Re: C# for java

>>> As there are certain constructs in C# (local objects stored on
>>> stack, listeners as integrated part of the language, etc.) that are
>>> not supported by the JVM, I would say this is impossible.
>>
>> By the same approach, having scheme run on jvm (no call/cc or
>> similar, etc) should be impossible aswell, no? One can for example
>> always emulate the presence of a stack, and merely put objects that
>> need to be removed when stack unwinds on it.
>
> It's been done, not just for Scheme, but for many other languages, too
> (see <http://grunge.cs.tu-berlin.de/~tolk/vmlanguages.html>). I cannot
> vouch for the quality or usability of the various code generators
> produced by all these projects, but the sheer number suggests it's
> pretty feasible for a wide variety of languages.


Yes, I know - as there is in fact several implementations of scheme that
run on JVM, I was bringing it up as a counterpoint to his claims, as scheme
is also a 'very much not java' language, really even more so than C#.


> There's an obvious motivation to do this: The extensive work that goes
> into improving the performance of the JVM and its portability gives a
> considerable advantage to the compiler writer.
>
> Google for "Scheme JVM" and you'll get plenty of hits, including the
> page I referenced above.


I know. I use one of them (SISC) almost every day.


Sander


+++ Out of cheese error +++

Return to the comp.compilers page.
Search the comp.compilers archives again.