to get my certs again.
jess has quit [Quit: back soon]
<pie_> pam seems kinda opaque, thats always great for a security system
<pie_> " Note that regardless of what the pam_unix documentation says, accounts with hashed empty passwords are always allowed to log in."
<pie_> heh how would you even do that
sparogy has quit [Quit: ZNC 1.8.2 - https://znc.in]
sparogy has joined #nixos-chat
__monty__ has joined #nixos-chat
<leonardp> for anyone mildly interested in chess: https://www.bongcloud.art/ <-- magnus carlsen's "chess art collection"
<__monty__> I don't understand what exactly is up for auction. It looks like you could just transcribe the animations?
endformationage has joined #nixos-chat
<pie_> Wow. Well, I just got kicked from #openssh because someone was very up in arms about me trying to get it working with an empty root password
<pie_> no warning about "i will kick you if you keep trying to make this work" or anything
<srk> :D lol
<lukegb> hahaha
<pie_> edited for brevity https://bpa.st/YB5A
<pie_> its funny but also quite irritating
<pie_> given that someone was actually giving it a shot
<sterni> lol
<sterni> maybe they have a policy against pie_ rubes goldberging
<pie_> haha
<pie_> look
<pie_> theres _no reason_ this shouldnt work!!
<pie_> pie_ in a nutshell: <pie_> theres _no reason_ this shouldnt work!!
<sterni> haha
<pie_> like if it was "here is the code and weve disabled this and here is why" that would be different
<pie_> if it was a deliberate policy decision, but as it is its "heres some arbitrary behaviour who knows why"
Synthetica has joined #nixos-chat
cole-h has joined #nixos-chat
AkechiShiro has joined #nixos-chat
<samueldr> anyone with a gitlab.com account can help me confirm something?
<samueldr> compare the two URLs, are they giving a 404 the same way?
<eyJhb> samueldr: yes
<supersandro2000> shows no diff
<samueldr> supersandro2000: you need to be logged-in
<samueldr> eyJhb: you were logged-in right?
<eyJhb> Yeah :)
<eyJhb> Visual inspection however! :p
<samueldr> supersandro2000: because when you're not logged-in it redirects you to the sign-in page
<samueldr> eyJhb: that's fine enough
<supersandro2000> 🤔
<samueldr> that confirms my assumptions
<samueldr> it's to prevent enumeration
<eyJhb> Would be insane if it didn't say 404 :D
<samueldr> private.or.public.gitlab.instance/projectname/secrets
<eyJhb> At least privacy vice
<samueldr> that would confirm you're trying to accesss something that exists or not
<eyJhb> The only difference between the two copy pasted was a nounce in a script
<samueldr> eyJhb++
<{^_^}> eyJhb's karma got increased to 9
<supersandro2000> that shows a diff
<supersandro2000> in `x-gitlab-feature-category:` which is set to projects for one URL and to nothing for the other
<samueldr> oh
<samueldr> uh
<eyJhb> Niiiiiiiiiiiiiiiiiiiiiiiiiiiiiice :D
<samueldr> sorry supersandro2000 :)
<supersandro2000> the -I is important for the headers
<samueldr> I went the eyJhb way and only did a superficial visual observation at first
<supersandro2000> the HTML is the same
<eyJhb> Who is opening the PR?
<eyJhb> :p
<eyJhb> MR* because, Gitlab
<samueldr> supersandro2000: report it to gitlab.com, tha may not be enough for a bounty, but we never know
<supersandro2000> and one cookie has no expires for one connection but that could be curl
<eyJhb> Go go supersandro2000 :p If you don't we will (I at least will) :D
<supersandro2000> first let me google if that is intended
<samueldr> supersandro2000: cookie behaviour ALSO differs
<samueldr> try two non-existing projects
<supersandro2000> could be CURL caching
<samueldr> the _gitlab_session cookie has no expiry if the project exists (and is private)
<samueldr> nah
<supersandro2000> and the HTML differs
<supersandro2000> terminal to smal
<supersandro2000> ⋮ │<html><body>You are being <a href="https://gitlab.com/users/sign_in">redirected</a>.</body></html> \
<supersandro2000> No newline at end of file
<supersandro2000> ⋮ 1 │<html>
<supersandro2000> ⋮ 2 │<head><title>301 Moved Permanently</title></head>
<supersandro2000> ⋮ 3 │<body>
<samueldr> that's odd
<samueldr> I always have the same HTML though
<supersandro2000> 🤔
<supersandro2000> oh I think I know what happens. The bottom part is from my website when I tested if the diff works at all 😂
<supersandro2000> derp
<samueldr> haha
<samueldr> but still, good catch on the headers!
<supersandro2000> so let me disclose that
<samueldr> be transparent that it was discussed publicly here, that if it would have warranted a bounty, but this made it so it is not eligible anymore, that they consider donating the bounty equivalent to the nixos opencollective
<supersandro2000> yeah I should add that
rajivr has quit [Quit: Connection closed for inactivity]
waleee-cl has joined #nixos-chat
neeasade has joined #nixos-chat
<ar> am I the only one who thinks it's weird to implement in kernel something networking-related that's completly non-standard and has no support anywhere else? >Add support for TCP over UDP https://reviews.freebsd.org/D29469
tomberek has joined #nixos-chat
__monty__ has quit [Quit: leaving]
<Irenes> that's really strange
<Irenes> I'm not confident that I fully understand what the kernel's role in it is
jess has joined #nixos-chat
<samueldr> is it me or does it feel like a missed opportunity that at the highest log level, kernel modules being loaded are not part of the logs?
Synthetica has quit [Quit: Connection closed for inactivity]
<supersandro2000> >> Support for this is already available in Wireshark, packetdrill, and uperf.
supersandro2000 has quit [Killed (hitchcock.freenode.net (Nickname regained by services))]
supersandro2000 has joined #nixos-chat