Why MS-extensions?

[Kit McCormick]

On Tue, 15 Jun 1999, Dirk Grunwald wrote:
>TW> Essentially we wanted to provide a true Open Source Java virtual
>TW> machine and class libraries which would support the two versions
>TW> of Java we currently have to deal with - one is Sun's Java, the
>TW> other, whether you like it or not, is Microsoft's Java.  By
>--
>
>At PLDI, Steve Lucco of Microsoft Research demonstrated a non-java JIT
>running a version of Word. He said that the JIT can translate to
>native code at 10-12 MByte/s -- Word was jit'd in 1/4 of a
>second. It's faster to JIT than to page. Earlier papers from Choulsa
>software (concerning the omniware product) demonstrated performance
>within 8% of an optimized application for a particular host across
>a set of applications.

ummm... 
that's because a large portion of such software is actually executing within
the host's object model, ie your talking about JITing relatively thin wrappers
around COM objects.  fast or not, it dang sure isn't cross-platform or scalable.


>Reading
>
>http://www.microsoft.com/PressPass/press/1999/Jun99/InnovationPR.htm

No thanks; it's probably just more spewtum.

>(particularly the portion about "Common Executable Format") may lead
>you to believe that the Linux/Java community has more important things
>to worry about than whether TVT has sold out the public Java effort
>(which, frankly, I doubt).
>Dirk Grunwald

of course.  this whole excercise is designed to make the linux/java community
anxious.  and well done, i must say.  even if the whole thing dies down after
JavaOne, MS will have gotten their money's worth (if, in fact, they have spent
any money on transvirtual).

As for TVT selling out, i hope you are right.  I really do.

kit