You sure that's actually what he said? A change in CRC proves the data
changed, but lack of a change does not prove it didn't.
"To quickly determine if rows have changed, we rely on a cyclic
redundancy checksum (CRC) algorithm. If the CRC is identical for the
>
"summary" functions, such as an MD5 hash. I wouldn't trust it at all
with a 32-bit CRC, and not much with a 64-bit CRC. Too much risk of
collision.
Small example of collisions for crc32:
0x38ee5531
Hundemarke
92294
0x59471e4f
raciner
tranchefiler
0x947bb6c0
Betriebsteile
4245
I had make a lot of work when choosing hash function for tsearch2. Also, I had
find that popular hash algorithms produce more collision for non-ascii
languages... CRC32 is more "smooth".
On dictionary with 332296 unique words CRC32 produces 11 collisions, perl's hash
function - 35, pgsql's hash_any - 12.
--
Teodor Sigaev E-mail: [EMAIL PROTECTED]
WWW: http://www.sigaev.ru/
---------------------------(end of broadcast)---------------------------
TIP 2: Don't 'kill -9' the postmaster
