honton (07/13/82)
In reply to Greg Johnson's comments of July 6.
The overall 'syntax' to apl is of the form:
aexpr f aexpr or g f aexpr
depending on whether you use the dyadic or monadic form,
and the extension when providing an axis is:
aexpr f[axis] aexpr
Now the compression and related operators follow this form.
The inner (and outer products) can be considared of this form
if you look at f.g as a single function. (this is stretching
things a little.)
Indexing, on the other hand, is different. We could get
around its different syntax with an indexing operator. ie.:
aexpr indx aexpr
with indx being whatever symbol you feel appropiate. (How about
a semicolon?)
With this form we can only index along one axis at a time,
the default axis being the last one.
aexpr indx[axis] aexpr
And if we want to index a three-dimensional expression to get
a single element, we must use the indexing function three times.
This shows how convenient indexing really is.
I think that this probably shows that the current syntactic
form would remain, and this idea is probably not the basis of
some new language.
chas (CWRU)
(..decvax!cwruecmp!honton)