This is the mail archive of the
mailing list for the GCC project.
[Bug fortran/24807] Fortran supports real*16, but not complex*32
- From: "schnetter at aei dot mpg dot de" <gcc-bugzilla at gcc dot gnu dot org>
- To: gcc-bugs at gcc dot gnu dot org
- Date: 19 Nov 2005 01:16:12 -0000
- Subject: [Bug fortran/24807] Fortran supports real*16, but not complex*32
- References: <firstname.lastname@example.org/bugzilla/>
- Reply-to: gcc-bugzilla at gcc dot gnu dot org
------- Comment #4 from schnetter at aei dot mpg dot de 2005-11-19 01:16 -------
The same error message was already given (a few lines further down) when the
kind had an unsupported value. An old style kind is given in a declaration as
"real*8", a new style kind as "real(8)".
For integer and real types, the old and new style kinds are identical. For
complex types, they differ, and the new style kind is half the old style kind.
Since the old style kind is not part of the Fortran standard, there is no good
way of naming it. One might instead call it the "size" of the type, since
real*8 is supposed to take up 8 bytes.
For a type "complex*15" it is not correct to say "complex kind 15 not
supported", since the "size" 15 should be mapped to the kind 7.5 (which is not
integer, which causes the error). A "complex*16" has kind 8.
The same error message is already give a few lines further down in the code.
Should we leave these messages as they are? Or should we change them to use
"size" instead of "old style kind"? Whichever we do, we should keep the