NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/comp/lang/fortran/4696

Text File | 1992-12-14 | 3.5 KB | 78 lines

Newsgroups: comp.lang.fortran
Path: sparky!uunet!walter!att-out!pacbell.com!decwrl!deccrl!news.crl.dec.com!dbased.nuo.dec.com!e2big.mko.dec.com!quark.enet.dec.com!lionel
From: [e-mail] (Steve Lionel)
Subject: Re: function name as an argument
Message-ID: <[e-mail]>
Lines: 65
Sender: [e-mail] (Mr. USENET)
Reply-To: [e-mail] (Steve Lionel)
Organization: Digital Equipment Corporation, Nashua NH
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Date: Mon, 14 Dec 1992 15:24:01 GMT


In article <[e-mail]>,
[e-mail] (David Lombard) writes:
|>In article <[e-mail]> [e-mail] (Steve Lionel) writes:
|>>                                     Indeed, the standard says:
|>>
|>>    If a character function is referenced in a program unit,
|>>    the function length specified in the program unit must be
|>>    an integer constant expression.
|>>
|>>            ANSI X3.9-1978, section 15.2, p15-2, lines 5-7
|>>
|>>Because VAX FORTRAN supports the usage above, we always pass character
|>>functions by descriptor, which includes the length.  But a standard-conforming
|>>program can't do this thus a compiler which chooses not to support this
|>>extension can get away without passing the length of character function
|>>arguments.  (When the function is actually called, the length of the function
|>>result does have to be passed, but that's another story.)
|>>
|>
|>If you're going to quote the standard, make sure you're right.

Sage advice. I believe I followed it in this case.

|>
|>See section 8.4.2 CHARACTER Type-Statement, line 46-:
|>
|>    If a dummy argument has a len of (*) declared, the dummy
|>    argument assumes the length of the associated actual
|>    argument for each reference of the subroutine or function.
|>
|>Thus, in a function or subroutine,
|>
|>    CHARACTER FRED*(*)
|>
|>is correct (IMHO the _proper_ declaration!).
|>

Your own advice is probably appropriate here.  A careful reading of the
paragraph you quote reveals that these lines only tell you what happens *IF*
the dummy argument has a len of (*); it doesn't say that such a len is legal
in this case.  If you keep reading to lines 5-10 on page 8-7, you'll see the
additional rule that if the program unit references the function, the length
must NOT be (*) but rather be an integer constant expression which "must
agree with the length specified" in the actual function specification. (It
was a restatement of the first part of this which also appeared in chapter
15 and which I quoted in the previous posting.) A length of (*) is legal, as
others have mentioned, only if the function is not referenced in the program
unit; perhaps it is passed as an argument to another routine.

The Fortran language standard is not as explicitly written as that for some
other languages (notably Ada), and there are times when the reader has to be
familiar with the standard as a whole, fitting together pieces mentioned in
diverse sections.  Even after many years of study, there are traps for the
unwary and not a small number of ambiguities.   I've been in the Fortran
business for over 14 years now, and I still find little surprises between
the lines.  Don't feel too bad if it trips you up from time to time.

--

Steve Lionel                    [e-mail]
SDT Languages Group
Digital Equipment Corporation
110 Spit Brook Road
Nashua, NH 03062