J3/01-299
Date:     06 August 2001
To:       J3
From:     Dan Nagle
Subject:  Sj Response to a WG5 comment (named scratch files)

One of the comments made by WG5 members at the recent London meeting
expressed a concern regarding named scratch files.  The criticism
was that we hadn't thought this through very well.  Fortran supports
named and unnamed files, unnamed files are made by specifying a
status of scratch.  Named files are not.  A specific question was
asked regarding the proper reaction of a program to a named scratch
file when the file name already exists.  J3 members present were
unable to provide a good answer to this question.  WG5 expressed
its desire to remove named scratch files from the draft.  This
paper is a vehicle for removing named scratch files from the draft
should J3 decide to do so, or alternatively, for replying to WG5
with our reasons why named scratch files should be kept in the draft.

To remove named scratch files from the draft:
(Yes, I know I need to provide line numbers.
Let's see what JoR says before I do it.  My pdf
version doesn't seem to have line numbers,
at least, I haven't found the switch.)

EDITS

(Restore the normative paragraph discussing the OPEN statement.)
[171: (paragraph after the constraints)] Add after the first sentence
"If the STATUS= specifier has the value SCRATCH, the FILE= specifier shall not be present."

(Restore note in discussion of STATUS=.)
[174: (after the second paragraph under STATUS=)] Add the Note
"SCRATCH shall not be specified with a named file."