On Thu, 4 Feb 1999, Mike Brenner wrote:

> Rich > ... The more object-oriented way is to export a function
>      > ... in place of the deferred constant
> The weak point of this is the inability to have static constants
> that are constructed from initializers requiring that function.
> Sometimes having a static constant is a performance advantage.


  You are correct for the case of a declared static constant in the
  declarative part of a subprogram.  I am referring to the deferred
  constant one sometimes exports from the public part of a package,
  rather than one declared locally to a subprogram.

  Of course the function would be in-lined, but that does not solve the
  issu you are citing. I agree that statically declared constants can
  be more efficient.