FreeBSD has serious problems with focus, longevity, and lifecycle

[Mark Blackman]

On 27 Jan 2012, at 03:26, Mark Linimon wrote:

> On Thu, Jan 26, 2012 at 10:52:44PM +0000, Mark Blackman wrote:
>> I suspect poor old RE is putting too much work into BETAs and RCs for
>> point releases. 
> 
> The counter-argument is that we have a lot more leeway to make mistakes
> on a .0 release.  We're not going to be cut any slack at all for shipping
> a badly regressed point release.
> 
> Some minor regressions are inevitable in software, but they do indeed
> need to be minor.
> 
> For how we're doing with regressions in general, see:
> 
>  http://people.freebsd.org/~linimon/studies/prs/prs_for_tag_regression.html
> 
> Now, it's true that many of the recent PRs are against 9.0, and many of
> the ones that aren't may be stale (certainly most of the pre-2010 ones),
> but these are the types of things that users really notice and become
> unhappy about.

All good points, although I'd guess there's some diminishing returns argument
for progressive RCs/BETAs, however probably only the RE team have a good feeling
for the sweet spot.

- Mark