lclint-interest message 117
From evans@cs.virginia.edu Fri Oct 18 14:49:34 1996
Date: Fri, 18 Oct 96 14:46:43 -0400
From: evans@cs.virginia.edu (David Evans)
To: lbr@mjolner.dk
Cc: lclint-interest@larch.lcs.mit.edu
In-Reply-To: Lars Balker Rasmussen's message of 18 Oct 1996 17:58:16 +0200 <0f6848nu1z.fsf@fraxinus.daimi.aau.dk>
Subject: Hi and problem
I'm not sure I agree that
f_t /*@checkedfornull@*/ f[] = { ... };
would be less arcane than,
typedef /*@null@*/ f_t nf_t;
nf_t f[] = { ... } ;
I think the Standard C const pointer syntax is confusing and even
experienced programmers often have a hard time figuring out exactly what
deeply nested * const * constructions mean, and I wouldn't want lclint
annotations to add to this confusion. By using the typedef, its clear
exactly what the null applies to. Its true that this might seem a bit
strange to a maintainer who isn't familiar with lclint, but they are
going to wonder about all those silly-looking comments anyway!
--- Dave
David
Evans
University of Virginia, Computer Science
evans@cs.virginia.edu