From: aty@ucssun1.sdsu.edu (young a t) Subject: Re: Fortran, f2c and so on... Date: 8 Jul 1993 18:56:15 GMT
In article <211geqINNli@rave.larc.nasa.gov> jcburt@gatsibm.larc.nasa.gov () writes:
|>In article <MUTS.93Jul2102313@muts.hacktic.nl> muts@muts.hacktic.nl (Peter Mutsaers) writes:
|>>On Thu, 1 Jul 1993 00:06:28 GMT, ken@phys.titech.ac.jp (Ken-ichiro Aoki) said:
|>> KA> I will also add that in my experience, it is robust. It has
|>> KA> been around long enough so I suspect it is bugfree than some of
|>> KA> the commercial products.
|>>
|>>It is, but alas does not work for some programs I have to use :(. The
|>>problem is with some obscure format statements, that use a variable
|>>size. I don't know the details, but I have the feeling that the format
|>>statement is not fully implemented.
|>>--
|>
|>Ummm...could it possibly be that the "obscure format statements" that
|>won't work under f2c are not conformant to the F77 standard, but are
|>actually a vendor's own personal "extension" to the F77 standard...
No, there are perfectly standard-conforming F77 statements that f2c barfs on.
As I recall, one of the most irritating involved writing multiple records
to an internal file. The f2c thing fails to advance the record counter,
tries to write the second record to the end of the first, finds the line is
then too long, and dies a horrible death.
I do *NOT* regard f2c as a robust way to compile F77 in real-world
applications. Works fine in "toy" programs, but has problems with
the *real* stuff.
--