On Wed, Feb 20, 2002 at 11:13:44PM +0000, Nicholas Clark wrote: > Sorry to be off topic. > > If I have a string of some length, containing characters between 0 and 255, > I'm wondering there's a fast perl way to make a 64 byte long bit vector > from it, with bits set for any character present in the original string.
I suspect 32 bytes is sufficient, no? > If I can do this quickly, then when I'm trying lots of things to see if $a is > a substring of $b, then I think I can optimise it. > I should be able to rapidly reject $b as a candidate by logical operations on > the two bit vectors in constant time whatever the length of the strings - if > $b doesn't contain all the characters in $a somewhere, then $a can't possibly > be a substring, so I don't need to bother trying index. Short of coding in C, I can't think of a way to do this without a Perl level loop, which would kill efficiency, I imagine. The best I can come up with this early is something containing vec ord for split > [yes, I'm still playing with Encode's compile script. > No, it's not a waste of time - I've got it taking about 37% less time > than it did 2 days ago. Much cheaper than throwing hardware at the world] On behalf of anyone who has ever compiled Encode, thanks! -- Paul Johnson - [EMAIL PROTECTED] http://www.pjcj.net
