On Thu, 24 Jun 2004, sad wrote:

> >  then lots of currently perfectly correct
> > programs break.  If they're the same, then ELSE has different meanings
> > depending on whether NULL is specified, and that's generally bad from an
> > understanding the language standpoint.
>
> i've already thougth on this
> new control structure needed
> but the name of the IF is perfect %-)

I was thinking that something like Cs switch might work. There's still a
question of the keywords because I don't like reusing case, but maybe
something of the general form:
 case foo
  is true
  is false
  is null

The general form looks more like:

 case <expression>
  [when <expression>] then <statements>
  [is (true | false | null)] then <statements>
  [else] <statements>
 end case

With (unthoughtout) semantics like
 The case expression is evaluated once (although the below
  describes things in terms of expressions of the case
  expression, it is meant to be indicative of the intent
  not the actual implementation)
 Each when/is/else clause is evaluated in order as follows:
  For a when clause, if the value of the case expression is
   equal to the value of the when expression the statements
   are run and the case is ended.
  For an is clause,
   IS TRUE: if case expression IS TRUE is true then the
    statements are run and the case is ended
   IS FALSE: if case expression IS FALSE is true then the
    statements are run and the case is ended
   IS NULL: if case expression IS NULL is true then the
    statements are run and the case is ended
  For an else clause, run the statements (since no preceding
    condition has succeeded)

This is basically an extension of the syntax and semantics of one
of the case expression.  I don't think the above works keyword-wise
possibly, but it might be a reasonable starting point.


---------------------------(end of broadcast)---------------------------
TIP 9: the planner will ignore your desire to choose an index scan if your
      joining column's datatypes do not match

Reply via email to