Gregory Stark wrote:
>
> I think the change to the hash functions needs to be called out in the release
> notes.
>
> If anyone stored any results of hashintN, hashfloat8, etc in their database or
> outside the database those results will have changed. It's fairly unlikely but
> there could be someone out there doing that.
>
> No intervention is required for normal expression indexes using those
> functions or hash indexes which will be rebuilt during a database upgrade
> anyways. I don't think hash_any itself changed so this wouldn't affect
> hashtext or any hash function which was already using hash_any.
Agreed. Added to the Migration section of the 8.3 release notes:
<listitem> <para> Internal hashing functions are now more uniformly-distributed (Tom) </para>
<para> If application code was calling and storing hash values using internal <productname>PostgreSQL</>
hashingfunctions, the hash values must be regenerated. </para> </listitem>
I assume people will realize this does _not_ affect md5(), or should I
add a mention of that too.
-- Bruce Momjian <bruce@momjian.us> http://momjian.us EnterpriseDB
http://postgres.enterprisedb.com
+ If your life is a hard drive, Christ can be your backup. +