(j3.2006) Are functions in 15.2.3 pure?

[Malcolm Cohen]

>Are the functions in 15.2.3 pure?

No.

>  It doesn't say so in 15.2.3.1, or in
>their individual descriptions.

No contradiction there then.  (Unlike the c14 case.)

>Should they be pure?  I can't think of a reason why not.

It would appear to be reasonable to make them pure, but they are not pure in 
F2003 or F2008.

I think the definition of purity in 12.7 needs an extra entry, something like
  - a module procedure in an intrinsic module that is specified to be pure
otherwise there is a contradiction between 12.7 and 14.x.  It might also be 
considered advantageous to allow vendors to supply additional pure procedures in 
an intrinsic module.

BTW, compiler_options() and compiler_version() are also apparently not pure.

Cheers,
-- 
................................Malcolm Cohen, Nihon NAG, Tokyo.