When you've spent a week on a bug which has a workaround, its time to
give up. The workaround is clearly documented in the code and is
registered in our problem database, and has survived porting from a Cray
C90 to a T3E, an MPP re-write and a further port to an IBM SP-2 and
regular runs under HP-UX. I'm confident that this one is a compiler
problem, not a code one, so use of other compilers on other machines is
not an issue, and after trying to remove the fix on the T3E, the problem
seems still extant in SGI's compiler after their port. I try removing
the hack about once a year with the latest compiler, always to no avail.
BTW, we've found and reported a fair few bugs with SGI and Nag's
compilers in our time, and have current suspicions over IBM's, but often
we'd rather do a re-code than argue a point of f90 syntax with the
vendor (I'm certainly wary of telling Malcom the Nag compiler is
non-conformant!)
John
--
John Bray, Room 337, Numerical Weather Prediction, UK Met Office
Work: [log in to unmask], http://fr0400/~frjb (internal only)
+ 44 (0) 1344 854035 (voice), + 44 (0) 1344 854026 (fax)
Home: http://www.cix.co.uk/~vicarage/
%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%
|