[fab] looking at our surrent state a bit

[Jeremy Katz]

On Fri, 2006-11-03 at 15:16 -0800, Karsten Wade wrote:
> On Fri, 2006-11-03 at 12:52 -0500, Jesse Keating wrote:
> > On Friday 03 November 2006 12:40, Thorsten Leemhuis wrote:
> > > Late march only once. I asked after FC5 on fedora-devel if we would
> > > stick to "late march and late September" in the future again (+ some
> > > delays if needed) and the answer was "no".
> > >
> > > But if we get our release to "round about late march and late September"
> > > in the future I'd be very glad.
> > 
> > It's something we shoot for, but flexibility is necessary for what we do.
> 
> Looking at the trending from Thorsten's list:
[snip specific dates]
> The nine months for FC5 brought the schedule back into the
> late-March/late-October flow again.  If this is important, I think we
> all have some possibility of influencing that this stays this way; and I
> think some influence may already be applied.  

I tend to agree that there's a lot of value in trying to align with
"something" like this... and we've been spastic at times as to what the
"something" we're aligning with is.  gcc has been the driver once, xorg
once, GNOME a few times... I think that GNOME might well be the
reasonable thing in general, but I do want to try to give us time to
reasonably integrate the GNOME release; the relatively identical
timeframes for FC5 was a bit painful from my POV.  Trying to align the
freeze for ~ 2 weeks after the GNOME release gives time just to make
sure that things are in good shape and have a little bit more of a)
hititng the confidence and b) getting all the components updated and
well tested.  So rather than mid-September/March, I'd try to aim for
freeze of early April/October.  Which, funny enough, is what we were
shooting for with FC6 :-)

While I somewhat share Jesse's concern about a slip of one release
short-changing the next, I think that the predictability might make it
worthwhile.  Especially since, realistically, 5 months vs 6 doesn't make
a huge difference.  

A bigger win is looking at trying to get more frequent and earlier
testing... but that's a topic for another thread.

Jeremy