Arthur:
Putting together SRFI-122 and then dealing with the (relatively few)
comments was really a learning experience for me.
I can't finalize this to my satisfaction within the 90 days allowed.
It's clear that I have a concrete vision of how arrays should work that
is different from the lower-level views of previous proposals, but I
have not been terribly effective in getting this vision across.
Additionally, my initial proposal had a bunch of bells and whistles that
I thought people would like, but which added complexity that nobody
seemed to appreciate.
Finally, I should add {interval|array}-permute procedures and maybe
remove the {interval|array}-distinguish-one-axis procedures. The latter
are important in practice, but the *-permute procedures are more
fundamental (*-distinguish-one-axis is the composition of *-permute and
*-curry) and of interest in their own rights. E.g., various views of a
CT scan are set up via *-permute + *-curry.
This is not a question of adding new routines, I think I need to have a
different set of base routines to have a coherent initial SRFI.
So I currently plan to withdraw the SRFI for now and resubmit later.
Do you have any thoughts or comments before I do so?
Thanks.
Brad
PS: Decided to CC this to SRFI-122 mail list in case anyone else has
comments.