This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: Signed int overflow behaviour in the security context

On Sat, 2007-01-27 at 02:47 +0100, Andreas Bogk wrote:

> I'm exactly talking about the semantics of "undefined" here.  It would
> be immensely reassuring if a compiler would at least interpret this as
> "unspecified, but consistent".  Even better would be a defined and
> documented semantics for everything left undefined by the standard.

I have read the admonitions to have this thread end with the good
final words about what is causing all this: a language issue. Still,
what you are saying here is that a compiler should decide
that the meaning of "undefined" be defined.

If you know what you want to happen on your microprocessor, it seems
much simpler to just give it suitable instructions (you know how)
than to add a decision tree of GCC switches that will essentially
establish another non-portable language in itself, in a most
complicated way (as per the suggestion of Paul Schlie, considering
the number of ifs in the compiler control language I see there.)

Do you C the facts? ;-)


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]