NetNews Usenet Archive 1992 #18/NN_1992_18.iso/spool/comp/programm/2432

Internet Message Format | 1992-08-22 | 1.6 KB

Path: sparky!uunet!cs.utexas.edu!wupost!waikato.ac.nz!aukuni.ac.nz!ecmtwhk
Newsgroups: comp.programming
Subject: Re: ANSI C, was Re: Teaching the basics
Message-ID: <[e-mail]>
From: [e-mail] (Thomas Koenig)
Date: Sun, 23 Aug 1992 02:17:56 GMT
References: <[e-mail]>
Organization: University of Auckland, New Zealand.
Lines: 25

[e-mail] (Boyd Roberts) writes:

>Prototypes are cheap hack, and they can lie, whereas the actual
>function declaration specifies it exactly.  The protypes just say that the
>calls agree with the prototypes, and that's not the same as the actual
>declaration.

>Putting extra info in the object would be a total win.  Why should I go to
>declare my functions twice, when the machine can do that stuff for me?

Prototypes are at least better than nothing, if you enforce their use.
ANSI-C really could not do much better without breaking all existing C
program up to that date. If you have declared your header dependencies
in your makefile correctly, you should not have any problem.

Actually, I like the Modula-2 concept much better: Put definitions
(declarations, in C usage) into separate files and let the compiler barf
if use or implementation (definition, in C usage) do not agree with
these. Finally, let the linker check wether the compiled definition and
implementation modules actually belong together. A pity Modula-2 does
not have either decent or portable I/O libraries...
--
Thomas Koenig, [e-mail], [e-mail]
The joy of engineering is to find a straight line on a double logarithmic
diagram.