On Thu, 9 Oct 2008, Malcolm J. Currie wrote:
> I had a chat with Pat.
>
> Can problems with COCO and SLALIB please be put on the Starlink
> development list where Patrick can see them. Pat is not a subscriber
> to the JAC stardev list. I've forwarded correspondence.
>
> COCO is 100% Fortran 77 kosher. The issue lies with the compiler.
> Some Fortran 95 compilers do not fully support deprecated features of
> Fortran 77. Pat has mailed g95 author Andy Vaught with details.
>
>> the format or whether only the first is. I assume that everything will
>> work if we explcitly use IFA(1), IFA(2), IFA(3), IFA(4) in the WRITE
>> statement?
>
Malcolm - did you note whether your g95 that seemed to work was newer or
older than the version we have at JAC? Aug12 2008.
> To quote Pat:
>
> While it seems a pity to dumb down the COCO code just to appease a
> buggy compiler, perhaps you will have to resort to this. I'd be
> grateful if the changes could be appropriately notated, giving
> authorship and reason, and retaining the commented-out original code.
>
> The LU usage is more complicated than it might at first appear,
> because COCO supports a number of I/O options, not just interactive
> use. However, it's quite possible that the choices I made regarding
> LU 6 reflect historical features of VAX/VMS, and can be improved.
Well, after I reverted to the CONVEX hack it all worked on my Mac with
gfortran. Including calling coco with input and output files so I think
the patch is fine. I tested it on linux as well.
--
Tim Jenness
JAC software
http://www.jach.hawaii.edu/~timj
|