On Sun, 29 Apr 2012 12:50:44 -0400 Robert Dewar <de...@adacore.com> wrote:
> On 4/29/2012 12:47 PM, Basile Starynkevitch wrote: > > > My biased point of view is that designing a processor instruction set (for > > POSIX-like > > systems or standard C software in mind) with function pointers of different > > size than > > data pointers is today a mistake: most software make the implicit > > assumption that all > > pointers have the same size. > > What's your data for "most" here? I would have guessed that most > software doesn't care. `dlsym` is the obvious hint (including the fact that the `dlfsym` proposal vanished), and also simply that most (probably nearly all) Linux/ELF systems and Unix systems have same size for data and function pointers. Also, because while coding since 1974 (teenager then) I never personally got a system where they have been different -with the important exception of some programming models for early PC/AT (and PC/286) systems of the MSDOS area (e.g. Borland Turbo C v1). IIRC correctly, CrayY/MP have various size for some data pointers so it had sizeof(char*) != sizeof (double*) but the function pointers had the size of some integral size which was the size of some data pointers (IIRC sizeof(void(*)()) == sizeof(long*) == sizeof(long) on that machine). And I did wrote that designing an ISA *today* for *Posix-like* systems [emphasis both on today and on Posix/Unix] with such a major difference would (IMHO) be a mistake. For instance, I don't think that porting the Linux kernel (or the FreeBSD one) to such an architecture (having data pointers of different size that function pointers) is easy. And GTK wants nearly all pointers to be gpointer-s, and may cast them to function pointers internally. Regards. -- Basile STARYNKEVITCH http://starynkevitch.net/Basile/ email: basile<at>starynkevitch<dot>net mobile: +33 6 8501 2359 8, rue de la Faiencerie, 92340 Bourg La Reine, France *** opinions {are only mine, sont seulement les miennes} ***
