mckenney@acf2.NYU.EDU (Alan Michael McKenney) (03/20/89)
[Please respond to the E-mail address at the *end* of this article,
and **NOT** to the address in the article header. The news posting
mechanism on my home machine is, as always, broken, so I have to
use a different one to post.]
A question on the 1977 FORTRAN standerd with respect to COMPLEX
variables:
are COMPLEX data required to be stored as a pair of REAL data?
The standard (page 4-4) says "The representation of a complex datum is
in the form of an ordered pair of real data. ... Each part has the same
degree of approximation as for a real datum." It is not clear to me
from this whether each part has to have the same (exact) representation
as a REAL datum. Every implementation I have ever seen has done this,
but it is conceivable that someone, somewhere, will do it different if
it is allowed (and perhaps for a good reason.)
The situation where it comes up is: we have a routine which has
various scalar arguments, some of which are REAL, but in most cases
where it will be called, the caller will have a COMPLEX variable
with zero imaginary part to pass. It would be simpler to just
write in the COMPLEX variable, rather than always have to cast
it to REAL. If (as it is for most implementations) the COMPLEX
variable is stored as a REAL array, then the callee will just get
the first element of that array, which will be a perfectly good
REAL number which is the real part of the complex number. However,
we want to be sure that we don't get bitten somewhere down the road
if our code gets to a machine which uses a different represenation
for the parts of a COMPLEX number that for a REAL.
And yes, I know, there are other ways to do this.
Please send E-mail, and I will summarize.
Alan McKenney E-mail: mckenney@acf8.nyu.edu (INTERNET)
Courant Institute,NYU ...!uunet!cmcl2!acf8!mckenney (UUCP)
--
Alan McKenney E-mail: mckenney@acf8.nyu.edu (INTERNET)
Courant Institute,NYU ...!uunet!cmcl2!acf8!mckenney (UUCP)