woods@hao.UUCP (Greg Woods) (11/10/83)
I don't know if this is the right newsgroup for this, but I don't
know where the hell else to put it.
We have recently brought up a VAX running 4.1aBSD UNIX. Our users
are extremely unhappy with the bug-filled FORTRAN compiler f77(1). Considering
that we just had a long and involved battle over VMS vs. UNIX and UNIX won,
there is a lot of "I told you so" going on, since superior FORTRAN was one
of the major arguments in favor of VMS. I know many UNIX hackers think
FORTRAN is a four-letter word, but in our line of work (scientific research)
we are stuck with it, like it or not. Some of the bugs complained about are:
1) The implicit statement does not correctly redefine variable types
(although it seems to work for function calls) when the -I2 flag is
in use (and maybe other times as well, but I have only documented
it when using -I2).
2) There is no BYTE variable, and the compiler will not let you
equivalence an integer to a CHARACTER*1 for purposes of byte-swapping.
3) Execution of the "stop" statement sometimes causes a core dump,
for no apparent reason.
4) SDB(1) will not print out variables in common blocks, and a new
"fixed" version I got from someone on the net a while back refuses
to read core files (it always says: "Bad core magic number").
5) The documentation is almost nonexistent. For example, nowhere is there
a list of what intrinsic functions are available, short of perusing
the sources in /usr/lib/F77. And the functions that are available
are inconsistent, i.e. there is a "dreal" but no "dimag".
Now for the real point of this article: Two questions:
1) We are about to go to 4.2BSD. Is the f77 that comes with that
version of UNIX any better than the one that comes with 4.1a?
2) Does anyone know of any other FORTRAN compilers that will run under
4.2BSD? We might even consider buying one, but of course we would
prefer something in the public domain.
Any help would be *greatly* appreciated!
Greg "I use FORTRAN" Woods
--
{ucbvax!hplabs | allegra!nbires | decvax!brl-bmd | harpo!seismo | ihnp4!kpno}
!hao!woodswls@astrovax.UUCP (William L. Sebok) (11/11/83)
> Now for the real point of this article: Two questions: > > 1) We are about to go to 4.2BSD. Is the f77 that comes with that > version of UNIX any better than the one that comes with 4.1a? > > 2) Does anyone know of any other FORTRAN compilers that will run under > 4.2BSD? We might even consider buying one, but of course we would > prefer something in the public domain. I had been commissioned by our Dept. Chairman, Dr. Ostriker to post a similar request, but I seen it has already been done for me. We, as another astronomical institution, are mostly a Fortran shop (personally I'm a FORTH person, with a bit of C, but that's another matter). Like it or not FORTRAN is as close as there is to a universal astronomical programming language, which (unfortunately) is why most astronomical sites with Vaxes run VMS rather than Unix. The same two questions asked above we would also like to ask. -- Bill Sebok Princeton Univ. Astrophysics {allegra,akgua,burl,cbosgd,decvax,ihnp4,knpo,princeton}!astrovax!wls
woods@hao.UUCP (Greg Woods) (11/11/83)
So far, the response to my request for F77(1) info has been overwhelming.
I will post a summary to the net, so you don't need to send me yet another
"let me know what you find out" letter.
GREG
--
{ucbvax!hplabs | allegra!nbires | decvax!brl-bmd | harpo!seismo | ihnp4!kpno}
!hao!woodsfair@dual.UUCP (Erik E. Fair) (11/15/83)
Berkeley has been spending a fairly large effort toward improving the
f77 compiler. In particular, they wanted to make the optimized code as
fast as VMS fortran. If you want details, talk to Bob Corbett
(ucbvax!corbett). I don't know if it will be in 4.2, although I would
assume so.
Erik E. Fair {ucbvax,amd70,zehntel,unisoft}!dual!fair
Dual Systems Corporation, Berkeley, Californiagwyn%brl-vld@sri-unix.UUCP (11/18/83)
From: Doug Gwyn (VLD/VMB) <gwyn@brl-vld> 4.2BSD comes with a complete new Fortran-77 that is supposed to be MUCH better. So does UNIX System V Release 1.1. I don't know how these f77 systems are related to each other. There should be no question of what intrinsics the language has, they are specified in the ANSI standard. If they aren't all supplied, that should be fixed. The only major surprise in the UNIX f77 is that files are positioned at their END after an OPEN, whereas one would expect the initial position to be the BEGINNING. (This was apparently done to provide a way to open a file for appending, since the ANSI OPEN does not have a MODE=APPEND clause.)
jsq@ut-sally.UUCP (John Quarterman) (11/21/83)
The story I heard was that the System V f77 was an early version of
the 4.2BSD one, and if not earlier, at least did come from Berkeley.
Not being a FORTRAN hacker, I haven't been curious enough to check this,
but have no reason to disbelieve it.
--
John Quarterman, CS Dept., University of Texas, Austin, Texas
{ihnp4,seismo,kpno,ctvax}!ut-sally!jsq, jsq@ut-sally.{ARPA,UUCP}