[SunRescue] Q on "optimal" OS for Sun4c machines, now that Solaris 8 won't run

[MikeHebeldruaga at pmail.net]

>>   For commercial OSes, the fundamental problem with supporting enormous
>>ranges of hardware is testing as much as anything else. Everything that's
>>supported has to be tested to some degree, and adding platforms can cause
>>testing work to grow exponentially. And systems of the correct type have
>>be to be available to developers and testers, which means they have to be
>>maintained... After a while, the temptation is almost overwhelming to pick
>>the platforms that the fewest customers care about and drop them, in order
>>to cut the development and testing efforts down to a more mangeable size.
>>Otherwise, any company will eventually be brought to a grinding halt,
>>anchored by the weight of supporting everything that it's ever done
before.

Testing is certainly an issue but if a company sees that the product is
still being viably used they should at least make an effort to support it.
Linux, NetBSD, etc. makes that old hardware viable and companies aren't
ready for that.  They've been working for so long on the premise that
"Anything beyond a certain age is completely useless" that they are having a
hard time accepting that people are still utilizing technology that is 10 -
15 years old.  They will eventually catch up but it's going to take a decent
overhaul of their structure to handle the new model.  For instance, if Sun
wanted to get old Sun users to buy new Suns then they should at least
partially support the old hardware.  (Releasing all data/utilities to the
public freely would be a good start.)  If they do that then they will almost
certainly have users that will buy new Sun hardware when they need to.  Sure
some will still buy old units, in fact many will, but others will just go
out and buy a brand new box.  And if Sun was "nice" or "helpful" then they
will be more likely to purchase a Sun box than a Dell or IBM or whatever.
Unfortunately for us most companies aren't ready to understand that yet.

>>   For free OSes, nobody gives a damn about testing, it seems. Do not,
>>repeat NOT, assume that just because something is listed as supported that
>>it will actually work. It's listed because it worked at least one time in
>>one configuration for one person somewhere. Unless it's a popular piece of
>>hardware (or the pet of a maintainer), odds are that it probably doesn't
>>anymore. I've been burned too many times to have any faith in the
>>supported-hardware lists.

Hmm... I seem to see just the opposite.  Because the open source developers
pool is so large they seem to run on more kinds of hardware than you can
count.  (When was the last time you saw Arcnet cards?  They're out there.  I
know people are using them.  I even have some, abeit not in use but they
work.)  The point is this: the open source developers want to be better in
the long run than Windows or xOS or whatever.  They know that to that they
have to be far more careful than M$ is because they know that their software
needs to run on as many things as possible and and run _better_ than other
things.

I'm not arguing that there aren't problems with the open source OS platforms
but what I will argue is that the compatability issues and "hardware
workability" issues are far less frequent than you might think.

Sincerely,

Mike Hebel