> On 31/05/07, Girts Zeltins <[EMAIL PROTECTED]>
> wrote:
> 
> > If Sun Microsystems will go this road then Solaris
> can become leader operating systems.
> 
> Windows does ok. How many does it run on?

Didn't it used to run on the Alpha for awhile?
Maybe even the egos at MS at least once understood that
a port can be a valuable exercise, even if it doesn't seem to
pay of right away.  (but there are limits to that, see below)

Back around early Solaris 9 when Sun was talking about dropping
support for x86, I know I said more than once (and can't have been
the only one!) that a port is an insurance policy, particularly when
running on a minority CPU architecture.

I think there are benefits of having at least 2 or 3 ports.  Insurance
against changes in which architectures remain competitive; and
deeper understanding of and safety handling various issues like byte
order.  And once the source is already set up to support multiple
architectures (as Solaris is), adding more is at least not as difficult as
going from 1 to 2 must have been.

> PPC would be good, but a lot of NetBSDs platforms are
> relatively
> unused. I don't see the
> benefit of porting Solaris to dead architectures.

Right, it's still far from easy, and the one-time work is only part of it;
it's the support that's a killer.  For anything that Sun couldn't find a way
to profit from, likely some other organization would have to make
some sort of commitment to support the architecture for X amount of time;
otherwise why put up with the clutter in the source tree for the additional
architecture?  Of course someone could still maintain their own as a
separate set of files and patches, but that would be that much more work
for them, and hardly collaborative either.

So a few ports are good, but a larger number requires some sort of
economic justification (how to make $$ from doing it and supporting it),
since programmers have to eat too (can't all be doing it just out of goodwill
in addition to a day job).

Embedded devices are an interesting case; they pretty much imply
a separate, minimized, and perhaps somewhat re-tuned distribution,
as well as special drivers and such.  I wonder for instance if it would
be practical to think in terms of Solaris on an ARM, since the Solaris
kernel probably makes a lot of design choices that assume the availability
of at least a few hundred MB of RAM, for efficiency's sake.  I do have some
old boxes that ran Solaris 7 or earlier ok on 128MB or less, but newer
versions tend to be sized for newer hardware (of course).  The peak
minimum seems to be the installer, which with CD or DVD media
would clearly have a range of speed-vs-RAM tradeoffs it could make;
but it would be great if it could make them more dynamically, although
I don't know that I'd expect that to do more than get one or two more
versions worth of life out of dinky old boxes.
 
 
This message posted from opensolaris.org
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to