Warner Losh | 17 Jan 22:09 2012

Re: FreeBSD has serious problems with focus, longevity, and lifecycle

On Jan 17, 2012, at 11:12 AM, John Kozubik wrote:
> Again, I'm not suggesting more snapshots - I am suggesting more real, bona fide releases.  This will help people.

I tend to agree with you.  Our release engineering process isn't serving the needs of users as much as it once
did.  When Walnut Creek was running release engineering, we had releases often because they wanted to make
money from their subscriptions.  This produced reasonably spaced minor releases and except for 4-5,
decently spaced major releases.  Even after the torch passed from walnut creek to others, there was still
either residual pressures to make the releases happen, or inherited mindset that keep on the same pace.

Today we have lost our way.  We have no major vendor pushing the process along to make it happen faster.  We have
no reason to get things done faster or differently than the volunteers are doing it.  So we're languishing. 
9.0 took forever to get out, and we didn't do stop-gap 8.x releases.  Our port collection has also gotten
bigger since those by-gone days so doing a release of the whole ports tree is taking longer to QA, so
pressure to do it more often meets up with resource constraints.  Our binary update tools lag considerably
compared to Linux, and there's a big reluctance to whole-heartedly embrace PBI as a possible solution. 
Maybe pkgng will help there.  Maybe the various attempts to get ABI stability to allow for easier
decoupling of FreeBSD base and FreeBSD ports releases.

But we have a real problem here.  One I don't have easy answers for how to solve.  One that likely has many other
root-causes than the few I've cherry picked for this reply.  The underlying balances that allowed the
early project to succeed have shifted, but we've not shifted with them.


freebsd-hackers <at> freebsd.org mailing list
To unsubscribe, send any mail to "freebsd-hackers-unsubscribe <at> freebsd.org"