John Polstra writes:
In article <[EMAIL PROTECTED]>, Dag-Erling Smorgrav <[EMAIL PROTECTED]> wrote:
This is wrong. caddr_t should be uniersally replaced with void *.
Not quite. There is (or at least used to be) a lot of code that assumed you could do address arithmetic on a caddr_t. You can't do that on a void *, at least not in ANSI C. I think gcc lets you do it, but it's an extension.
As I have discovered. I specifically looked for this, and my misreading of the spec is now clear. :-)
Yes, but relying on this during fixing out caddr_t may break use of other compilers.
Now, that OpenWatcom is released, the FreeBSd port of it should follow. And maybe someone will try to compile the kernel and world with it. If that would work, this would be great, because the watcom compiler generates much better code than gcc does, even than gcc -O3 (and all known optimizations on).
So long, Jens
To Unsubscribe: send mail to [EMAIL PROTECTED] with "unsubscribe freebsd-current" in the body of the message
