Just following up on a few recent postings on Solaris/x86:
A driver for the Adaptec 6260/6360 chipset, including its instantiation on
the Adaptec 1522 controller, is in the works, and should be available within
a few months at most.
Sun CD-ROM drives work with Solaris/x86 2.1.
Solaris 2.3.2 is indeed the merged SPARC/x86 release, and the two will
stay in lock-step thereafter. Early Access release is in a few weeks.
This SunSoft site (Los Angeles) is the best anecdotal evidence for the stability
of the x86 product. We ran our development operation on Solaris/x86 2.1
pre-releases for months, ran it on the official Solaris/x86 2.1 release for
over a half-year, and are currently running it on Solaris 2.3.2 internal
releases.
This article is being written on Solaris 2.3.2 on a clone-quality 486/33 that
has been on my desktop for over a year running the releases listed above. My
machine *never* panics, hangs, etc. Only power outages, earthquakes, etc.
cause it to reboot. Admittedly, I am a manager rather than a developer and
thus use email, Frame, Wabi/Excel, and internal Sun graphical apps more than
I use development tools, modems, etc. But this organization as a whole does
use development tools *EXTENSIVELY* and *SUCCESSFULLY* and *EXCLUSIVELY*
on Solaris/x86.
There was a thread on binary compatibility with UnixWare. All Sun can do
is follow published standards, and we have done so. Solaris/x86 2.1 conforms
to the Revised Edition of the iABI, which incorporates the revised edition
of the generic SVR4 ABI. Some of the problems mentioned in that thread, such
as the dynamic linker name and the gettimeofday() problem, are simply
differing interpretations of the standard. In the gettimeofday() example, it
is pretty clear that UnixWare was very marginal in its conformance, as it
included this function in a statically bound library when it should have been
dynamic. In both these cases, much to our chagrin, marketing pressures forced
us to "emulate" UnixWare by putting in hacks to fix these problems. We expect
these pressures to subside extensively as ISV's continue to do native
Solaris/x86 ports (which are just recompiles from Solaris/SPARC). But in
general, all we can do is follow the standards and work with standards bodies
to fix the grey areas and add extensions for important new API's, both of
which we are doing.
My condolences to the fellow who found out the Solaris/x86 2.1 Answerbook
is $495, I was kind of surprised at the price myself when I verified it.
But remember, in the corporate environments which comprise the main target
market, there will be perhaps 1 or 2 AB's bought for every 100 desktops
running Solaris. Given that this is essentially a low-volume product,
the price-point makes a little more sense.
**TURN ON HYPE-DETECTOR, STOP READING IF YOU WANTED FACTS AND NOT OPINIONS!!**
Make no mistake about it, SunSoft is committed to Solaris on x86 (IMHO;
I am not writing this as a representative of Sun, see dislaimer). Since
releasing 2.1 last May, dozens of patches have been produced, many new drivers
have been written and released on update diskettes, and extensive work has
been done with OEM's to ensure proper support for their hardware and with ISV's
to attract applications.
There are hundreds of people at SunSoft whose sole job is
engineering, marketing, and selling the x86 product. Our interest in its
success and our commitment to it is far greater than any of the subscribers
to this newsgroup, as our livelhood depends on it. We're not stupid. The
product will continue to improve -- although the technical quality is already
high, it will get higher; the supported hardware list will grow extensively;
the number of applications available will mushroom; and the product will
become *the* Intel UNIX implementation in time! You heard it here first... :-)
Abe Ellenberg
Manager, OS Engineering
SunSoft, Inc.
DISCLAIMER: Although I am a manager at SunSoft, this article does *NOT*
necessarily express the views, policies, or product plans of SunSoft --
you'll have to go to Ed Zander for those, or at least someone in marketing!
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:08:55 CDT