Fedora 11 schedule proposal

[Jeremy Katz]

On Wed, 2008-11-12 at 10:03 -0900, Jeff Spaleta wrote:
> On Wed, Nov 12, 2008 at 9:46 AM, Jeremy Katz <katzj at redhat.com> wrote:
> > Otherwise, we're really saying that the release date of release n+1 is
> > release n's date + six months (which is what we used to do).  And the
> > problem with that is that any slips over time meaning that we end up
> > running into major holidays for a release date and have to have a
> > long/short cycle to reset things.
> 
> Does it make sense to ask the following questions?
> 
> Historically, how much do we tend to slip per release?  Disregarding
> the month during F10 run up specifically during the infrastructure
> rebuild.  Are we somewhat consistent with regard to slippage? Is there
> an average slippage? The sample size is pretty small so I admit its
> not a statically valid measure.   What's the most we've slipped in the
> past, that is not the direct result of significant system downtime?

It's not an unreasonable question to ask.  It's a little harder to
actually get the data since we change our schedule pages.  I think John
Poelstra had tried to do some data mining leading to the proposal at
hand, but I'm not finding where he had stuck that off-hand.  I can try
to dig through some mail archives if I'm mistaken

My off-the-cuff response is that recent slippage has been on the order
of two to three weeks, outliers of 4-5 but most of those have had
extenuating circumstances.

> Accounting for historical slippage, were does F12 release land? Mid Nov?

If we do two to three weeks for F11 from May 28th, then we get the
middle of June.  If we add that to the shortened F12 schedule (since
shortening it more seems unreasonable, at least to me), it is scheduled
for mid November.  And any slippage putting us into the
Thanksgiving/Christmas quagmire.

Jeremy