[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: OK, how about...




Alan,

> One thing I've wanted that the proposal doesn't cover is the ability to
> reorder the arguments.


There are cases where it is indeed useful to permute the
arguments of a procedure with a simple notation.
Yet, I am convinced that the chance of confusion
is considerably higher with numbered variables than with
properly named variables and this is not worth it.

That is the primary reason why I have chosen not to include
this feature into the SRFI. If you need to permute the arguments,
write down a lambda-_expression_.

There are some secondary reasons, too. In the first place,
one has to limit the number of possible arguments by an
argument of the "640K should be enough for anybody"-type,
which I generally distaste. Limiting to 30 arguments might
not seem severe, but if you deal with computer-generated
code it can easily become a limit. There is no way out (I
tried) but if you start introducing (<> 1) (<> 2) .. or
the like then you are loosing the real benefit (a compact
notation for specialization) very quickly.

That is why I have a preference for not introducing
numbered arguments.

Sebastian.