[Mono-list] Re: Java VM for .NET

[Miguel de Icaza]

> > At first your JVM-JITter creates IL-code, and then the .NET-VM creates
> > native code. I think, this needs a lot of time.
> 
> I can't see that it'll be drastically slow. Besides, if this is running for
> server side processes (where Java is used most), then JITting will be
> relatively minor as servers tend to stay running.

or compilation can be done once (.class to cil) and then reused.  I
think that this particular performance problem is not important at all. 
Besides, without actual performance tests, it is hard to predict.

> > But where is the advantage to compile Java-Binaries Just-In-Time to
> > .NET-Binaries?
> 
> JITting would be slower than converting ahead of time, but it also gives the
> advantage that there are no additional steps required to get it to work.
> Compile .java to .class, and drop the files into a native JVM, or into the
> JVM-to-.NET project, and either way, they just work.
> 
> I think it's a great idea, personally...

Yes.  That is exactly the kind of project I wanted to see happen (Gaurav
had looked at this, but he has been really busy lately).

Miguel