home *** CD-ROM | disk | FTP | other *** search
- Newsgroups: comp.sys.sgi
- Path: sparky!uunet!charon.amdahl.com!pacbell.com!ames!kronos.arc.nasa.gov!iscnvx!netcomsv!netcom.com!jchauvin
- From: jchauvin@netcom.com (John H. Chauvin)
- Subject: Bug Report: FORTRAN 3.10 compiler
- Message-ID: <1992Nov7.030148.11596@netcom.com>
- Organization: Netcom - Online Communication Services (408 241-9760 guest)
- Date: Sat, 7 Nov 1992 03:01:48 GMT
- Lines: 41
-
-
- I have discovered two additional bugs in the 3.10 release of the
- FORTRAN compiler. Both bugs have been reported to Technical
- Support.
-
- (1) The -g and -listing compiler options when used together will
- sometimes result in a core dump from /usr/lib/fcom. I have not
- found any indicator to predict when this will happen. The work
- around is not to use the two options together.
-
- (2) The -listing compiler option incorrectly lists undeclared real
- variables as real*8 NOT real*4. For example, examine the
- following code:
-
- A = 1.0
- B = 2.0
- C = A + B
- STOP
- END
-
- What is the data type for variable A? Assuming the code was
- compiled using the default compiler options, variables A, B, and C
- should be real*4 (This fact is not listed in the manuals or man
- pages. It should be!). This can be check by using dbx. Yet if you
- compile this code with the -listing option, the variables are listed as
- real*8.
-
- I have not fully explored this bug, but it does bring in to question the
- accuracy of the other information provided by the -listing option.
-
- I encourage other users to use this forum to list any bugs they may
- discover. Many hours of work could be saved if users are aware of
- outstanding bugs not included in the release notes. Any comments
- are welcome
-
- jchauvin@netcom.com
- Northrop ATDC
- Pico Rivera, California
- --
- John H. Chauvin jchauvin@netcom.COM
- Netcom - Online Communication Services San Jose, CA
-