Path: utzoo!utgpu!jarvis.csri.toronto.edu!rutgers!rochester!bbn!csd4.milw.wisc.edu!bionet!agate!saturn!usenet
From: usenet@saturn.ucsc.edu (Usenet News Account)
Newsgroups: comp.lang.fortran
Subject: Re: Character aliases are Satanic extensions
Summary: Warnings are a very good thing
Message-ID: <7747@saturn.ucsc.edu>
Date: 31 May 89 15:02:28 GMT
References: <598@mbph.UUCP> <100@unmvax.unm.edu>
Organization: U.C. Santa Cruz, CIS/CE.
Lines: 36
Sender:

Organization: University of California, Santa Cruz;  UCO/Lick Observatory
Keywords: Compiler Warning Messages
From: mongo@helios (Mongo Package)
Path: helios!mongo

In article <100@unmvax.unm.edu> brainerd@unmvax.unm.edu (Walt Brainerd) writes:
>In article <598@mbph.UUCP>, hybl@mbph.UUCP (Albert Hybl  Dept of Biophysics  SM) writes:
>> the innuendo--ONLY STUPID PROGRAMMERS USE VENDOR EXTENSIONS!
>IGNORANT programmers also use vendor extensions (there is a lot of
>difference).  Also INTELLIGENT programmers use extensions ....
>We seem to have two proposals (please don't anyone say these are MY proposals):
>1)  No vendors should be allowed on standards committees.
>2)  The Fortran standard should state that the processor does not
>    conform to the standard if it accepts extensions.

If the Fortran 8x standard ends up *requiring* that a Fortran compiler be
able to generate Warning messages, then things are headed in the right
direction.  With such warnings, anyone who writes code that might ever
be compiled by a different compiler can know about it.  I would even prefer
it if the warnings were ON by default.

I make rather liberal use of the non-standard "include" statement in FORTRAN
77.  I do this knowing that I would write a Pre-Processor to duplicate the
"include" functionality if I were using a compiler which did not have it.
The utility of "include" to me is sufficient that I would not mind having
to watch Warning statements roll past on the screen.  I prefer to think that
this does not make me STUPID or IGNORANT.

Finally, I personally think that any Fortran/FORTRAN compilers that accept
Hollerith should be required to *ALWAYS* print a warning message.  When
an extension has been specifically DISALLOWED, the programmers should be
forced to notice this Every Time the code is compiled.  Vendors should look
upon this as providing a service.

Steve Allen
No, my employer may not share these opinions.