home *** CD-ROM | disk | FTP | other *** search
- Path: sparky!uunet!cs.utexas.edu!sun-barr!ames!haven.umd.edu!darwin.sura.net!jvnc.net!yale.edu!think.com!Think.COM!sandee
- From: sandee@Think.COM (Daan Sandee)
- Newsgroups: comp.lang.fortran
- Subject: Re: programming style question
- Date: 12 Aug 1992 20:13:25 GMT
- Organization: TMC
- Lines: 20
- Message-ID: <16brd5INN946@early-bird.think.com>
- References: <1992Aug12.193739.23817@news.Hawaii.Edu>
- NNTP-Posting-Host: telecaster.think.com
-
- In article <1992Aug12.193739.23817@news.Hawaii.Edu>, tholen@galileo.ifa.hawaii.edu (Dave Tholen) writes:
- |> If OPENing standard output, what good is trapping an ERR= condition,
- |> because there is no output device to which an error message can be
- |> written? Of course, one could use the STOP 'Error message here'
- |> construct to get around this problem.
-
- In some implementations, a STOP statement is written to stdout *and* stderr.
- Hopefully, an error on preconnecting stdout and stderr will be handled
- gracefully by the system. But I wouldn't advise a STOP message if you find
- a problem on stdout.
-
- You can write an error message to any file (stderr is the obvious first
- choice). Normally, this would occur if stdout is redirected to a file and
- (a) the file is write protected or (b) the device is full (or something).
- More interesting is when you get an error on writing to stderr. Then there
- is no obvious solution.
-
- Daan Sandee sandee@think.com
- Thinking Machines Corporation
- Cambridge, Mass 02142 (617) 234-5044
-