Re: cc can't build 32-bit executables on amd64

I tend to get snippy towards the end of release cycles, and I apologize 
to those I've offended or have been needlessly rude to.  But please hear
me out on what I have to say here....

What makes a release cycle stressful is not the bug tracking and fixing,
or the building of bits, or anything like that.  It's that many people
seem to wait until the very last minute to test the release and speak
up with their issues.  I don't know if it's because people politely
assume that all of the issues are already known and are being worked on,
therefore their effort and their voice isn't needed, or if it's because
people are reluctant to test anything until others have worked the bugs
out for them.  Either way, whether I try to rush a release in 1 month
or let it draw out to 3 months, there is always the last minute push by
those who are concerned about a certain problem or a missing feature and
who desperately hope that something will be done about it.

First of all, each release is a step in the overall process.  While it's
always unfortunate and undesirable to have bugs in releases or have
missing features, it's even more undesirable to hold releases
indefinitely until "all the problems are solved".  It amounts to
throwing away all of the work has happened on the hopes that future work
might happen.  Releases need to happen, and they need to happen on a
regular basis.  If something is broken or missing in a release, then it
can be worked on in the next release.  6.1 is just a precursor to 6.2,
and 6.2 can't happen until 6.1 is released.

Second, we need more people testing early on in the release process.
The BETA builds are meant to give people a baseline of what the expected
functionality should be so that bugs can be found and fixed.  The RC's
are just meant to be a final polishing of cosmetic things.  Waiting
until RC1 or RC2 to test UFS quotas or ask why 32-bit compiling doesn't
work is a bit late.  If these things had been brought up 3 months ago,
or even 1 month ago, there is a much higher probability that they could
have been addressed.  If no one wants to do any serious testing (with
the notable exception of Kris Kenneway, Peter Holms, and a few others),
then I might as well just skip the majority of the release cycle and
just stamp out RC1 followed immediately by RELEASE.  The release cycle
doesn't work if people put off testing until others test for them.  So
please, when 6.2-BETA2 happens, everyone reading this email please give
it a spin, even if it's just in a QEMU or VMWare session (and it is
rediculously easy to test a CD with those, I do it all the time for my
day job).

The release process is about balancing the need to "get it done" with
the need to "get it as good as possible".  The time to identify and fix
problems is during the BETA phase, so please help us to make that
happen.  And if we get to the RC phase and something is broken or
unfinished, then please help us to get it resolved for the next release.

Thanks,

Scott