Re: RE : Re: Re: Function memoization

>>>>> "Florent" == Florent Georges <[email protected]> writes:

    Colin> This is no good. That means two functions calls that pass the
    Colin> value 3 will not (usually) be memoized.

    Florent>   Maybe it would make sense to say:

    Florent>     - if the argument is an atomic item (or a sequence of
    Florent> 1 such item, as these are indistinguishable), compare by
    Florent> value (string() or data()?);

Good.

    Florent>     - if the argument is a node, compare with the "is"
    Florent> operator;

Good.

    Florent>     - else (so if it is a sequence of at least 2 items),
    Florent> compare by identity (the two sequences must be the same
    Florent> one).  I think such a "sequence identity" doesn't exist
    Florent> in the CRs.

Re-formulate that as compare each item of the sequence of identity as
determined above.

    Colin> Now if long strings are being passed as arguments, then indeed
    Colin> the cost of recognizinf them may be less than the cost of
    Colin> running a small function.

    Florent>   This is the /raison d'être/ of this extension.  The
    Florent> implementation doesn't have to guess itself where to use
    Florent> memoization, the developer will tell it that.

    Florent>   Does it make sense?

I guess you are right. If the developer makes a bad decision - tough.

It's analogous to xsl:key.





























	

	
		
    Florent> ___________________________________________________________________________
    Florent> Yahoo! Mail réinvente le mail ! Découvrez le
    Florent> nouveau Yahoo! Mail et son interface révolutionnaire.
    Florent> http://fr.mail.yahoo.com


-- 
Colin Adams
Preston Lancashire