Path: utzoo!utgpu!news-server.csri.toronto.edu!mailrus!cs.utexas.edu!samsung!munnari.oz.au!bruce!goanna!ok
From: ok@goanna.cs.rmit.oz.au (Richard A. O'Keefe)
Newsgroups: comp.lang.fortran
Subject: Re: problem with "pe" format
Message-ID: <3200@goanna.cs.rmit.oz.au>
Date: 8 Jun 90 11:03:32 GMT
References: <90155.122858GL4@psuvm.psu.edu> <90158.091706GL4@psuvm.psu.edu> <8159@ubc-cs.UUCP>
Organization: Comp Sci, RMIT, Melbourne, Australia
Lines: 41

In article <8159@ubc-cs.UUCP>, buckland@cheddar.ucs.ubc.ca (Tony Buckland) writes:
> In article <90158.091706GL4@psuvm.psu.edu> GL4@psuvm.psu.edu writes:
> >Thanks to everyone who e-mailed me about the problem I was having with the
> >"p" format specifier. As everyone noted, this really *is* standard (although
> >very strange) FORTRAN behavior.

> >I'd be curious to know if anyone can think of a sitution where you'd actually
> >*want* the magnitude of the output to be changed like that.

Several people have suggested why you would want P at all, though no-one
as yet has pointed out that decimal scaling can be done by the formatting
routines without _any_ extra roundoff error.

As to why you would want it to be sticky, consider two facts:
(a) you want pe, pf, pg, so it looks as though it might be simpler
    for the "p" part to be a separate formatting command
(b) but you also want p10e and the like, and it's simplest to
    explain (and implement!) if p just sets a global variable.

Another possibility would have been for everything that uses the 'p'
flag to reset it to its default value, but then you'd have to write
	10(p,e)
rather than p,10e and unfortunately that might not interact with
Fortran's rules about the last set of parens in a format just the way
you would want.

So my surmise is that the expectation was that if you wanted scaling
you were probably printing out a table and wanted everything scaled
the same way, and that if you didn't want everything scaled the same
way it would be clearer if everything in the format had its own scale.

A rule of thumb for keeping things straight, then, is that if you
use P in a format at all, you should

	either put P first in the format (so you _know_ when it is
	set and that it applies to everything)

	or give each relevant item its own P	
-- 
"A 7th class of programs, correct in every way, is believed to exist by a
few computer scientists.  However, no example could be found to include here."