Your "simple adverb" upd is almost identical to the solution
([:> [: f&.>/\.&.|. <@],<"[EMAIL PROTECTED])
I suggested. I would have appreciated if you have mentioned this.
"It uses ideas from several of the contributions" seems rather inadequate
to
me.
R.E. Boss
Thanks for your note. I value the time that everyone spent on this. I
agree upd has many elements similar to your form and I am pleased to
acknowledge that but the Quintana adverb has most of the structure there
too. For me the main message has been 'treat boxing even more seriously'.
Nearly all objects in my J code are boxed because I want to keep relevant
nouns together and want a whole class of verbs to operate on them. The J
structure of using suffix scan proves a very clean computationally efficient
method for this updating problem for such objects. The seq and upd adverbs
provide simple tools to help write informative code exploiting that
structure.
On my tests your form is about 30 percent slower than upd and seq but I have
not explored the reasons for this. All three are much faster than recf and
that gain is worth while.
My post has not appeared in the programming forum messages I have received
so your message also confirms one thing I wondered about - whether it had
been received at all.
Thanks,
Fraser Jackson
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm
