NetNews Usenet Archive 1992 #30/NN_1992_30.iso/spool/comp/lang/fortran/4715

Internet Message Format | 1992-12-15 | 3.0 KB

Path: sparky!uunet!stanford.edu!snorkelwacker.mit.edu!ai-lab!life.ai.mit.edu!burley
From: [e-mail] (Craig Burley)
Newsgroups: comp.lang.fortran
Subject: Re: function name as an argument
Date: 15 Dec 92 03:17:25
Organization: Free Software Foundation 545 Tech Square Cambridge, MA 02139
Lines: 46
Message-ID: <[e-mail]>
References: <[e-mail]>
    <[e-mail]>
    <[e-mail]>
    <[e-mail]>
NNTP-Posting-Host: apple-gunkies.gnu.ai.mit.edu
In-reply-to: [e-mail]'s message of Tue, 15 Dec 92 01:59:58 GMT

In article <[e-mail]> [e-mail] (David Lombard) writes:

   In article <[e-mail]> [e-mail] (Steve Lionel) writes:

   >Absolutely true.  But even if you compiled all the routines together, what if
   >you added another caller of SUB which passed FUNC declared as a different
   >length?  Unless you additionally require procedure inlining, there is no way
   >that SUB can know which routine called it and which length to statically
   >allocate.  This is why the standard requires the length of a character
   >function to be declared as an integer constant expression in the program
   >unit which calls the function.  And this is why an implementation (such as
   >VAX FORTRAN), which does allow the (*) in the referencing unit, passes the
   >length along with the routine address, so that it may do run-time allocation
   >for the function result.
   >
   NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO NO!
   [useless repetitions of above line deleted]

   PLEASE, read sections 8.4.2 and 15.9.3.1 of the standard.  They explain why
   all implementations of the language must maintain and pass around the
   length of each character variable.  All implementations must allow the (*)
   in subprograms.  This is not some mystical example of the great god VMS.
   If an implementation does not support (*), then it is simply non-conforming.

This is a case of _your_ not reading the subject heading.  The discussion
is about passing character _functions_, not _variables_.  It _is_
standard-conforming for an implementation to not support CHARACTER*(*) CFUNC
if the program unit does CFUNC(...) anywhere.

   A word to the wise, the standard is not a document one can casually read.

Neither, apparently, is a USENET posting.  :-)

   The information is `scattered' throughout the document.  One must `swim'
   in the standard to understand it.  This is not meant as an impeachment of
   the standard, just a statement of fact.

Or, one could read the subject heading if one is unwilling to read through
earlier portions of the thread.  :-)

Admittedly, this kind of confusion is one of the reasons we all love
Fortran so much.  :-)
--

James Craig Burley, Software Craftsperson    [e-mail]
Member of the League for Programming Freedom (LPF) [e-mail]