FreeTDS Development Group

[entropy AT freetds.org]

James K. Lowden wrote:
> I have an opinion I want to check with the old hands out there, because
> the GNU C compiler issues a warning I think should be ignored.  
>
> Given code like this:
>
>         if (a && b || c && d)
>                 /* ... */
>
> the compiler "recommends" parentheses, encouraging:
>
>         if ((a && b) || (c && d))
>                 /* ... */
>
> Please.  To he who would read the code, I say RTFM.  If K&R is too long
> for you, here's a short version:

The compiler doesn't need the parens, but humans do.  If precedence were 
as simple and obvious as you claim, you wouldn't still be finding 
precedence bugs like

>         if((ret=func() != NULL))
> instead of
>         if((ret=func()) != NULL)

in such a mature production code base.

> is a good one.  I fixed one like that last week.  So is:
>
>         memset( ptr, len, '\0')
>
> a favorite of mine, not least because '\0' is silly: the value argument is
> defined as int.

Well the value should go in argument 2, but I don't see any problem with 
'\0'.  Would you think "memset(ptr, 'x', 23)" was silly?  Would 
"memset(ptr, 120, 23)" be clearer?  Specifying '\0' makes it explicit 
that "this is a character", and making such things explicit makes code 
easier to read.  IMNSHO, of course.

> Am I missing something?  Are there compilers out there that would get this
> wrong?  Do the parentheses give the compiler information it doesn't
> otherwise have?  Or is this just nanny nagging we can turn off by default?

You can probably turn it off, but I don't see anything to be gained from 
doing that.  I like the parens, and I easily get a headache from reading 
too much smart-ass code written in the "lean and mean" style without 
them.  However, this might just be evidence that I have some kind of 
brain damage from too much Scheme programming in my formative years.  YMMV.

--
Cheers,
entropy