On Tue, 2003-02-18 at 10:07, Tom Lane wrote:
> Neil Conway <neilc@samurai.com> writes:
> > Some minor nit-picking follows below. I tend to be a bit of a
> > style-nazi, don't mind me :-)
>
> > - the length argument to snprintf() includes the terminating NUL byte,
> > so code like pg_avd.c line 334 is off-by-one:
>
> > char buf[256];
> > /* ... */
> > snprintf(buf,255,"...");
>
> Actually the preferred coding of this is
>
> snprintf(buf, sizeof(buf), ...);
>
> which is both correct and impervious to subsequent alteration of the
> declared size of buf. I get antsy whenever I see a snprintf with a
> literal-constant size argument, because it's a mistake waiting to
> happen. Use sizeof() when you can, or at least a #define.
OK, I will make those changes in the next version along with all the
changes mentioned in the other emails I have received. I will try to
have it in the next two days or so.
Matthew
