[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
- To: (BUG LISPM) at MIT-AI
- From: RMS@MIT-AI
- Date: Wed ,21 Jun 79 13:21:29 EDT
I have made TYPEP of two args work, and made TYPEP of one arg
return useful things for all reasonable data types.
However, before I publish it, a couple of things might want to be changed.
First of all, TYPEP of two args expects the type second
so that thefirst arg can always be the random object.
But SIGNP takes its operation first. I think it would be better
for TYPEP to take the type first. Then the arglist would not be
very explanatory of how to use it with one arg. So maybe TYPEP
of one arg should be just for compatibility, and a new name found
for the one-arg case (it could get rid of the "P"). Do you think
that "TYPE" would be bad? Or what about "DATA-TYPE" (which right
now returns a symbol starting with DTP- and probably is never used).
Anyway, it all works, but I'm suspicious of the way colons are used in it.
How much do colons in QFCTNS work? What happens if a symbol which is
in GLOBAL appears in QFCTNS with a colon (say, it is being used as a keyword
by someone who doesn't want to depend on the fact that it's also a function
in global, or in KWDPKG)? Does this screw the world? Or is that only
if it's in the cold-load? The old definition of TYPEP had some things
of this sort; some may still be there.