[Nottingham] Response on awareness day

[Graeme Fowler]

On Thu, 2005-11-03 at 20:28 +0000, Matthew Sackman wrote:
> That doesn't make complete sense tbh - most good distributions offer a
> fairly effective upgrade mechanism which should take care of that, and
> let's face it, you have to do that with Windows anyway within 0.3s if
> you want to have any chance of not being hacked...

Updates (as distinct to upgrades) yes, but I think what Andy's getting
at is the preponderence of actual releases of distributions.

OK, so Debian are world renowned for _not_ releasing a complete new
STABLE distro until world+dog agree 100% (which rarely happens) so the
release schedule for that particular distro feels sluggish, to say the
least [yes, I'm wearing asbestos pants]. But they cheat by spooging
oodles of new stuff into UNSTABLE and TESTING which, well, just work.
And Apt is a perfect way of pulling them in and removing them again.

However - Mandr*, SuSE, RedHat, Yellowdog, Lin(dows|spire), hell even
LinuxPPC.org as was - all suffer from "release panic" IMO and shovel (or
have shovelled) complete new distributions out between them at a truly
amazing rate. If I were a complete newbie now, I'd be terrified at the
number of distributions ("the great thing about standards being that
there's so many of them to choose from") and the number of releases of
them. How to compare DongleLinux 1.2 to BangleLinux 5, against a
background of umpteen 9.x and Core 4 and Potato releases; it must be
just about bloody impossible.

Any unit shifter (in terms of hardware) has a pretty hard time IMO
specifying which distro they want to install, as their lead time
probably eclipses the devel/test/release cycle of your average distro by
a factor of two. Imagine turning to a specific distro and committing to
purchase (or simply provide) and install (say) 10k units in 2006 Q1,
only to find that mid Q1 they rev up and turn another one out. Who wants
your hardware then?

It's a minefield ;-)

Graeme