901 and calculus primitives (D. et. al.):

If you want D. , and it completely meets your requirements, then continue
to use 807.

If you want D. type facilities, but were frustrated by its limitations,
bugs, and opaqueness (C code), then use the new 901 facilities. The
implementation is all in J and is amenable to study and improvements. It
may take a bit of time for this to filter through to common practice in the
J community, but 901 takes the hard step. This is a good thing!

The C for these primitives was a lot of complicated code. That code was a
lot of baggage to drag forever forward to new releases for the benefit of a
few users. Users who will, in fact, be much better served if they move to
the new framework.

Compatibility is a good thing. But it can be carried to far. If you want J
frozen circa 401, run 401.

All published code (regardless of language or platform) implicitly or
explicity is tied to the fact that it runs unchanged on only certain
versions. No code of any interest will run on all back and future versions.
It is best when code has a statement like: runs on j807, won't run on
earlier version, may or may not run on later versions. The pacman addons
has some mechanism for this.

A major J release (8 to 9) is when we introduce non-compatible changes.
J901 is a big step forward and positions for the rest of the 90x series. If
you want to be a pioneer, move to j901 and if D. is important to you,
become part of the community effort to see that the replacement rocks. If
you want to stay a bit behind the curve, j807 is solid and will be there
for a long time.
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to