Bmaybe we thread this for discussion
BI'll restate in a better way than my emotional 
earlier.
earlier.BGitHub
Summary Revise scope of Universal Blue Main Images such that kmods (and associated kernel packages) are no longer included. Description The purpose of this change is to streamline build processes(c...
BSo I want to put together a set of of PRs in main, asus, surface, bazzite, and bluefin which demonstrate this scenario
BBI'll probably drop a distinct PR to demonstrate my thoughts better...
I appreciate the work on this PR, but I am most concerned about future maintenance... so, seems we should have a "legacy" and "normal" path that's a little easier to follow., so I'll probably branch from this branch for my PR, because I'll build on this work.
I appreciate the work on this PR, but I am most concerned about future maintenance... so, seems we should have a "legacy" and "normal" path that's a little easier to follow., so I'll probably branch from this branch for my PR, because I'll build on this work.
ESorry. I've been wanting to give this a proper response but we're about to leave for the next day and a half or so. If we changed the implementation a little bit (prevented tagged PR images from being used outside the org so that things proceed as normal), I think it would be awesome to do
EThe limitation I ran into here was that as soon as someone from an external repo opened a PR, it fell apart for a couple reasons that I'm going to have a hard time explaining here
EWe can jump into tty or something later tomorrow and maybe even today if I have any time
EReal quick, while I can, the cosign pairs have to match and the user also needs to be a part of that repo if I'm not mistaken. Meaning an outside PR from a different repo, will refuse to push to the tag
EAgain sorry
ETwo different things being discussed here
EDisregard all that ^ lmao
Bi think this was all meant for the other thread 
E@bsherman figured I'd necro this since F39 is due soon (release got pushed back a little). Just want to clarify what a legacy and normal path looks like to you. Do we split the Containerfiles away, removing all legacy stuff from the new stuff, then adapt the build workflow?
Bappreciate the necro
Bthe last week got really busy
Bi'm working on that PR now
EAwesome!
BI'm almost done... Containerfile/scripts are finishing testing now...
will just need to tweak the workflow
will just need to tweak the workflow
Bdo you want me to make a distinct PR or just update yours?
EEither way would work for me. Whatever is easier for you
Bwell, i'd prefer fewer PRs for less confusion, otherwise i don't care
Bsee what I did here if interested: https://github.com/ublue-os/main/pull/375/files
I have NOT updated workflows/build.xml yet... doing that now
I have NOT updated workflows/build.xml yet... doing that now
EI'll have a look 
EThis is much cleaner and much more straight to the point. Less confusing as well
Bthank you
Bit is working better than I had feared, too, which is good
EThat's awesome. Setting up CI shouldn't be too bad
Balmost ready to push again
Byeah, the main issue will be adding logic to the downstream repos
BDunno if you saw, but I did add some improvements to the README for akmods repo
https://github.com/ublue-os/akmods#how-its-organized
https://github.com/ublue-os/akmods#how-its-organized
EYeah. I'm going to need to investigate more of what that's going to take. I'll have a look
Bwe're going to have 2 resulting artifacts from this push... I'm getting rid of
*-nokmods and instead... all *-main images will be nokmods by default... but on F37/F38 there will also *-kmodsETook my medicine on an empty stomach and it just hit hard. One sec. Brb
Bso... the
but
Bazzite and Bluefin are more complicated becasue they build from main/nvidia/asus/surface repos.
*-nvidia images... to keep current behaviors the same... will need to build from *-kmods for F37/F38 but *-main for F39 and forward...but
asus and surface already build from *-nokmods so they just need to switch to *-main since those will always be "nokmods"Bazzite and Bluefin are more complicated becasue they build from main/nvidia/asus/surface repos.
EThis sounds great to me
Bugh
Bi just re-read our proposal
Bthis approach conflicts with the proposal
EAckk
Badding that timeline concession really made this hard 
EYeah. I was hoping to knock this out earlier but I've been busy as well
Bok, i'm going to think on the matrix a bit more but WITH the proposal up on my screen
ESounds good. I'll get things pulled up as well and assess what this should look like
Bok, just pushed again
Bi THINK this will work per the proposal
EAwesome. I'll have a look
Bhmm
