Roger wrote:
>  In the implementation, it was (is) a useful fiction to assume
>  that memory is unlimited, because then the (implementation)
>  code can be a lot simpler.  Similarly, it was (is) a useful
>  fiction that there is no stack limit.

Yes, but the sad thing is we often have enough room to make the former fiction 
believable (useful).  Not so with the latter, in my
experience.

>  Did you know that you can change the stack (recursion) limit?

Yes, but unfortunately, it involves modifying the J binaries.  Also, I do not 
know how much is "enough" in advance.

>  I implemented Memo 
>  http://www.jsoftware.com/jwiki/Essays/Memo instead 
>  because it offered a more immediate and dramatic benefit.

Yes, and I appreciate M., it is quite a coup for the J programmer.  Thank you.  

However, even with  M.,  recursion is limited, primarily for two reasons.   The 
first is extrinsic:   M. currently only provides
value to scalar, numeric verbs.   This may change with future improvements to 
the adverb    In contrast, the second issue is
intrinsic and cannot be solved:  M. only provides values to verbs whose inputs 
are repeated.  For example, it will never speed up a
recursive string parser (e.g. for arbitrarily-parenthesized mathematical 
expressions).

-Dan


----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to