On 11 Jul 2008, at 17:08, Brian Aker wrote:
Hi!
On Jul 5, 2008, at 7:20 PM, Stewart Smith wrote:
then again, I like Jim's idea of having NUMBER and STRING and none of
this fluffing about with 153 types of number.
I spoke to some Intel folks about this yesterday. Basically 1,2,4,8
are efficient. Almost everything really be treated as 4 though.
3bit is extremely inefficient and internally would be treated as
4bit (so saving space in memory depending on alignment is all you
will get).
BTW Date is 3 byte, we should look at that.
I am for keeping 1,2,4,8, but that is all. I am up in the air on
signed/unsigned.
Remove support for 64 bit unsigned type... It would remove a bunch of
special-case handling..
If you want to support 64bit unsigned, consider adding in support for
128bit signed type (HUGEINT?).
By making the largest integer type signed, you can then handle all
unsigned types smaller than it by simple masking and then you also
don't need to carry around an extra signed/unsigned flag for special
case processing like we must for unsigned bigints in mysqld today.
Regards,
Antony.
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
