Re: Does anyone have a better scm_string_hash ?

[Marius Vollmer]

Stephen Compall <address@hidden> writes:

> /* This one comes from `recode', and performs a bit better than the above as
>    per a few experiments.  It is inspired from a hashing routine found in the
>    very old Cyber `snoop', itself written in typical Greg Mansfield style.
>    (By the way, what happened to this excellent man?  Is he still alive?)  */
>
> size_t
> hash_string (const char *string, size_t n_buckets)
> {
>   size_t value = 0;
>
>   while (*string)
>     value = (value * 31 + (unsigned char) *string++) % n_buckets;
>   return value;
> }

This is essentially(?) the one used by glib.  We use it with 37
instead of 31 and like glib, compute the module after the loop.

    unsigned long 
    scm_string_hash (const unsigned char *str, size_t len)
    {
      /* from suggestion at: */
      /* http://srfi.schemers.org/srfi-13/mail-archive/msg00112.html */

      unsigned long h = 0;
      while (len-- > 0)
        h = *str++ + h*37;
      return h;
    }

-- 
GPG: D5D4E405 - 2F9B BCCC 8527 692A 04E3  331E FAF8 226A D5D4 E405