Hi,

On Fri, 2008-06-27 at 14:13 +0200, Ákos Balázs wrote:
> Hi Gerrit, Carsten,
> 
> sorry for the late reply, I was a bit busy in the last week.
> 
> On Mon, Jun 16, 2008 at 5:14 AM, Gerrit Voss <[EMAIL PROTECTED]> wrote:
> > yeah my fault ;-), but I treat reinterpret_casts quite carefully,
> > especially since gcc started to use structural information
> > to optimise. So that's why some static_casts are still there
> > if I did not hit a compiler complain. This is a little bit optimistic
> > for template code, especially if I miss to build some of the tests
> > as in this case ;-(
> 
> What do you mean by using structural information? I thought
> reinterpret_casts by definition only change the type of the object
> pointed to, but perform no implicit conversions whatsoever. Can you
> give me a URL or other such pointer where this gcc optimization is
> described?

as far as I looked into it has to do with aliasing assumptions, in 
particular the fact that the standard allows the compiler to assume 
(with -O3 and -fstrict-aliasing) that pointers of different types can
not point to the same address. At one point alignment seems to enter
the game too.

google for 

warning: dereferencing type-punned pointer will break strict-aliasing

which is the warning gcc gives if it detects this kind of situation.

kind regards,
   gerrit



-------------------------------------------------------------------------
Check out the new SourceForge.net Marketplace.
It's the best place to buy or sell services for
just about anything Open Source.
http://sourceforge.net/services/buy/index.php
_______________________________________________
Opensg-users mailing list
[email protected]
https://lists.sourceforge.net/lists/listinfo/opensg-users

Reply via email to