[PC-BSD Testing] Rolling release criticism

Claudio L. claudio at hpgcc3.org
Wed Dec 18 04:13:00 PST 2013


On 12/18/2013 01:15, Mike Barnard wrote:
>
> Claudio,
>
> The idea of STABLE reflects a picture of a pre-tested, clean, few if 
> not zero vulnerabilities. RELEASE speaks of just that, tested up to a 
> point we are comfortable releasing for users to play around with. Just 
> thinking, what if STABLE remained as thus, stable with the packages 
> updated to this branch every two months as you suggest. Any one 
> picking the stable version knows they are picking production quality 
> PCBSD. RELEASE then remains where packages are shipped in as fast as 
> you can and this runs as the semi-stable platform for near cutting 
> edge technology. This then leaves CURRENT with all the experimental 
> stuff, and bleeding edge technology.
>
> Follow the FreeBSD tree. Whoever wants a stable production system goes 
> for STABLE, if you want to test new packages, run RELEASE and if you 
> want get into the bleeding edge, take CURRENT
>

Either I got it backwards or you did. I think the name is a bit 
misleading, but I understood RELEASE is more stable than STABLE.
STABLE gives you the mental picture of production quality, but in 
reality it means stable as in "pushed out as soon as it stopped 
crashing", versus CURRENT, which is "still might crash".
The STABLE branch is a development one, while RELEASE is the production 
quality one. PCBSD being a rolling release works more like STABLE, 
getting new packages as they come (but the base system tracks RELEASE on 
freebsd, to increase confusion).
In fact it's so confusing, we should forget about STABLE and RELEASE 
names. The idea is to have one PCBSD "EDGE" and one PCBSD "PRODUCTION" 
branches. Or something like that.

Claudio





More information about the Testing mailing list