|
Boost : |
From: Ullrich Koethe (koethe_at_[hidden])
Date: 2006-10-19 14:34:07
Hi again,
since there was no response to my earlier questions about undefined
behavior, I'd like to bring this up again:
GIL makes extensive use of reinterpret_cast and gil_reinterpret_cast. The
former is, for example, used in code similar to the following:
short * imagedata = new short[3*image_size];
rgb16s_t * rgbdata = (rgb16s_t *)imagedata;
gil_reinterpret_cast is defined as
template <class OutPtr, class In>
OutPtr gil_reinterpret_cast(In * p)
{
return static_cast<OutPtr>(static_cast<void *>(p));
}
I suspect that undefined behavior results in both cases. IIRC, there is no
guarantee that the memory layout of an rgb array is equal to that of a
corresponding scalar array with three times as many elements (although
this will probably work in practice). Likewise, a static_cast from void *
is only allowed back to the original pointer type, not to a different type.
Can someone more familiar with the fine points of C++ confirm these
suspicions? If they are true, I think the issue should be resolved before
inclusion into boost.
Regards
Ulli
-- ________________________________________________________________ | | | Ullrich Koethe Universitaet Hamburg / University of Hamburg | | FB Informatik / Dept. of Informatics | | AB Kognitive Systeme / Cognitive Systems Group | | | | Phone: +49 (0)40 42883-2573 Vogt-Koelln-Str. 30 | | Fax: +49 (0)40 42883-2572 D - 22527 Hamburg | | Email: u.koethe_at_[hidden] Germany | | koethe_at_[hidden] | | WWW: http://kogs-www.informatik.uni-hamburg.de/~koethe/ | |________________________________________________________________|
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk