Also - never before have I seen OS devs move a tenth so fast to address usability concerns, UX, and

Also - never before have I seen OS devs move a tenth so fast to address usability concerns, UX, and reliability. It's already made a huge difference for me personally, and it's exactly the sort of thing I aspire to and built my major around. I really appreciate it and I look forward to having time to contribute more directly

jstone•2/27/23, 4:00 PM

We're able to move fast because we're leveraging some of the best features of containers, which is making a build system that's relocatable, easy to offload onto other hosts, can be structured in whatever makes the most sense for our needs, and is easy to normalize with one pull spec composing an entire file tree compared to having to test against any number of package configurations

CChaiQi Also - never before have I seen OS devs move a tenth so fast to address usabilit...

dreamyuki•2/27/23, 4:01 PM

Ublue feels like the future of linux desktop

Jjstone We're able to move fast because we're leveraging some of the best features of co...

ChaiQiOP•2/27/23, 4:02 PM

Good stuff!

j0rge•2/27/23, 4:03 PM

yeah it's easier when you start from scratch

jstone•2/27/23, 4:03 PM

Semi-relevant, but we're able to piggyback off of the shoulders of tech giants in the best way possible https://xkcd.com/1988/

xkcd: Containers

All services are microservices if you ignore most of their features.

j0rge•2/27/23, 4:03 PM

plus we can move faster than fedora because we don't have to use pagure, bugzilla, etc.

Marco (马可)•2/27/23, 4:04 PM

I absolutely agree with jstone, but appreciate your sentiment. This wouldn't have really been possible 2 years ago - it's right at the edge of a tipping point in how we build desktop distros

dreamyuki•2/27/23, 4:06 PM

Imagine if ubuntu switch to same model as silverblue rn
With OCI image etc

j0rge•2/27/23, 4:09 PM

they won't

j0rge•2/27/23, 4:10 PM

they won't even finish ubuntu core desktop so they're mostly actively avoiding it outside of iot

dreamyuki•2/27/23, 4:12 PM

Ubuntu core desktop is planned?

j0rge•2/27/23, 4:14 PM

yeah it's been around for years

j0rge•2/27/23, 4:16 PM

https://github.com/canonical/ubuntu-core-desktop

GitHub

GitHub - canonical/ubuntu-core-desktop

Contribute to canonical/ubuntu-core-desktop development by creating an account on GitHub.

ChaiQiOP•2/27/23, 4:16 PM

that is... a sad-looking preview, for what it represents

dreamyuki•2/27/23, 4:17 PM

Last update is 1 month ago

j0rge•2/27/23, 4:18 PM

yeah it's only a side thing

j0rge•2/27/23, 4:18 PM

but also who knows what's private

jstone•2/27/23, 4:56 PM

Canonical's take: "what if we deployed a distro as a bunch of containerized images that get updated side-by-side?", Silverblue's take: "what if we deployed a distro as one containerized image that gets updated in one place?"

ChaiQiOP•2/27/23, 4:58 PM

Heh
Sounds about right

dreamyuki•2/27/23, 5:00 PM

Silverblue way sounds better

ChaiQiOP•2/27/23, 5:01 PM

Normally it would sound worse to me
But the way it's implemented makes it far better

ChaiQiOP•2/27/23, 5:02 PM

Because it does the opposite of sacrifice modularity to achieve that singular image

jstone•2/27/23, 5:03 PM

I think snap has a weakness that its developers weren't "brave enough" to go all-in with snapifying a distro. If they decided to have an entire OS as one big snap, it would basically be a much less efficient ostree

jstone•2/27/23, 5:05 PM

They probably didn't know how to make large images that could be extended like how rpm-ostree can install layered RPMs, so the next best thing is to break up a distro into a bunch of snaps

ChaiQiOP•2/27/23, 5:10 PM

Honestly I think it feels a lot cleaner and more reliable to just have one big eldritch monster I have to go to special effort to even see which does everything from running the DE to browsing to text editing to who knows what, instead of having a hundred snaps cluttering several obvious places for a few apps and a bunch of pieces of an OS

ChaiQiOP•2/27/23, 5:15 PM

Eventually I'll unlayer a bunch of that, once I figure out why distrobox refuses to integrate

Jj0rge @hermmm you should be able to rekickoff the PR builds, lmk if you want me to do ...

xyny•2/27/23, 6:07 PM

I think I did it, nothing changed though.

j0rge•2/27/23, 6:29 PM

k I'll take a look in a bit

j0rge•2/27/23, 9:08 PM

alright, lunch and then github time!

jstone•2/27/23, 9:28 PM

Oh hey, does github CI go faster if it's pulling images from ghcr.io instead of quay.io? I'd imagine there'd be a higher chance of getting a cache hit if it's pulling images that're on the same host

j0rge•2/27/23, 9:28 PM

I didn't notice one way or the other when we switched out the images

j0rge•2/27/23, 9:44 PM

@KyleGospo heya after I review and merge herm's stuff into base I'm thinking just removing vanilla's thing from the base image, keep 37 with herm's improvements and then basically call it "finished", but then do 38 with the new wizard, etc.

j0rge•2/27/23, 9:44 PM

Can you keep the copr building it though?

Kyle Gospo•2/27/23, 9:44 PM

yep, won't touch that

j0rge•2/27/23, 9:44 PM

also

j0rge•2/27/23, 9:45 PM

I was doing GHA stuff last night, there's tons of ways to have github build and host RPMs

j0rge•2/27/23, 11:19 PM

sigh

j0rge•2/27/23, 11:19 PM

this permissions problem makes testing impossible

j0rge•2/27/23, 11:20 PM

It can't push any packages into the registry

Marco (马可)•2/28/23, 1:28 AM

It feels like a critical blocker

hyperreal•2/28/23, 1:32 AM

My next step for Vauxite would be creating flatpaks for common Xfce desktop packages

hyperreal•2/28/23, 1:33 AM

And somehow getting them to be installed by default and removing the GNOME ones

hyperreal•2/28/23, 1:33 AM

anyone here have experience creating flatpaks?

j0rge•2/28/23, 1:36 AM

I don't, but I found in the installer that it installs 0 flatpaks, so when we make an ISO we don't have to rip out stuff, we can just add what we want, which will be a welcome change, heh