qyliss changed the topic of #spectrum to: A compartmentalized operating system | https://spectrum-os.org/ | Logs: https://logs.spectrum-os.org/spectrum/
colemickens has joined #spectrum
* colemickens waves
<qyliss> yay!
<qyliss> That validates that Irenes[m]' Matrix instructions work :)
<qyliss> Put that on the website (I forgot before): https://spectrum-os.org/participating.html#irc
<colemickens> Very prompt of you :) thanks!
<qyliss> Not very prompt -- I meant to do this 6 days ago!
edrex has joined #spectrum
* edrex waves to Cole Mickens , qyliss
<qyliss> Hello!
<Irenes[m]> yep! you're welcome :)
<Irenes[m]> hi Cole
<colemickens> Hello! Thanks! I've been trying to join the room for... oh gosh too long now.
<colemickens> I mentioned SpectrumOS in #sway and got the low-down.
<colemickens> Confused as to why the normal join method didn't work though.
<Irenes[m]> it's a bug in the bridge
<Irenes[m]> it is possible that the cause involved manual cleanup to room aliases that the bridge maintainers did at some point, but that's not fully understood
<qyliss> Has anybody got in touch with the bridge operators about it?
<qyliss> I've never used Matrix so I wouldn't know what to say.
<colemickens> I've reported at least two rooms broken, once since last Nov and they remain broken, seems to be a known-ish long-standing (set of?) bug(s).
<qyliss> :(
<colemickens> Seems like it might be a tricky component.
<colemickens> It makes me really sad actually, I can't evangelize Matrix (esp to certain folks who happen to also be in broken rooms) with the current bridging experience.
<colemickens> I'm not sure it's a good use of my time to try to fix though, I don't know Matrix, or IRC, or the language it's written in.
<colemickens> Anyway, Rust and Nix are more my style. :) Speaking of which...
<colemickens> Going to ask random questions about Spectrum if it's cool
<colemickens> https://spectrum-os.org/git/ random, but is it custom software or just the footer stripped of software info or?
<qyliss> It's cgit with a non-default theme
<colemickens> Any regular meetings even on IRC? Any f2f (zoom/jitsi?) or just irc+ml?
<qyliss> The original theme comes from https://git.causal.agency/src/tree/www/git.causal.agency/custom.css, and I modified it a bit to make it work better in some layout edge cases
<qyliss> There are no regular meetings or anything at the moment, largely because I'm the only person working on Spectrum at the moment.
<qyliss> But if there was demand we could set something up.
<qyliss> I suggested recently that streaming myself working on Spectrum might be a good way to get people acquainted with what's going on if they're interested in helping
<qyliss> And a couple of people were interested, so I'll probably look into that.
<qyliss> Once there's a bit less churn on crosvm I was going to write a ML post about some Rust tidy up somebody could help with if they were interested.
<colemickens> qyliss: is there a roadmap for crosvm or just a general feeling that there's some features landing and then some quiet?
<colemickens> or do you mean churn you're doing in the downstream fork/
<qyliss> the latter
<qyliss> To explain what I'm working on:
<qyliss> Upstream crosvm doesn't really have any way for two VMs to communicate
<qyliss> That's obviously quite important for Spectrum
<qyliss> So for most of this year I've been working on adding that, essentially.
<colemickens> I see.
<colemickens> Would there be value in translating the design document into large chunks of efforts that will have to be made?
<qyliss> Probably, but that sounds like a lot of work.
<qyliss> And it might not be very clear at this point.
<qyliss> For example, I didn't expect to have to implement this feature in crosvm nearly this early.
<qyliss> But it became clear to me when I was talking to some potential contributors at 36c3 that almost everything they could work on actually depends on it.
<qyliss> There's not even a milestone in my grant agreement for all this crosvm work -- it's not even funded beyond GitHub Sponsors
<qyliss> But I realised that we need it.
<colemickens> So, here's my thought: I wonder if TWI posts can be valuable for driving user+dev interest, even if it's just a post that re-introduces the general idea/motivations for the project, with the blurbs you just gave me.
<colemickens> And then the next few things that might be unblocked.
<colemickens> I didn't take a super close look (esp at branches and other repos) so I didnt know there was such active dev or that you're about to unblock potential parallel work.
<colemickens> I guess 36c3 people did? Did you give a talk?
<qyliss> No, I was just introduced to a couple of interested people
<qyliss> Thank you for all these ideas, btw. It's nice to see someone take such an active interest.
<qyliss> So, my concern with a TWI post is that not much happens week-to-week, from an outsider's perspective.
<colemickens> Just a thought, and I don't say it without thinking of the fact that writing such an update is not necessarily trivial to just throw out.
<qyliss> Like, every week this year it would have been "worked a bit more on crosvm, still doesn't work", up until last week
<qyliss> In which case it would have changed to "worked a bit more on crosvm, now it works kind-of"
* colemickens nods
<qyliss> I've so far avoided giving talks or anything because I'm wary of promising things to to many people that I end up not delivering on.
<colemickens> Well, maybe that idea wasn't the best one for now. The streaming one hits it a bit though, if you had a feed of them on spectrum-os.org. idk, I feel like maybe I'm projecting though, I don't really know how people evaluate whether to look past the first page, or dig into projects in early stages. The design doc certainly entices.
<qyliss> I do write on my blog when interesting stuff happens
<qyliss> It's long overdue a status update
<colemickens> Ah, cool!
<qyliss> (I'm sort of supposed to do one every couple of months)
<colemickens> The most recent most is certainly interesting to me.
<qyliss> I just noticed that it comes up first or second in a bunch of search engines if you search for "virtio wayland", which is nice
<qyliss> Because prior to that post there was basically no information on it out there.
<qyliss> I've also been thinking I should add a list of links to the homepage
<qyliss> It's kind of difficult to navigate with them mixed up in the text
<qyliss> And I could link to my blog on there.
<qyliss> colemickens: anyway, thanks very much for the suggestions. maybe talk more about this tomorrow, but now I must sleep.
<qyliss> I appreciate your interest.
<IdleBot_e923fa8a> Hm, I did not think about VM-to-VM virtio-wl as a prerequisite for some contributions; would any of them be unblocked by re-packaging my nsjail setup to get rid of the Lisp part, to provide Nix-installable CrosVM+netns+9p+VNC experience as an early prototype?
<IdleBot_e923fa8a> 9p being rust-9p here; I guess rust-vnc could be packaged too without much trouble
<IdleBot_e923fa8a> OK, rust-vnc last auth/crypto, but the communication will be entirely inside sockets and net NS-es, so should not be too bad
* colemickens has sunk many hours into building sommelier. including since this post was written :(
<colemickens> oh hey, maybe that's useful though, I have some code the xdg-stable v1 update. I'll look and see if you've done that yet.
<qyliss> IdleBot_e923fa8a: this isn't just about virtio_wl
<qyliss> This is inter-VM communication in general.
<qyliss> If I can do one virtio device I can do them all, pretty much.
<qyliss> colemickens: I have. What I could really use now is wl_compositor v4.
<qyliss> It would be very nice not to rely on that hack.
<colemickens> oh! okay, that's a specific thing. It'd be nice to get involved. I'll put it on the list for the weekdn.
<qyliss> :D
<qyliss> It didn't occur to me that a Sommelier package was likely to be useful to other people, but it sounds like maybe it would be
<qyliss> Spectrum nixpkgs has some really nice infrastructure for building Chromium OS components.
<qyliss> (pkgs/os-specific/linux/chromium-os/common-mk/default.nix)
<colemickens> Sommelier would let you bypass current restrictions around wayland+wlroots+xwayland
<colemickens> Or rather, workaround it by cheating?
<colemickens> But the stars are aligning for a proper fix for that anyway, so I'm less interested in sommelier for that reason now.
<colemickens> qyliss: yeah, I'm trying to dig into spectrum's nixpkgs, uh, realizing I don't know how to read the git log when it's not rebased :S
<colemickens> I should've just added a remote for it and checked it out locally anyway, I'm sure I can figure it out then, or use a GUI client :P
<qyliss> colemickens: git log origin/master..spectrum/master
<qyliss> If you do have a go at hacking on Sommelier, start-vm.nix on the crosvm-demo branch gives you a really easy way to test.
<colemickens> okay!
cole-h has quit [Ping timeout: 260 seconds]
<IdleBot_e923fa8a> qyliss: well, isolated networking is an option for VM-to-VM communication…
<hyperfekt> reading about xdg-desktop-portal is very promising. i feel like it's filling the niche that would've needed the most effort to implement and would be most crucial for adoption
<Shell> hyperfekt: it’s honestly one of the most exciting things in desktop Linux to me right now.
<qyliss> IdleBot_e923fa8a: for _some_ VM communication yes, but if we did that, the code we were able to write would end up having to be rewritten later
<qyliss> So might as well get it right the first time
<IdleBot_e923fa8a> Well, maybe I indeed underestimate the dependence on «which kind of socket» in the parts of SpectrumOS that I do not expect to use; thanks for helping me measure my systematic bias a bit better
<qyliss> Which kind of socket isn't really relevant, of course.
<qyliss> But I can't just shovel virtio drivers over a network.
<qyliss> Andlots of things are implemented in existing virtio drivers.
awordnot has quit [Ping timeout: 268 seconds]
awordnot has joined #spectrum
cole-h has joined #spectrum
MichaelRaskin has joined #spectrum
MichaelRaskin has quit [Ping timeout: 255 seconds]
MichaelRaskin has joined #spectrum
tilpner_ has joined #spectrum
tilpner has quit [Ping timeout: 258 seconds]
tilpner_ is now known as tilpner
vilhalmer has joined #spectrum