Re: [Valgrind-developers] valgrind: r8311 - branches/FORMATCHECK/coregrind

[Julian S. <js...@ac...>]

On Wednesday 02 July 2008 14:08, Bart Van Assche wrote:
> On Tue, Jul 1, 2008 at 10:02 AM, Dirk Mueller <dm...@gm...> wrote:
> > VG_(printf) uses "%t" as a "xml-escaped %s" attribute. This does not if
> > gcc format checking is enabled, as %t is a nonstandard extension, which
> > it doesn't understand. I've asked you some time ago wether it would be
> > okay to either map it to "%s" and use a ToXml function for the argument,
> > or to use a completely different (non printf-like) function for xml
> > output. I guess the latter makes more sense, as line formatting (printf)
> > doesn't really make sense with xml output, the stream output can be
> > abstracted into something else.

I suppose it would be a good thing to move towards separating the XML
messaging completely.  So, yes, the latter sounds better to me too.

> How about the code below ? The patch below defines two variants of the
> same function, namely VG_(xml_message)() and VG_(message)(). Format
> string checking is disabled for the former and enabled for the latter.
> This way the %t format specifier does not trigger a warning when
> passed to VG_(xml_message)().

Am I right to understand that VG_(xml_message)() and VG_(message) are
functionally identical, or is that not so?

J