I voted "release on time", because the delay between woody and sarge
was a kind of traumatic experience at work or with sysadmin colleagues.
It is OK to "release when it's ready", but IMHO with a max time between
releases. 1.5 years seems OK for me. I have to say that backports.org is
of great help to bear waiting for the next stable release, especially for
the desktop use. And no, using testing for desktop use in production is
not an option
I can remember how painful it was to install the old woody kernel
on a fairly recent server with hardware RAID. The aim is to use Debian in
production. It is OK to spend few hours to install Debian, but not days.
Bosses will just say "go ahead with Fedora/Ubuntu/etc."
I would have vote "Support hardware that requires sourceless firmware"
for second choice. While I support strongly Debian's (and others, like
OpenBSD) activism against sourceless firmware, I think we should
find a balance between philosophy and pragmatism (sorry RMS). Let's
face it: OS is ony useful if it runs on your hardware. Geeks now how
to choose hardware to match FLOSS OS, average joe not. And in
production environment, you don't have always the choice of server
hardware. When I try to convince friends/family to try Debian on
their personnal Windows PC, if the hardware (e. g. network interface,
wireless or not) is not recognized straight ahead, the experience is
over. Ubuntu is a popular choice around me because
it installs flawlessly most of the time, espacially on laptops.
The best could be a message from the installer like "OK, the installer
detected some hardware that needs sourceless firmware. Philosophically,
the Debian project dislike blabla. Would you llike to continue with or
without those firmware". Then, average joe can be sensibilized to a
problem he/she probably never heard before.
Anyway, thanks for the great work!