On Wed, 1 Dec 2004 23:13:21 -0500 (EST), Todd Vierling <tv@duh.org> wrote:
> On Wed, 1 Dec 2004, Sean Davis wrote:
> 
> > > > I still fail to see the point in releasing broken code.
> > >
> > > This is a long-since-ignored argument.  If broken code were never released,
> > > there would never be a release.
> >
> > I'm less concerned with the fact that a release might (will) have bugs
> > than the fact that it takes so long to get from the initial branching
> > to release - an every-two-years release cycle is very embarrassing,
> 
> [with introspective NetBSD hat on, paying homage to the Zucker brothers]
> 
> "We're familiar with it."
> 
> [with hat removed]
> 
> Basically, this is a really bad, known problem that releng has been trying
> to fix for some time.  Sure, there's an intellectual tug-of-war at times
> between people with different opinions on how "stable" the system is wrt
> releaseability.  Much of that had to do with the fact that a cyclic, formal
> releng process didn't really exist until 1.6.  Even then, the releng team
> itself didn't get much pull with the rest of the development team until well
> into the 2.0 branch.
> 
> 2.0 also had its own share of interesting technical problems on that front
> that would take forever to describe here, but I do believe that The NetBSD
> Foundation as a whole is hoping not to repeat that in the future, with 3.0
> and beyond.

Please understand that I do not - and did not - mean to sound
accusatory. I'm well aware that NetBSD doesn't have the manpower of
the other BSDs for releases and things, I was just voicing my views
regarding the extreme gap between releases (and the functionality gap
between -current and whatever the latest branch is).... I track
current for that reason. Sure I catch the occasional really obnoxious
bug, but it's usually fixed quickly. Hell, I can't remember a
show-stopper bug that affected me since the tl(4) bug back in...
1.5[A-Z]+ I think?

-- 
Sean