-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

David Brown wrote:
> On Wed, Jan 02, 2008 at 08:06:21PM -0800, Christopher Smith wrote:
>> Not handy. Miguel said that CLR's support for struct's and generic types
>> (and a more sophisticated type system in general) provides it with some
>> performance advantages, but if you stick to using capabilities that both
>> can use, the JDK JVM was more aggressive in its optimizations.
> 
> That sounds more like an implementation/maturity issue.  More effort put
> into CLR runtimes could make them better.

Yes, that's what I meant when I said, "It's a better architecture. The
JVM's are still a better implementation."

The CLR guys have a lot of experience writing JIT's, and a lot of
compiler experience going back even further.  I think the reality is
that Sun has been able to put more effort from some of the best experts
in the world on JIT's. It is entirely possible this will continue to
provide them with an insurmountable lead.

In the end though, none of that matters. When I look at a JIT for a new
technology I develop, I want three things: no bugs, easy to build my
technology with, and fast as humanly possible. The JVM wins on two out
of three (I'm sure the CLR's generics support made that aspect of Cobra
easier to work on). Not that I disagree with Chuck's decision to target
the CLR (I think for a first run "easy to build" tops the others), but I
suspect a JVM implementation of Cobra would prove to be useful.

- --Chris
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.6 (GNU/Linux)
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFHfR1mOagjPOywMBARAobvAJ95IYXVEKDBVYisb6ZIHGpsYLob4ACdGyml
7kgrfgN5sfjMa69v2fPD6Gw=
=dCX6
-----END PGP SIGNATURE-----

-- 
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg

Reply via email to