qyliss changed the topic of #spectrum to: A compartmentalized operating system | https://spectrum-os.org/ | Logs: https://logs.spectrum-os.org/spectrum/
ghavil has joined #spectrum
cole-h has quit [Quit: Goodbye]
<qyliss> Think I'll try to do a stream this evening (central Europe time)
MichaelRaskin has quit [Ping timeout: 264 seconds]
<pie_[bnc]> not sure ill be able to drop in but ping me?
MichaelRaskin has joined #spectrum
MichaelRaskin has quit [Read error: No route to host]
leah2 has quit [Ping timeout: 240 seconds]
thefloweringash has quit [Ping timeout: 240 seconds]
aranea has quit [Ping timeout: 240 seconds]
aranea has joined #spectrum
leah2 has joined #spectrum
thefloweringash has joined #spectrum
edef has quit [Ping timeout: 252 seconds]
Irenes[m] has quit [Ping timeout: 240 seconds]
colemickens has quit [Ping timeout: 240 seconds]
IdleBot_57398a17 has quit [Remote host closed the connection]
<qyliss> slight problem: my laptop overheats when using the internal graphics to stream
Irenes[m] has joined #spectrum
colemickens has joined #spectrum
MichaelRaskin has joined #spectrum
cole-h has joined #spectrum
edef has joined #spectrum
<pie_[bnc]> :I
<qyliss> yeah...
<qyliss> Actually not using the graphics seems to work well enough, but I'm also having audio issues
Irenes[m] has quit [Ping timeout: 252 seconds]
edrex has quit [Ping timeout: 260 seconds]
colemickens has quit [Ping timeout: 240 seconds]
thefloweringash has quit [Ping timeout: 256 seconds]
IdleBot_3d9d0c82 has joined #spectrum
<qyliss> Okay, I think we'll be able to make this work
puck has quit [Read error: Connection reset by peer]
puck has joined #spectrum
thefloweringash has joined #spectrum
<qyliss> Will be streaming quite soon I think
<cole-h> Make sure to not show gopass windows like colemickens did :P
<qyliss> You will need either VLC or mpv because I haven't set up a web player
<MichaelRaskin> Wait, mplayer would not suffice?
<MichaelRaskin> (hmmm, I wonder if literally single video element would be a good enough web player)
<qyliss> it would not be
<qyliss> If it supports h264/aac over rtmp it'll work
<qyliss> browsers do not support rtmp
edrex has joined #spectrum
colemickens has joined #spectrum
Irenes[m] has joined #spectrum
<qyliss> oh, right -- people who netsplit: there's going to be a stream soon
<MichaelRaskin> Hmm, Nginx seems to support transcoding to HLS on the fly?
<qyliss> yes, but I have not set that up and I want to get on with streaming :P
<MichaelRaskin> Heh
<MichaelRaskin> Ouch Ouch Ouch «run a local VLC to transcode to ogg and let your Nginx proxy it»
<qyliss> My laptop is close to melting with this as it is
<MichaelRaskin> I mean, it's ouch-worthy in any case
<qyliss> Indeed :P
<qyliss> Okay, if I do a stream now, is anybody going to tune in?
<qyliss> cc cole-h hyperfekt
<cole-h> I'm here and ready to ask (mostly) stupid questions :)
<MichaelRaskin> I am pretty sure that if the link is not posted, nobody will tune in
<cole-h> ;^)
<hyperfekt> qyliss: right now not because i'm on a rare visit to my mom and we're about to watch a movie but you can definitely expect me to tune in in the future
<qyliss> awesome :)
<qyliss> rtmp://hyperion.kookie.space:11011/alyssa.tv
<qyliss> Please tune in and let me know what you see
<multi> qyliss: firefox in dark theme, and stacked alacritty and gnu emacs
<qyliss> lovely
<cole-h> Yep, same
<cole-h> I can hear youuuu
<cole-h> Except it's left channel only it sounds like
* multi doesn't have headphones handy, can't check audio
* cole-h takes off right headphone then
<cole-h> :P
<multi> huh, TIL that emacs is up to v28
<cole-h> From HEAD it is
<cole-h> Yep, emacs is fullscreen
<tazjin> yes
<MichaelRaskin> Hmmm, not sure I know how to make mplayer respect proxy… giving it unlimited network seems to work. Should there be sound?
<qyliss> yes
<cole-h> There is for me
timj has joined #spectrum
<samueldr> my left ear hears that
<cole-h> Known issue for this stream
<samueldr> yeah, didn't look carefully *enough*
<cole-h> :D
<samueldr> = af toggle "lavfi=[pan=1c|c0=0.5*c0+0.5*c1]" ; show-text "Audio mix set to Mono"
<edef> hmm
<samueldr> addingt his to .config/mpv/input.conf will make = re-mix as mono
<tazjin> in VLC you can rightclick -> Audio -> Stereo mode -> Mono
<edef> i'm definitely not seeing anything in mpv
<edef> but VLC works :3
<MichaelRaskin> Mplayer gives picture just fine
<samueldr> yep, searched that for something else in the past for mpv, the gobbledeegook I shared is about that
<MichaelRaskin> Not the sound
<ashkitten> oh, i see the stream now
<edef> hmm. no video for me after restarting my vlc
<ashkitten> hm left channel only yeah
<edef> i can't seem to recover video now, very strange
<cole-h> Woof
<MichaelRaskin> OK, after some effort configuring VLC it seems to agree doing both video and sound
<ashkitten> got it!
<ashkitten> mpv rtmp://hyperion.kookie.space:11011/alyssa.tv --af="lavfi=[pan=1c|c0=0.5*c0+0.5*c1]"
<edef> i've only gotten audio on mpv so far
<edef> i had video on vlc briefly
<MichaelRaskin> vlc: Audio → Stereo Mode → Left
<ashkitten> sometimes it doesn't have video, i had to restart it once to get video back
<edef> ooh
<edef> now i have video
<tazjin> it works perfectly fine for me on vlc
<cole-h> samueldr: Thanks for that binding, works flawlessly
timj has left #spectrum [#spectrum]
<edef> qyliss: there's an Infallible type in std
<edef> (it's also in core, fwiw)
<cole-h> ^
<edef> when the never type stabilises, Infallible will become an alias for it
<edef> at least, that's what the docs promise
<tazjin> is there any functional difference between ! and Infallible?
<tazjin> or is it just syntax?
<tazjin> I'd expect both get optimised similarly?
<edef> ! is somewhat special
<cole-h> One is unstable, one isn't :P
<edef> it's a genuine bottom type
<tazjin> oh, I see
<edef> and thus implements every trait, since invoking with a ! parameter already involves proving falsity
<edef> or every trait that has it in parameter positions for every thing
<edef> but that bit is controversial
<edef> and the primary reason it hasn't been stabilised, aiui
<ashkitten> ooh i like the rainbow variables, i just noticed that
<ashkitten> i wonder if there's a nvim plugin for that
<cole-h> Rainbow delimiters is fantabulous
<ashkitten> ooh
<ashkitten> i've been trying to get vscode working but it doesn't like when the dev environment isn't loaded at startup
<tazjin> we looove dense argument lists
<ashkitten> i don't have very much context for what you're doing
<samueldr> (don't take it wrong, but I've been doing other things with you in the background, so I've not tried to follow :))
<samueldr> scheduled reminders of what's going on may be good anyway
<cole-h> +1
erictapen has joined #spectrum
<ashkitten> yeah that's much more clear now
<samueldr> get other people doing your bidding!
<cole-h> >:)
<tazjin> I can't unsee the "devcie" typo in the arg description
<samueldr> ashkitten is impressionable enough to start contributing to nice projects :3
<ashkitten> oh hush
<ashkitten> samueldr is referring to how when my phone stopped booting into android i spent a few days trying to get mobile-nixos working on it
<ashkitten> i don't have a lot of energy to contribute to big projects, though
<samueldr> s/trying to get/getting/ :)
<ashkitten> well, yes, but i never did get wifi to work
<samueldr> you had it going as successfully as I do the other devices
<samueldr> (for now)
<cole-h> Why to_string vs String::from?
<cole-h> (Just seems like String::from is used above, rather than to_string)
<ashkitten> it's more that i haven't had the energy to understand what parts of spectrum i would have the knowledge and experience to work with
<ashkitten> i'll look at it some day
<cole-h> Woo, already contributing!
<MichaelRaskin> Mwa ha ha. You have experience with making things boot despite their opposition? It would be nice to have a faster-booting (i.e. smaller) kernel config for booting in CrosVM for Spectrum purposes. If it is also sufficient to run NixOS tests in Qemu, even more people will be grateful
<samueldr> for nixos integration it'd have to work on aarch64-linux too, or be drop-in replacement enough though
<samueldr> (commenting about "If it is also sufficient to run NixOS tests")
<ashkitten> i could attempt to look at the kernel, but i don't actually have any experience with linux kernels outside of my short time working with mobile-nixos (and that was mainly just fighting with an oem kernel)
<samueldr> (if that was about running nixos tests in crosvm)
<samueldr> though I think I see now it must be about the minimal config
<MichaelRaskin> For NixOS integration it needs to be a drop-in replacement anyway — people will want to run the same test switching _only_ the kernel to make sure there are no catches
<samueldr> it'd be nice to know the actual tested minimal kernel config required for nixos
<MichaelRaskin> CrosVM for NixOS tests is probably too complicated (even to spec)
<MichaelRaskin> ashkitten: well, minimal kernel is not about editing kernel, it is about a simple-ish description of kernel config for Nixpkgs kernel builder that results in something nice and small and usable
<ashkitten> right, yeah
<ashkitten> i mean, i can look at it and probably just disable everything and progressively enable things until it passes tests
<ashkitten> seems simple enough
<samueldr> btw, ashkitten, `make nconfig` doesn't fail with being unable to find ncurses stuff
<ashkitten> i haven't tried getting spectrum working on my computer yet. is there a quickstart doc for devs?
<MichaelRaskin> There us no full Spectrum to get working
<cole-h> Completely unrelated to all that's going on, but: what monospace font do you use? e.g. in your sway bar
<qyliss> iosevka
<samueldr> s/(.*)=m$/# \1 is not set/
<cole-h> That's what I used for a while as well, thought so.
<cole-h> I like that it's so configurable as well.
<samueldr> as a tall monospace I used M+ for a while, quite neat
<samueldr> but not customizable
<cole-h> I'm using JetBrains Mono right now, because my utf8 tall bars in weechat look strange in alacritty with Iosevka (e.g. some overlapping of glyphs)
<ashkitten> sounds good to me
<ashkitten> i'm building that now, see if it works
<ashkitten> oh right, i've never used a mailing list before haha
<cole-h> At least 45 minutes
<qyliss> Oh lol the stream died
<cole-h> (I had to restart my mpv for unrelated reasons)
<cole-h> Oh so it did
<samueldr> what a coincidental time to die
<cole-h> :'(
<qyliss> That's because of a hilarious bug
<cole-h> "Wow, it's been over an hour" -> *ded*
<erictapen> ^^
<qyliss> Let me see if I can restart it
<qyliss> Stream back? You might have to restart your player.
<samueldr> it restarted
<samueldr> didn'T need to restart mine
<erictapen> same here
<samueldr> yes
<erictapen> yes
<cole-h> Stream's back and didn't need to restart either
<edef> yep
<cole-h> Audio good
<samueldr> oh no
<cole-h> :D
<samueldr> that's one way to know how long you've been streaming for
<cole-h> 👀
<ashkitten> i am building a kernel right now
<ashkitten> generally a mainline nixos kernel takes around 12 minutes on my machine i think?
<cole-h> Thanks for the stream! Very enlightening -- hope to see another one soon-ish when I can actually pay full attention :^)
<samueldr> I figure the plan still requires /dev/kvm to work, right?
<samueldr> damn you qualcomm!
<samueldr> they can do KVM, but it's gated behind their firmware not giving it out
<samueldr> something about how aarch64 does the equivalent to intel "rings"
<samueldr> the virtualization features are gated for fancy servers
<samueldr> (and possibly the new upcoming ARM surface)
<samueldr> if you were to buy one of the current ARM windows laptops you can't have hardware virt :(
<samueldr> it's really qualcomm here
<samueldr> yes, they can
<samueldr> those are qualcomm laptops :)
<ashkitten> risc cpus generally don't have special instructions for virtualization i thought, since they don't need to?
<samueldr> aarch64 chromebooks are either rockchip (rk3399) or mediatek (some awful name)
<ashkitten> hm, i should get a better cpu cooler to be able to run at a higher clock while compiling stuff
paragraph has joined #spectrum
<samueldr> even other aarch64 SoCs generally don't control the EL tightly like Qualcomm firmwares does
<samueldr> so kvm works on raspberry pis 3 and 4
<samueldr> on allwinner devices
<samueldr> all of them for now
<samueldr> but there's like 4 or 5 models
<samueldr> they're basically glorified cellphones, with a keyboard and screen
<samueldr> the way the firmware and OS is setup is as... not good... as it is for qualcomm-based android devices
<erictapen> I just saw it in my Mastodon timeline
<paragraph> somebody boosted your mastodon post
<ashkitten> hello i'm back, what's going on?
<ashkitten> looks like it's wrapping up the kernel build
<erictapen> I'm also stuck at building linux atm btw^^
<samueldr> if you need some pointers / guidance in figuring out an aarch64 setup for spectrum later, I'll be glad to advise
<ashkitten> ,pointers
<ashkitten> oh wait
<samueldr> (no bot here)
<ashkitten> ahaha
<ashkitten> demo works good
<samueldr> I was sure it was going to go that way
<ashkitten> i run sway
<ashkitten> how do i exit the demo tho lol
<samueldr> it is now your life
<ashkitten> eh, pkill crosvm
<erictapen> thanks for streaming! I think next time I'd even imagine to enter it into my calendar if you want to announce it upfront.
<qyliss> Awesome! I'll see if I can do that.
<ashkitten> hmm, how do i get a base kernel with nothing enabled in nixpkgs to start?
<qyliss> You sorta don't, IIRC
<qyliss> Module support is mandatory, for example
<ashkitten> oh, it is?
<qyliss> I think so
<qyliss> One thing you can do is just supply your own kernel config file, which would be fine for us
<ashkitten> that sounds reasonable, yeah
<qyliss> And just do make menuconfig or whatever
<qyliss> lol still one person tuned into the stream
<samueldr> (make nconfig, make menuconfig may have a kernel-mainline bug)
<samueldr> in fact make nconfig now also has a bug
<samueldr> (I can't simply use a patch as it has to work through all revisions)
<samueldr> ncurses changed the way it builds its .pc files, which I think should be okay, there's been 0 regressions in nixpkgs from that change
<samueldr> but the linux kernel links in a weird way, which I think makes our magic LD_FLAGS break when linking
<qyliss> ncurses kernel config stuff is my one kernel patch :D
<qyliss> (which improved the error message if you don't have them)
<ashkitten> where is the git repo for the linux-cros kernel?
<samueldr> ncurses <6.1-something always had the -L/path/to/lib/dir in its .pc files, for pkg-config, but starting wtih ~6.2 it doesn't anymore
<samueldr> a real fix for mainline should be figured out though
<ashkitten> it amuses me how much cpu horsepower cloning a kernel takes
<ashkitten> kernel repo*
<qyliss> ashkitten: Spectrum currently uses the chromeos-4.19 branch of that repo
<ashkitten> ty
<qyliss> (because that's what Google is currently shipping)
<ashkitten> how do i use a config file for the kernel build?
<samueldr> I wonder if they do, I've never been able to build a gru-scarlet config that works like theirs do
<qyliss> ashkitten: pkgs.linuxManualConfig
<ashkitten> cool, ty
<qyliss> I'm not sure how you'll use that with linux_cros...
<qyliss> Maybe with linuxPackages_custom?
<qyliss> lmk if you get stuck and I can try to get something to work
<MichaelRaskin> ashkitten: if you manage to make it not literally a fixed full .config, it might be easier to maintain across kernel versions (and to reuse)…
<ashkitten> i'll start with a full config at first because that's the easiest thing
<qyliss> MichaelRaskin: OTOH you run into bugs like https://lore.kernel.org/lkml/87wob4tf9b.fsf@alyssa.is/, which has me pretty skeptical of Nixpkgs' approach.
<MichaelRaskin> Well, I would say it is a true upstream bug, though
<qyliss> I suppose, but even then it's hell to actually debug a kernel configuration the nixpkgs way
<samueldr> I've been dumping config files in mobile nixos, which is pretty much a non-issue because of how OEM kernels never get a significant upgrade :(
<qyliss> yikes :(
<MichaelRaskin> Actually, a lot of stuff might be not just about a smaller kernel, but about kernel command line to avoid some costly choices
<MichaelRaskin> I wonder what is the true overhead of using everything via modules, and also of using innitramfs
<MichaelRaskin> Which surely simplify debugging
<ashkitten> ugh, kernel config stuff sucks
<ashkitten> i'll work on this later
erictapen has quit [Ping timeout: 258 seconds]
erictapen has joined #spectrum
paragraph has quit [Ping timeout: 240 seconds]
<hyperfekt> ashkitten: idk if this is still relevant, but i usually configure any paths to tools that vscode needs to invoke nix-shell - but there's also an extension that lets you load a nix-shell environment for each window
<ashkitten> hyperfekt: it's kinda icky with rust stuff i think
<ashkitten> i don't remember the exact issues though
<hyperfekt> ashkitten: hm, i seem to have rust-analyzer and cargo working. haven't looked into non-rust dependencies yet tho
<hyperfekt> too bad i missed the stream. the movie was good tho. i'll be returning home on wednesday so looking forward to joining one of the next ones
<qyliss> When would people like another one?
<Irenes[m]> we haven't shown up to one yet but we'd love to... notionally, if it's earlier than about this time of day we're less likely to be able to attend, but we know you're in Europe so that's probably not realistic
<Irenes[m]> no other constraints on our end
<MichaelRaskin> Hmm, you might be also fine with a late-morning European time stream?
<Irenes[m]> oh yeah quite possibly
<Irenes[m]> our night meds kick in around 05:00 GMT, so anything before that is fine
<MichaelRaskin> Ah, I have never seen qyliss posting something that early
<Irenes[m]> yeah ah well
<Irenes[m]> when we were younger we didn't have these constraints, but medicating our bipolar stuff creates serious and inflexible limitations
<Irenes[m]> and we're better off this way really
<Irenes[m]> it's just too bad when things like this don't work
<MichaelRaskin> Sure — I was just checking for the low-hanging fruits.
multi has quit [Quit: bye]
multi has joined #spectrum
erictapen has quit [Ping timeout: 256 seconds]