4.2 and F77: Summary of Responses
Greg Woods
woods at hao.UUCP
Sun Nov 20 06:16:38 AEST 1983
The following is a brief summary of responses to my article asking
about f77(1) and 4.2BSD. My own comments are enclosed in []'s. This is
not intended to be objective; few of the responses were. This is the gist
of what people had to say, as I read it.
1) Most of the responses were of the "let me know what you find out"
variety. There are many other frustrated FORTRAN users out there,
including at least two who had been preparing to post similar
requests themselves. [Potential compiler hackers/writers, take note!
More evidence that a FORTRAN newsgroup (how about net.lang.f77?)
is needed.]
2) Those in the know (meaning some people at Berkeley and some people
who have already brought up 4.2 on their systems) all say that the
code produced by the 4.2 version of f77 is *much* faster than that with
4.1a (which we have), and incorporates "many bug fixes". Most of them
felt that all the bugs have not yet been found/removed, but the majority
of them have. [This, if true, is good news for us. ] Take heart, FORTRAN
users!
3) No one knew of any other FORTRAN compiler running under Berkeley UNIX.
[We still would be interested in finding out about any such beast.]
4) About BYTE variables: two people pointed out that this is not part
of the standard, and therefore we shouldn't complain that they
"implemented the language you asked for instead of the one you wanted",
to quote one person. [However, that doesn't eliminate our need for it.
One other person suggested hacking the compiler to support an
INTEGER*1 declaration. An excellent suggestion, but no one here has
the knowledge and/or time to accomplish such a task. If anyone else
does this, please let us know!]
5) Several people stated that the ANSI FORTRAN 77 standard is sufficient
documentation for f77, since it implements the standard and only the
standard. [Upon hearing that, one person here commented that it is
like having a dictionary for a reference manual. The ANSI standard
documents can hardly be considered user-friendly!]
6) A couple of people sent me trivial C programs to do some of the
things I complained about not being able to do in f77. [I have
already written lots of such routines. Our problem is that not
all our FORTRAN users know C, so this option is not always
possible for us.]
7) The consensus was that the new debugger, DBX(1), does a much better
job on F77 programs than the current sdb(1). [Again, if true, that
is good news for us!]
We now have our 4.2 tapes, and will probably be going to 4.2 in the
next couple of weeks (that is, if our second VAX 11/750 ever arrives!
:-) If anyone wants to hear the opinions of our FORTRAN users, send
me mail and I'll be happy to accomodate. As usual, if there are
sufficient such requests, I'll post to the net instead.
Greg Woods, FORTRAN hacker
--
{ucbvax!hplabs | allegra!nbires | decvax!brl-bmd | harpo!seismo | ihnp4!kpno}
!hao!woods
More information about the Comp.unix
mailing list