Parallel Object Oriented Programming Language

[cx] writes "There is one here at a U-Virginia college site. "

Parallel Object Oriented Programming?

[|<amikaze]

POOP?

Well OK...

[azephrahel]

It looks pretty neat, BUT...

Paralell programming in C++ objects isn't really hard (and paralell programming with Java is even easier). And from their page, this really is just a modified c++ (remember c++ is the language, not the back end. How they impilmented it might be different..but maybe not).
So other than a neat project, is it really worth it? Probably not. Its an extention to a language to handle problems the language can already handle.
Now a rules based extention, that would be neat! Err say, didn't slashdot run an article on a r++ or something like that compiler a few years back? Any updates?

Is Mentat dead?

[BigTimOBrien]

The professor involved in Mentat looks to have either left UVA or is no longer focusing on Mentat as a viable tool. I believe that Mentat morphed into Legion, and that there is now a project called Centurion which is a proving ground for Legion.

The real question that would be interesting to answer is this: how has the emerging interest in peer to peer technology changed the approaches to massively parallel computation? or has it changed the approach at all?

When I tried to use Mentat initially in 1997 it was a forced fit. There was a piece of legacy Fortran 90 code about 100,000 lines long and Mentat was used purely as a way to spawn simulations in parallel over a net of 30 computers, it was a horrifying thing to learn to use, and I was always wondering why we didn't just rewrite the thing in C++ and use something more standard like PVM. I'm not saying that Mentat/Legion is a bad idea, it is just that for the application it didn't fit at all.

The pages for Legion look very out of date, has the Legion project hit a funding problem? Is Legion dead? or has it morphed into something entirely different?