> Simon sez:
> > but I don't think anyone's yet ready to use it "for real".
> Work here, being complex, has a dominantly imaginary component, but the Lab's
> Computing Division (er, I guess that would include me) has just added Linux to its
> list of officially supported OSes. This is more of a reaction than a leadership
> position, as there are several groups already putting together compute farms of
> many PCs running Linux for physics analysis -- collider physics is perfect for the
> Beowulfian coarse-grained parallel approach, with enormous numbers of modest
> events having no (known) interactive effects, much like FoRKmail.
> Many physicists here are moving from unix workstations to dual-boot Linux/NT PCs
> for their desktops. Cost is the main driver, but cheap games probably doesn't
> hurt the cause any. And even though the FP performance of the Intel parts can't
> touch an Alpha, the bang for the buck is still enticing.
We use linux every day at my work. The programmers have dualboot NT/Linux
workstations. I personally leave mine on Linux unless I'm doing something NT
specific. I just find it a lot easier to work remotely on other unix boxes for web
stuff on Linux, the NT telnet client seems particually retarded. Our main webserver
is solaris, but all of our other servers, mail, news, development webserver boxes are
all linux. I personally think its the best unix available for the intel platorm.
NetBSD/FreeBSD are good, but I prefer linux. I couldn't be paid enough to run
anything on ANY SCO product. I think one of the best side effects of the rise of
Linux and the BSD unixes is SCO going on the decline. I haven't had much experience
with intel solaris.