This is the mail archive of the
gcc@gcc.gnu.org
mailing list for the GCC project.
Re: Bug 15718 dispute: showstopper or WONTFIX?
On Wed, Jun 02, 2004 at 06:51:26AM +0200, Gabriel Dos Reis wrote:
> | > Due to a calling convention bug in the Alpha backend, the compiler
> | > may not accept some constructs involving function calls using native
> | > complex numbers either as parameter values or return value.
> |
> | > That undoubtedly needs polishing.
> |
> | It understates the issue considerably. LAPACK 3.0 hits this bug.
>
> In which ways does it understate the issue?
If you say "may not accept some constructs", some would read that to
suggest that only corner cases are hit, and only sometimes. If particular
language features are always broken, then we should just say so.
> | I believe that *every* Fortran program using an inline function
> ^^^^^^^^^^^^^^^^^^^^^^^^
> | with a complex argument hits the bug, though I'm not certain.
> ^^^^^^^^^^^^^^^^^^^^^^^
The key is "every". We don't get to say "may" (indicating possibility)
if the answer is "every" (certainty).
Unfortunately the Alpha box I was using seems to be down, so I won't
be able to run tests immediately.