On 05/02/2016 01:24 PM, Stephen Gallagher wrote:
There is strong engineering value in having two releases per year:
release
early, release often. There are many projects that develop through Fedora that
get thrown into disarray when our cycle gets too far out of whack (prominent
examples being GNOME and glibc)
The distribution is made out of 14k+ components most of which of those
components completely out of sync with the Gnome and Glibc one which
means components that make up different products are not synced with
those two components cycle, which means that you cannot have one
product's release cycle be synced with another or bound to it's release
cycle.
Even historically QA did not manage to deal with this release cycle when
distribution was smaller and when it was just one "generic" release but
despite all the evidence of this not working through the years you still
push this one forward which means in other words Red Hat wants the
distribution cycle to be forcefully synced with Gnome and does what it
takes to do so which is why it's back on Gnome's release cycle despite
everything indicating it should not be on that cycle at the cost of the
community and quality of the distribution.
If there is genuine interest of start releasing fedora on time you will
not achieve that goal by not doing or blocking mass rebuilds, you either
need to stabilize anaconda development earlier in the cycle or find
another installer for the distribution that can exist outside the
distribution release cycles and does not have to be rewritten like
people are getting paid for it every cycle.
JBG