>
> >> I've just finished working on the type
> >> conversion algorithms so understand the current "atttypmod" field a
> >> bit better, but have not decided how to extend it to multiple fields.
> > divide it into two 16 bit integers ?
>
> At the moment it already _is_ a 16 bit integer, so it would have to be
> divided into two 8 bit integers. Still OK, but then it must be a
> positive number, so one field can be only 7 bits. I was thinking of
> trying to solve the problem generally so that a type definition can also
> define a "type support type" similar to the current atttypmod, but which
> could be single or multiple numbers, or a string, or...
use unsigned short, that is 8 bits.
>
> Don't know if it would be generally useful though; still thinking about
> how to implement different character sets and collation sequences for
> strings and it seems like this might help.
>
> > A mathematical package exists for infinite scale decimals, I think
> > it was part of a 56 bit RSA cracking effort. It has all thinkable
> > operations defined...
>
> Well, if you find it again let us know ;) In the meantime, the 64-bit
> integers are probably the best candidate implementation.
Yes, the 64-bit idea is good.
--
Bruce Momjian | 830 Blythe Avenue
maillist@candle.pha.pa.us | Drexel Hill, Pennsylvania 19026
+ If your life is a hard drive, | (610) 353-9879(w)
+ Christ can be your backup. | (610) 853-3000(h)