Thread

1. default operator class: btree or hash

   Dmitry G. Mastrukov Дмитрий Геннадьевич Мастрюков <dmitry@taurussoft.org> — 2003-12-01T05:12:54Z

   Hi, all!
   
   I'm preparing new release of uniqueidentifier datatype. In new relaease
   in addition to existing btree operator class I've implemented hash
   operator class. But what class should be default for this datatype?
   Uniqueidentifier is used for identification, so main operation with it
   is '=', on machines with /dev/urandom '<','>' operations have no meaning
   (parts for date and time are randomized). May be hash operator class
   should be default for uniqueidentifier?
   
   Regards,
   Dmitry
   
   
   
   

2. Re: default operator class: btree or hash

   Alvaro Herrera <alvherre@dcc.uchile.cl> — 2003-12-01T13:33:23Z

   On Mon, Dec 01, 2003 at 08:12:54AM +0300, Dmitry G. Mastrukov wrote:
   
   > I'm preparing new release of uniqueidentifier datatype. In new relaease
   > in addition to existing btree operator class I've implemented hash
   > operator class. But what class should be default for this datatype?
   > Uniqueidentifier is used for identification, so main operation with it
   > is '=', on machines with /dev/urandom '<','>' operations have no meaning
   > (parts for date and time are randomized). May be hash operator class
   > should be default for uniqueidentifier?
   
   btree is a better choice anyway because it allows for more concurrent
   operation and it's far more tested.
   
   -- 
   Alvaro Herrera (<alvherre[a]dcc.uchile.cl>)
   "No reniegues de lo que alguna vez creíste"