--- John Randall <[EMAIL PROTECTED]> wrote:

> A rude awakening comes to most J programmers who use primitives at
> small rank on large amounts of data and then try the same thing with
> user-defined verbs: the program grinds to a halt.  The difference is
> that the former frequently have integrated rank support while the
> latter do not, and there is a significant cost whenever a verb is
> started up.  I suspect this start-up cost is the killer when you try
> fine-grained recursion in J.

This is a very good observation. I believe to overcome this would
be possible if J could JIT-emit the machine code for the inside of the
loop so that the compound expression be calculated at the same
cost as a primitive expression.

The JIT-emission has been done for byte-code languages since SmallTalk,
http://strongtalk.org/ and J tacit expression could be seen as bytecode, 
so JIT might be possible.



 
____________________________________________________________________________________
The fish are biting. 
Get more visitors on your site using Yahoo! Search Marketing.
http://searchmarketing.yahoo.com/arp/sponsoredsearch_v2.php
----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to