Re: [AMBER-Developers] Usage of long symbol names, KIND, DFLOAT, etc.

From: Joe Krahn <>
Date: Tue, 31 Mar 2009 18:50:07 +0100

Robert Duke wrote:
> On kind params, I get kind of ill over the lack of standardization as to
> what a kind of type n implies. It is an open invitation to disaster,
Fortunately, compiler writers are less clueless, but I don't know how
people feel about relying on the non-standard convention that KIND ==
byte size.

> about 4 vs 8 byte integers. Joe, I would be interested to know if they have
> done anything about this in fortran 2003; I have not looked yet, and in
> general have not been interested because I don't want to force adoption of
> something that recent (never inconvenience your user for your own
> convenience...).
They have become only slightly less clueless. But, the people that had a
clue were out-numbered and left. KIND values are still abstract. The one
advantage is that you can use the ISO C Binding features to match C
types, such as double and int32_t.

The stubborn rationale about the abstract KIND is that a system could
theoretically have 2 different floating representations using the same
byte size. In particular, there is more than one form of quad-precision
currently in use. Also, character KINDs could indicate different
character sets, but nobody actually uses character kinds yet.

Of course, there is no reason not to have a KIND selection function
based on byte size, and no reason not to have a standard SIZEOF() function.


AMBER-Developers mailing list
Received on Wed Apr 01 2009 - 01:16:25 PDT
Custom Search