Re: [PATCH] - improve sprintf buffer overflow detection (middle-end/49905)

[Martin Sebor <msebor at gmail dot com>]

On 09/21/2016 08:37 AM, Alexander Monakov wrote:
> On Wed, 21 Sep 2016, Martin Sebor wrote:
> > On 09/21/2016 01:11 AM, Markus Trippelsdorf wrote:
> > > The patch uses "nul" instead of "null" throughout.
> >
> > Yes, that's intentional.  NUL and null are alternate spellings for
> > the same character. I went with nul to distinguish it from the null
> > pointer and used all lowercase as per the GCC convention.
>
> Can you elaborate which guideline suggests spelling that in lowercase?
> It seems quite strange to me, especially given that the documentation
> added with the patch uses "NUL character" (which I believe to be a more
> common form), but then warnings use "nul" (without the "character" iiuc).

I don't know if there is a specific guideline that applies here,
but it's my impression that conventionally GCC messages use
lowercase letters.  The only guideline I know about is in the GNU
Coding Standard that calls for messages not to begin with capital
letters (or end in a period;  though now that I've checked more
carefully there are a fair number of messages that ignore that
suggestion).

That said, if the lowercase nul is offensive it can certainly be
changed to some other spelling (as can any of the other messages
emitted by the pass).

Martin

PS Yes, the C and C++ standards refer to NUL as the null character.
NUL is an abbreviation, used extensively (for example) by POSIX.