Re: a big limit of mathematica?

By "rewrite" I mean *a* kernel for computational mathematics,
and nothing else. Call it a toolbox if you want.
Sort of "reverse Matlab". In Matlab symbolic services f
orm a separate toolbox bridge Here the
computational kernel should be a largely a bridge to public
domain libraries, of which there are several excellent ones.

The symbolic kernel should be separate from the computational
one w/each interacting with the front end. One Cell property
would be the appropriate kernel it will execute.

Interweaving high performance numerics and symbolics in the
same pot risks the danger of "bug instability": fixing a bug
introduces 10 more. Beside the appropriate data structures are
different. Uncontrollable complexity has bothered many projects:
eventually the whole programming corps is busy just chasing bugs.

.

Quantcast