Re: poor m4 hash performance

[Eric Blake]

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

According to Paul Eggert on 6/4/2006 10:24 PM:
> Eric Blake <address@hidden> writes:
> 
>> right now, both the 1.4 branch and head use an algorithm borrowed from emacs:
> 
> That algorithm could be improved.  M4 should do all its hash
> computation using size_t, not a mixture of int and size_t.  And it
> shouldn't slow itself down to do case-folding, since M4 is
> case-sensitive.  And it shouldn't assume that int is 32 bits wide.

Agreed (I had already noticed that using a signed hash was wrong, but had
not picked up on the case-folding).  I will go ahead and commit your patch.

> 2006-06-04  Paul Eggert  <address@hidden>
> 
>       * src/m4.h (hash_table_size): Now size_t instead of int.
>       * src/m4.c (hash_table_size): Likewise.
>       (main): Adjust to this; use atol rather than atoi.
>       * src/symtab.c: Include <limits.h>, for CHAR_BIT.
>       (symtab_init, lookup_symbol, hack_all_symbols):
>       Use size_t for sizes and indexes, not int.
>       (hash): Likewise.  Don't case-fold in the hash function.
>       Shift by 7, not 3, for consistency with gnulib/lib/hash.c.
>       Don't assume hash word is 32 bits.
> 

- --
Life is short - so eat dessert first!

Eric Blake             address@hidden
-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.2.1 (Cygwin)
Comment: Public key at home.comcast.net/~ericblake/eblake.gpg
Comment: Using GnuPG with Mozilla - http://enigmail.mozdev.org

iD8DBQFEhCGK84KuGfSFAYARAg1hAJ4+g+oo7u6RlHPxTNLHmryP6Qo2dwCgqIjS
SLdMy9iasNTEWlXeBOnCpdw=
=e67k
-----END PGP SIGNATURE-----