Re: [PATCH 1/3] evaluate: warn on identical exprs around '&&'

[Josh Triplett <josh@joshtriplett.org>]

On Sun, Aug 28, 2011 at 02:57:36PM +1200, Chris Forbes wrote:
> > Also, how does your patch handle expressions like this: *x++ && *x++
> > Or this: f() && f()
> 
> Those cases will both warn. Do I need to be looking for potential side
> effects, and considering these expressions "probably correct" in those
> cases?

Those expressions can certainly make sense in correct code, unlike cases
with identical side-effect-free expressions; that doesn't make such
expressions a good idea, though.  You might want to check for
side-effect-free code (I think Sparse already has a mechanism to test
for that), and split that into a separate warning option.  Whether that
option should stay on by default or not, I don't know; that will take
running it over some real codebases and evaluating the benefit.

- Josh Triplett