Re: [gentoo-dev] [RFC] Future of gentoo's stable and unstable trees: what are your thoughts?

[Sergei Trofimovich <slyfox@gentoo.org>]

[-- Attachment #1: Type: text/plain, Size: 2309 bytes --]

On Mon, 24 Jul 2017 23:22:44 +0000
Peter Stuge <peter@stuge.se> wrote:

> Thank you for working on this.
> 
> Sergei Trofimovich wrote:
> > Can this proposal make a difference and make gentoo better and
> > easier to work with?
> > 
> > Does it try to attack the right thing?
> > 
> > Does it completely miss the point?  
> 
> I hold a perhaps radical view: I would like to simply remove stable.
> 
> I continue to feel that maintaining two worlds (stable+unstable)
> carries with it an unneccessary cost.
> 
> Based solely on how excellently unstable (and similar approaches before
> using Gentoo) works for me in practice, I believe that skipping stable
> and instead focusing efforts on resolving problems reported in unstable
> a little quicker would yield a much better end result - and would net
> positive dev time.

Good point.

Stable is used by Gentoo to guard against wide-spread bugs sneaking
into everyone's systems: SIGSEGVing bootloaders (hard to recover),
crashing at startup browsers (hard to find a safe point to rollback),
hosed toolchains (hard to diagnose in time), widespread build breakages
due to incompatible API (or ABI) changes upstream (hard to recover).

It takes time to identify and devise mitigation for new issues. What
would be the mitigation mechanisims for those when we know something
is broken? Currently we say STABLE should work better as packages
there had wider and longer testing.

Why would removing stable speed things up?
Due to smaller amount of bugs to deal with?

Do you think Gentoo needs KEYWORDS at all?

Should packages be tracked as "seemingly working" on the arch
or package.mask is enough?

> > Does it sound fun?  
> 
> Sorry, no, not to me. It sounds like "double" overhead. :\
> 
> 
> I consider dev time a precious resource. Devs should need to do as
> few things as possible, to keep things going, and should be able to
> immediately reuse as much input from the wider community as possible.
> 
> More troubleshooting and fixing "hard" problems, less routine work.

Can you clarify what exactly you see currently as a routine work
on the dev side WRT stable?

Fixing bugs for non-latest packages?
Tracking manually lists for stabilization?
Something else?

Thanks!

-- 

  Sergei

[-- Attachment #2: Цифровая подпись OpenPGP --]
[-- Type: application/pgp-signature, Size: 195 bytes --]