On 13 January 2015 at 15:39, Matthew Miller <mattdm(a)fedoraproject.org>
wrote:
On Tue, Jan 13, 2015 at 12:24:53PM -0600, Dennis Gilmore wrote:
> > > > The reason for "tentative" was mostly so we could get
feedback
> > > > from rel-eng on whether this is actually feasible.
> It is not even close to feasible.
[...]
> It is still not reasonable to get everything done in that period of
> time.
Okay, so, in an ideal world, what would it take to bring the mass
rebuild time down? This schedule is, nominally, 5 months and 10 days
from the F21 release, and if we consider that to include a built-in one
week slip buffer for each of alpha, beta, and final, it's exactly 6
months — our theoretical target.
If we have bottlenecks that make it impossible to construct this kind
of six-month schedule, we should at least eye what it would take to
reduce them. Is this a hardware problem? Storage speed? People power?
You forgot "too many packages?" There are 15842 packages in Fedora 21
and
16230 in Rawhide. That is a lot of packages that have to be rebuilt
possibly multiple times due to FTBFS, multiple architectures, etc.
2.5 weeks is 25200 minutes. That means a mass rebuild is doing 0.6 packages
a minute across 3 architectures. That is pretty darn fast
--
Stephen J Smoogen.