DN>
  > From the foregoing, I would conclude that implemnters of C/C++ do not
  > use exp(log()) any longer for coding pow(). Instead, they use some
  > more specialised algorithm, possibly with special cases for [small]
  > integer roots.
DN>

  The commentary in \BCOS2\SOURCE\RTL\COMMON\POW.ASM is too long to post
  here, but reading it will reveal that Borland uses two separate
  algorithms for implementing the pow() function (one of which is an
  optimisation for powers of integers), along with a handful of code to
  handle some special cases.

  > JdeBP <
___
 X MegaMail 2.10 #0:

--- Maximus/2 3.00