On 01/07/2010 11:42 AM, Richard Guenther wrote: > On Thu, Jan 7, 2010 at 12:29 PM, Andrew Haley <a...@redhat.com> wrote: >> On 01/06/2010 07:24 PM, Joshua Haberman wrote: >>
>> The response I think you may get is that this is already settled. >> I've seen this same argument many times (in mailing lists and in >> comp.lang.c) over the last decade or so, and the result has always >> been the same. > > But I still can't find proper language in the standard that restricts > pointer dereference nor pointer conversion (apart from the alignment > issue, but that's only where it constraints bijectivity of pointer > conversion). In every discussion that I have seen, the wording in 6.3.2.3 Para 7 has been deemed to be sufficient to constrain pointer conversion in all cases: the wording there has always been interpreted as "these are the only things you have permission to do; everything else is forbidden." It may be that there is some other way to interpret that section, but I doubt that the TC will take such a view. Perhaps a non-normative footnote would help. Andrew.