> How about doubling integer precision to 64 bits and float/double precision > to 128 bits? It's in line with 64 bit CPU capabilities, and personally, I > wouldn't mind it a bit if it would make numerical processing in PHP slower > on 32 bit systems. I think the advantages are much too great for that. If > this is ever gonna happen, I think PHP6 would be great timing for it. > If there were such a type as 'zend_long' used universally that might be at least possible (even configurable as a compile-time switch), but as it is *EVERYTHING* that touches a zval expects a long. On 32bit platforms that's a 32-bit number and there's no trivial way to change that.
Athlon64s and Opterons are cheap nowadays anyway. I picked up a MB/CPU combo at Fry's last week for like $200. -Sara -- PHP Internals - PHP Runtime Development Mailing List To unsubscribe, visit: http://www.php.net/unsub.php