Raul Miller <rauldmil...@gmail.com> wrote: > The aspects I like are the compiler and the esql. I am of the opinion > that having additional perspectives is nice.
Majorly nice, yes. > From a J perspective, I think that a compiler that handled a subset of > the language (even if it was only a rank 0 subset) would be a good > thing. This would extend the utility of the notation (albeit with > maybe a cost in elegance). Having a compiler that handles a subset of the language was attempted with PyPy. It paid off RIDICULOUSLY well. PyPy started its life as a toy that nobody would use (attempting to implement Python in Python); thanks to the restricted compiler (RPython) it became the fastest way to run serious Python programs, and is becoming a language-independent JIT platform, as well as a testbed for some seriously major advances in computer science (incremental real-time GC or software transactional memory). It might actually be easier to initially implement restrictedJ using PyPy's RPython rather than adding a compiler to J. (On the other hand, I tend to agree that it would be nice to provide some better turnkey application support. Using RPython wouldn't give that.) Hm, maybe instead of 'RJ' or 'RestrictedJ' it should be called 'JinTime'. > Raul -Wm ---------------------------------------------------------------------- For information about J forums see http://www.jsoftware.com/forums.htm
