gchristensen changed the topic of #nixos-chat to: NixOS but much less topical || https://logs.nix.samueldr.com/nixos-chat
ehmry has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
ehmry has joined #nixos-chat
evanjs has quit [Read error: Connection reset by peer]
<joepie91> tech demo of my new project: http://joepie91-home.cryto.net:3500/ (example query: stm)
evanjs has joined #nixos-chat
<cole-h> oh nice
<joepie91> other vendors to be added, and more specialized search engines like this to be built :)
<joepie91> (for other types of data)
<joepie91> the idea is crowdsourced development of tailored scrapers
<joepie91> to basically build many different search engines that are optimized for a given purpose
<joepie91> and that actually semantically understand the data they're working with
<joepie91> there's a matrix room at #seekseek:pixie.town for those interested in following along :P
ashkitten has quit [Quit: WeeChat 3.0]
<cole-h> I'm interested, but not interested enough to join matrix :DDD
<joepie91> lol
ashkitten has joined #nixos-chat
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-chat
<samueldr> joepie91: neat! now we only need a sci-hub, but for datasheets archiving
<joepie91> yeahhh
<joepie91> samueldr: I do plan on mirroring indexed datasheets, but I also fully expect some manufacturers to go "hey don't do that"
<samueldr> yeah
<joepie91> the full search indexes will be available as a public feed though
<samueldr> joepie91: send save queries to archive.org?
<samueldr> then provide an archive link?
<joepie91> that's one possible approach, but I'll need to think carefully about the best approach there
<joepie91> and probably discuss it with archive.org first
<samueldr> I think it's a situation of all of the answers above!
<joepie91> :P
<joepie91> if you're on Matrix, there's #seekseek:pixie.town btw
<joepie91> oh I mentioned that here already
<joepie91> it's definitely time to sleep :D
<samueldr> n/p
<samueldr> and 'night!
<joepie91> night :)
<cole-h> o/
<aaronjanse> Welp, I think I've spent a full 12 hours today trying to create a wireguard netns
<aaronjanse> I managed to get a container that's wireguard-only, but I can't figure out how to create a normal netns
<aaronjanse> I've been trying to follow https://www.wireguard.com/netns/, a bunch of Nix stuff via github code search, and lots of stackoverflow queries
<samueldr> gchristensen: but what's a reference?
<aaronjanse> Thanks for the link! That looks helpful
<gchristensen> hah
<aaronjanse> ^^samueldr asking the real questions
<gchristensen> a reference is what a reference does
<samueldr> that's self-referential
<aaronjanse> Hmm I'm not sure what "referential" means here
<aaronjanse> ./s
<aaronjanse> lovesegfault: Early this morning (read: 5 am, late last night) I managed to get a wireguard interface via wg-quick that doesn't force my entire system to go through wireguard
<aaronjanse> I ended up learning a lot about basic kernel networking stuff in the process
<aaronjanse> My new idea is to have something replicating the wg0 interface in both the main namespace and another namespace
<aaronjanse> Maybe I'm looking for a bridge??
<aaronjanse> "networking.wireguard.interfaces.<name>.interfaceNamespace" WHAT
<aaronjanse> .... that would have saved me a lot of time!
<lovesegfault> :^)
<aaronjanse> As said by someone I know, "classic nixos"
<aaronjanse> Ofc there's an option for it
<ajs124> aaronjanse: I'm just using systemd's JoinsNamespaceOf thing for that
<ajs124> Which kind of sounds like an option I just made up, but that actually exists
<aaronjanse> One day we'll just need to `services.antigravity.enable = true;`
<gchristensen> (insert xkcd()
<lovesegfault> Uh, does this SVG show up okay for yall? https://raw.githubusercontent.com/ayu-theme/ayu-colors/master/colors.svg
<ashkitten> kinda funky
<lovesegfault> Here the labels are all fudged
<gchristensen> here too
<ajs124> in its infinite wisdom, my kde decided that I want to see this svg in krita. rasterized. anyways, labels are weird as well.
<lovesegfault> if I use imagemagick to convert to jpg its still fudged
<lovesegfault> but if I use chrome it displays fine
<ashkitten> firefox displays weird
<lovesegfault> seems like it _only_ works in chrome
<lovesegfault> bizarre
* lovesegfault tries inkscape
<ashkitten> inkscape isnt very good iirc
<samueldr> the y= values are baked-in wrong
<lovesegfault> works in inkscape :)
<ashkitten> lmao
<lovesegfault> nvm
<samueldr> but svg in general isn't very intercompatible
<lovesegfault> it starts good
<lovesegfault> but then inkscape fudges the bottom half
<lovesegfault> lol
<lovesegfault> cursed
<samueldr> too many ways to break while totally following the spec
<aaronjanse> Oh yeah, I was playing with that. It looked useful for putting services behind a VPN ajs124
<samueldr> I assume any svg is meant to be consumed strictly by the software that produced it
<samueldr> if it works elsewhere, then it's good luck
<samueldr> (for more complex SVGs)
<samueldr> you should see the horrible thinkgs adobe illustrator exports and calls "SVG"
<aaronjanse> lovesegfault the svg is all wonky on Firefox mobile
<lovesegfault> aaronjanse: I think it _only_ works on chrome
<samueldr> here I assume it's "baked-in" wonky
<aaronjanse> But that looks like a nice color scheme 👀
rajivr has joined #nixos-chat
<lovesegfault> aaronjanse: ye, I use ayu-dark for everything
<lovesegfault> what does chrome use to render svg?
<lovesegfault> skia?
<samueldr> oh wow
<samueldr> it does render properly in chrome
<lovesegfault> :D
<aaronjanse> (side note the nixos option worked *flawlessly*)
<samueldr> lovesegfault: I think it's whatever in their webkit fork
<aaronjanse> I'm hyped
<lovesegfault> anyone with an apple device to test if it works in safari?
<samueldr> lovesegfault:
<samueldr> I know the issue
<samueldr> the user agent styles for svg are "broken"
<samueldr> in chrome, check the "computed" tab
<aaronjanse> Ah
<samueldr> find font-size
<samueldr> 13px for <rect>
<samueldr> 12px for <text
<samueldr> <text>*
<samueldr> and well, em are relative to those sizes!
<samueldr> it's been baked for the different sizes!
<lovesegfault> O.O
<samueldr> meanwhile in firefox, 12px both ways
<samueldr> so using `em` while relying on the user agent stylesheet is a fool's errand it seems!
<lovesegfault> samueldr++
<{^_^}> samueldr's karma got increased to 328
<samueldr> that's an odd discrepancy between user agents
<samueldr> and weird how <rect> *somehow* does not have the same default font size as <text>
<lovesegfault> that's what weirds me out the most
<lovesegfault> like... why?!
<samueldr> browsers: not even once
<aaronjanse> Write once, deploy everywhere^Wsomewhere
<infinisil> Consider this: Ramen in the morning
<infinisil> Or in general, hot food after waking up
LnL has quit [Quit: exit 1]
<samueldr> infinisil: coffee?
<samueldr> oatmeal?
LnL has joined #nixos-chat
LnL has joined #nixos-chat
LnL has quit [Changing host]
<samueldr> toasts?
<samueldr> tea?
<infinisil> Hmm
<infinisil> Nah those don't count :P
<samueldr> I'm considering it
<infinisil> I kind of really want Ramen right now, but it's almost bed time
<infinisil> So I thought, why not make it in the morning :)
<samueldr> I don't think there's a reason really
<samueldr> "breakfast" food is... food... when you think about it
<infinisil> WANT
<infinisil> Yeah I think I'll just cook some ramen tomorrow morning, which will be a first for me :)
<infinisil> But I am hungry now.. and I don't really wanna sleep with an empty stomach..
<infinisil> Damn these human bodies, needing constant maintenance
<aaronjanse> Oooh I think I'm gonna have Ramen tomorrow, too
<aaronjanse> Thank you for the idea, infinisil
<infinisil> Nice :D
<aaronjanse> Also, I think I might try to extend a nix LSP to let you jump-to-definitions via flake inputs
<aaronjanse> So you could ctrl click on packages and nixos options in some scenarios, maybe
<aaronjanse> Even cooler would be docs when you hover over a nixos option
<infinisil> Also nice, a step in the right direction!
<aaronjanse> I feel like it would significantly change the way I personally use Nix
<lovesegfault> infinisil: how are you skinny?
<lovesegfault> I feel like every time we talk it's about you eating more ramen
<lovesegfault> share your secrets
<infinisil> Hehe
<infinisil> I actually eat very little sweet
<infinisil> s
<infinisil> If I'm getting hungry in the evening, I'd rather wait a bit so I can have something proper
<lovesegfault> "something proper" is code for ramen?
<infinisil> Yeees :P
<infinisil> Or other pasta too
<njha[m]> nix for human body when
<infinisil> Honestly most of my hot meals are pasta
<njha[m]> `nix eat ramen`
<infinisil> spaghetti, ravioli, ramen
<infinisil> Newly soba
<aaronjanse> Bet njha
<aaronjanse> Same here infinisil.. I cook many varieties of alfredo
<siraben> `nix build .#ramen`
<siraben> > ramen
<{^_^}> undefined variable 'ramen' at (string):492:1
<siraben> > chocolateDoom
<{^_^}> "<derivation /nix/store/jvjc2y7p8rl983q3kmyphcrf68cw198x-chocolate-doom-3.0.1.drv>"
<cole-h> OK, tomorrow I'm gonna try the boot counting PR.
<cole-h> maybe we can finally get it merged soon
<{^_^}> #84204 (by danielfullmer, 50 weeks ago, open): [WIP] nixos/systemd-boot: boot counting and automatic fallback
<cole-h> s/_/)/
waleee-cl has quit [Quit: Connection closed for inactivity]
<lovesegfault> cole-h: getting that merged would rock
* lovesegfault wishes there was a way on putting bounties on nixpkgs PRs
<lovesegfault> I'd put $10 in the pot for that PR
<lovesegfault> easy
endformationage has quit [Quit: WeeChat 2.9]
lopsided98 has quit [Ping timeout: 264 seconds]
lopsided98 has joined #nixos-chat
slack1256 has quit [Remote host closed the connection]
<lovesegfault> ,locate bin/gsettings
<{^_^}> Found in packages: glib.bin
lopsided98 has quit [Ping timeout: 264 seconds]
<lovesegfault> cole-h: are you still around?
<cole-h> Yessir
<lovesegfault> do you have the code to gen an aarch64 sd image handy?
<lovesegfault> perfecto :)
<lovesegfault> cole-h++
<{^_^}> cole-h's karma got increased to 133
* lovesegfault squints
<lovesegfault> cole-h: looks like nixos-aarch64-images doesn't support the raspberry-pi?
<Mic92> lovesegfault: the raspberry pi is already support by the aarch64 image of nixos
<lovesegfault> Ah, I see
<lovesegfault> Mic92: might want to add it to the "supported boards list"? Or even just a footnote somewhere :D
<lovesegfault> Mic92++ (for the project)
<{^_^}> Mic92's karma got increased to 49
<Mic92> lovesegfault: feel free to edit it.
<lovesegfault> on it
<{^_^}> Mic92/nixos-aarch64-images#2 (by lovesegfault, 7 seconds ago, open): readme: mention raspberry pi support
<lovesegfault> actually
<lovesegfault> I'm dumb
<lovesegfault> totally misunderstood what you said
* lovesegfault closes PR
lopsided98 has joined #nixos-chat
<Mic92> lovesegfault: well, one could just passthru the unmodified image for aarch64
<Mic92> I don't think this support rpi4 so
<eyJhb> lovesegfault: Did you add it to the wiki instead?
cole-h has quit [Ping timeout: 265 seconds]
evanjs has quit [Read error: Connection reset by peer]
evanjs has joined #nixos-chat
<lovesegfault> seems like the rpi4 image is totes borked
<lovesegfault> Oh wow, it's been broken since 2021-01-11
<lovesegfault> last ok build: https://hydra.nixos.org/build/134720986
lopsided98 has quit [Ping timeout: 264 seconds]
__monty__ has joined #nixos-chat
lopsided98 has joined #nixos-chat
Mic92 has quit [Quit: WeeChat 3.1]
Mic92 has joined #nixos-chat
Raito_Bezarius has quit [Ping timeout: 264 seconds]
aranea is now known as mewra
Raito_Bezarius has joined #nixos-chat
aei has quit [Ping timeout: 264 seconds]
Baughn has quit [Ping timeout: 246 seconds]
das_j has quit [Ping timeout: 240 seconds]
ajs124 has quit [Ping timeout: 240 seconds]
ajs124 has joined #nixos-chat
das_j has joined #nixos-chat
lukegb is now known as lukeg
Baughn has joined #nixos-chat
Baughn has quit [Ping timeout: 256 seconds]
<siraben> Does anyone have koka-lang packaged?
<siraben> got them path issues https://github.com/koka-lang/koka
<siraben> Oops meant to paste https://github.com/koka-lang/koka/issues/163
<{^_^}> koka-lang/koka#163 (by siraben, 6 days ago, open): Koka fails to find libraries when built with Nix
cole-h has joined #nixos-chat
Raito_Bezarius has quit [Ping timeout: 264 seconds]
endformationage has joined #nixos-chat
waleee-cl has joined #nixos-chat
lunc has joined #nixos-chat
<gchristensen> wow ovh is in a bad spot with sbg
<srhb> sbg?
<LinuxHackerman> Strasbourg data centre location
<gchristensen> SBG2 had that big fire, and then a couple days ago SBG1 had a bank of batteries nearly catch
<srhb> Ah
<srhb> I mean, how the hell...
<srhb> Then again, if your batteries are the smoky kind...
<gchristensen> batteries make good fire ...
<__monty__> Did they get the same batteries for both sites?
<gchristensen> probably, SBGx are all on the same property
<srhb> The one time I experienced a full-site DC outage it was also the batteries, as I recall. They were moved a distance away from the site afterwards.
<srhb> Guess that's not a bad idea, where possible.
<srhb> It is morbidly fascinating to follow their rebuild at a distance.
Baughn has joined #nixos-chat
kcalvinalvin has quit [Quit: ZNC 1.7.4 - https://znc.in]
kcalvinalvin has joined #nixos-chat
<cole-h> lovesegfault: It is time. See you on the other side.
cole-h has quit [Quit: Goodbye]
<lovesegfault> time for what?!
cole-h has joined #nixos-chat
<cole-h> still boots without any "failed" services -- now to add a failing service and test that
cole-h has quit [Client Quit]
cole-h has joined #nixos-chat
rajivr has quit [Quit: Connection closed for inactivity]
Baughn has quit [Quit: ZNC 1.6.2+deb1 - http://znc.in]
<cole-h> lovesegfault: Sorry; time for boot counting PR testing :D
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos-chat
<samueldr> someone should test in a V'
<samueldr> VM*
Baughn has joined #nixos-chat
<cole-h> there's already a VM test written for it
<cole-h> I'm testing on a real system :P
<cole-h> "real"
cole-h has quit [Quit: Goodbye]
<gchristensen> what's the test for|
cole-h has joined #nixos-chat
cole-h has quit [Client Quit]
<samueldr> [14:23:31] <cole-h> [...] time for boot counting PR testing :D
<gchristensen> oh cool
cole-h has joined #nixos-chat
cole-h has quit [Client Quit]
cole-h has joined #nixos-chat
Baughn has quit [Quit: ZNC 1.6.2+deb1 - http://znc.in]
<cole-h> Great, it works!
<cole-h> One final test...
cole-h has quit [Quit: Goodbye]
cole-h has joined #nixos-chat
<lovesegfault> cole-h++
<{^_^}> cole-h's karma got increased to 134
Baughn has joined #nixos-chat
<cole-h> In the words of GLaDOS: "Great success!"
<bbigras> isn't it "Huge success?"
<drakonis> it is huge success
<cole-h> whatever
<cole-h> lol
<gchristensen> Medium success!
<f0x> okay i guess success!
endformationage has quit [Quit: WeeChat 2.9]
<bbigras> you did it, I suppose.
endformationage has joined #nixos-chat
<lovesegfault> I'm eating the bread satan trampled on trying to do that split-output linkFarm cole-h
<cole-h> lol
endformationage has quit [Client Quit]
<srk> now we need that on pi4 :D
<gchristensen> satan-trampled bread?
<srk> uefi and systemd-boot
<srk> now I'm wondering what satan-trampled bread is
<ashkitten> bread that satan trampled on, clearly
<samueldr> srk: uefi is trivial if you don't mind prebuilt artifacts
<srk> hope so, gonna try soon-ish
<samueldr> though it'd be nice to have it packaged
<samueldr> well, BUILT and packaged
<samueldr> but anything UEFI in device-tree land is going to be painful
<samueldr> since the mainline linux kernel kind of assumes (even though they say they don't) that you'll be using whatever device tree they authored for a given kernel revision
<samueldr> I have some notes about a solution for that, well... two different methods even
<samueldr> but it'd require a lot (well... not that much) development work
<srk> uboots dtb ignored?
<srk> or no passtru mechanism?
<samueldr> it's
<samueldr> it's more complex than that
<srk> oh cool :)
<samueldr> when booting through the "extlinux compatible" mechanism with u-boot, you end up loading the appropriate dtb file for the given kernel revision
<samueldr> but otherwise, you wouldn't!
<samueldr> you would use whatever FDT was already in place in memory during u-boot's run
<samueldr> which in theory should be fine, but in practice will be lacking when the kernel updates
<samueldr> same with Tianocore
<srk> ah, fun
<samueldr> because there is no mechanisms in any UEFI bootloaders to load a dtb file for a given generation
<samueldr> I guess that's because most distros only care about one kernel at a time
<srk> have another intermediate linux bootloader doing extlinux.conf & kexec?
<samueldr> so they just make the firmware load the only dtb file
<samueldr> srk: that would be an option, but not a *good* one
<samueldr> one concept is to shove that task back into the kernel code
<srk> iirc it's done that way on openpower
<samueldr> they want a specific dtb? then load it yourself!"
<samueldr> srk: yeah, but that's on openpower... with their linux-based boot flow
<samueldr> we were talking UEFI
<samueldr> the kernel already had all the bits available in the source tree to associate a given platform to a given dtb file... so it should, as early as possible, pick them from an appended initrd file
<samueldr> it could even add a /linux,dtb-compatible property to handle that better!
<srk> interesting, sounds good
<samueldr> this way *whatever* provenance it is, UEFI, not UEFI, the kernel would be able to load what it wants
<samueldr> for the given generation
<samueldr> another option would be a UEFI shim
<samueldr> do you know execline?
<srk> not really, recently stumbled on it
<samueldr> an "execline-like" UEFI shim that serves to do only one thing: load the appropriate FDT in memory then "exec" into whatever EFI program
<samueldr> this assumes that bootloaders end up executing an EFI program and not doing something else
<samueldr> which is good enough I think
<samueldr> execline's philosophy is that the program reads whatever cli args it handles, then `execve` into leftover command-line arguments
<samueldr> that's what's important here for my example
<samueldr> this shim idea, though, is extremely specific to being started by EFI, and execing into EFI
<samueldr> but should be feasible enough
<lovesegfault> gchristensen: it's a portuguese expression :P
<lovesegfault> you use it for when you talk about a task or event that was really horrible
<srk> samueldr: sounds like it, even buildable on nixos
<lovesegfault> "I had to eat the bread Satan trampled on in college!"
<samueldr> srk: what do you mean buildable on nixos?
<srk> samueldr: building the efi shim with nix
<samueldr> yeah, it's just software
<lovesegfault> "US bureaucracy is insane! I ate the bread Satan trampled on just to get a visa"
<samueldr> two dfferent links
<samueldr> this is what you were describing earlier
<samueldr> I'm not confident openpower *actually* does that, or if it does openpower differs enough from ARM platforms
<srk> nice!
<srk> not sure either, but it's open so we can take a look.. :D
<samueldr> because I found literally NO mentions of passing forward the nodes
<samueldr> the /memory node
<samueldr> and when I was asking around in the #devicetree channel
<samueldr> no one had any clues on how to do this with kexec
<samueldr> (AFAICT there really is no way)
<samueldr> they all assumed you'd just keep whatever FDT is in memory
<samueldr> even though, given what I **experienced** with many boards, it'll break in many different annoying ways!!!
<srk> uff :) scary channel. /me hesitates a bit
<samueldr> meh, nothing scary
<samueldr> they had a good attitude
<samueldr> they didn't assume I was doing something wrong
<samueldr> simply didn't saw that before
<samueldr> haven't seen*
<srk> good to know there's such chan, only knew about #u-boot
<samueldr> srk: just to be clear, that pi4 uefi thing I linked earlier is in no way related to u-boot
<samueldr> it's tianocore, "the" reference UEFI implementation
<samueldr> it's in no way related to u-boot's implementation
<samueldr> I highly suggest investigating it *too* whenever you do your rpi+UEFI thing
<samueldr> the best setup for a raspberry pi 3 I had was using their pi3 tianocore build
<srk> cool, will take a look. there are two devices I want to try it with - pi4 and omnia
<srk> maybe netboot would be even better solution to boot rollbacks and these but with router it's kind-of bootstrapping issue. NAS with multiple ifs I guess
<srk> s~and~with
<samueldr> yeah, if you want the same setup on misc. ARM platforms you're probably better off with u-boot
tomberek has joined #nixos-chat
<samueldr> I uh... may need to add a flashing image warning to the PR adding plasma mobile apps
endformationage has joined #nixos-chat
<ashkitten> i love that the "these weeks in firefox" blog posts have a section for "resolved bugs (excluding employees)" because it makes it sound like some of the bugs work at mozilla
<ashkitten> also yesssss theyre finally finally removing native theming for content
<ashkitten> no longer will dark gtk themes make your browser unusable
Dotz0cat has quit [Ping timeout: 264 seconds]
Dotz0cat_ has joined #nixos-chat
<aaronjanse> ashkitten: I can't find that in the release notes. Does that mean that websites can no longer choose light/dark mode based on the system theme when using Firefox?
<ashkitten> in nightly
<ashkitten> css dark theme hint will still work
<ashkitten> but it will have browser native form element themes instead of using your gtk theme
<gchristensen> das_j: we can get a full rebuild out in 24h if need be, but even if it is 4 days we're doing as well as the well funded distros
<das_j> I see the regular struggle with merging
<das_j> not the hydra build
<das_j> and backporting
<gchristensen> I don't think it'll be an issue, this sort of patch should go right to master
<das_j> oh good to know
<das_j> I thought it would also go through the regular staging(-next) stuff
<gchristensen> part of why we keep staging staging: it keeps master healthy and crisis level patches like heartbleed can get out right away
Raito_Bezarius has joined #nixos-chat
<aaronjanse> I'm slightly disappointed that I cannot use `nix run` as a shebang on a flake.nix, but oh well
<samueldr> I love software that segfaults on a missing undeclared dependency from their cmakelists
<samueldr> makes me real confident it'll work fine with features I haven't or couldn't test yet
<cole-h> I remember opening a PR against a project to fix a missing dependency
<cole-h> that was fun
<samueldr> according to my packaging work, $other_software shouldn't run, at all
<samueldr> since over a year ago
<samueldr> (I won't name them... it might be that I'm doing something wrong)
<samueldr> but it builds fine!
konubinix has quit [Ping timeout: 264 seconds]
matthewcroughan_ has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
matthewcroughan has joined #nixos-chat
jess has quit [Quit: K-Lined]
matthewcroughan has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
j has joined #nixos-chat
matthewcroughan has joined #nixos-chat
matthewcroughan has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
matthewcroughan has joined #nixos-chat
<elvishjerricco> So I know there's the bash syntax `{a..f}` for expanding to `a b c d e f`, but is there a way to expand to something like `-o a -o b -o c -o d -o e -o f`?
<samueldr> kind of, but will not split as you'd like: "-o "{a..f}
<samueldr> it will create 6 arguments, not 12
<lovesegfault> does anyone have the incantation to dump the depgraph of a drv to graphviz handy?
<lovesegfault> I lost it in my bash history