Hi All,
I'd like to raise the topic of post-151 development.
Jon (Tibble/Meths) - obviously you've been doing great work on the
prestable builds. I think I asked when we should ship to /stable, and
basically I got the impression you weren't comfortable doing that until
we had most of the CVEs fixed and could support it from a security
standpoint.
I don't think that's realistic right now, as CVEs are coming out faster
than we can fix them.
So as such, I'd like to propose that one of the next prestables be a
candidate for being shipped to "/release", rather than "/stable" and the
notion of /stable be abandoned for now.
We could do something like aim to push a /release every 6 to 12 months,
and we can aim to backport critical bug and security fixes to it from
/dev. But it will give users a bit of stability and free us up to move
/dev forward faster.
That fits in with OpenSolaris, where /release wasn't guaranteed free of
security holes (you had to pay for that). Obviously we can still ship
CVE and bug fixes to it, but it sets realistic expectations and
hopefully addresses your concern with using the name "/stable"
The goal with the above proposal is that once users have /release, we
can push forward with /dev and moving things from SFW/JDS/etc over to
the small oi-build that prestable currently builds from.
The work I was doing on /experimental was idealistic, and because Jon
kept updating /dev faster than I could keep up with it, /experimental
was pretty much constantly broken and not fit for purpose. It was a
giant leap too far, and unworkable.
If we push /release, then we can recommend users who want stability
switch to it, and we could, for example, aim at shipping /dev releases
every 4 to 6 weeks, as Jon has already been doing with the prestables.
Just to avoid any confusion, the /experimental repo I created was based
on the latest prestable at the time I generated it, plus an overlay of a
complete build of https://hg.openindiana.org/oi-build.
What I'm proposing is instead that we base future /dev releases on this:
https://hg.openindiana.org/sustaining/oi_151a/
where we slowly move things from the legacy consolidations, over to the
skeleton oi-build you can see here:
https://hg.openindiana.org/sustaining/oi_151a/oi-build/
Obviously we can source recipes from userland-gate and illumos-userland.
Why do it this way? Because it lets us take baby steps, push out regular
releases, and generate ISO images. It's not a big giant leap, and to a
large extent it's something Jon Tibble has kind of already been doing it.
Jon, what do you think of this idea? Is it workable?
Thanks,
Alasdair
_______________________________________________
oi-dev mailing list
[email protected]
http://openindiana.org/mailman/listinfo/oi-dev
