Re: AC_CHECK_TOOL changes, take 2

[Ralf Corsepius]

On Fri, 2004-03-19 at 22:39, Kevin Ryde wrote:
> bonzini <address@hidden> writes:
> >
> >  The only case when this may be
> > +useful is when you really are not cross-compiling, but only
> > +building for a least-common-denominator architecture: an example
> > +is building for @code{i386-pc-linux-gnu} while running on an
> > address@hidden architecture.
> 
> The upward case is also useful (but how often it's used I don't know).
Well, I strongly disagree.

Trying to use an i686-pc-linux-gnu toolchain targeting i386-pc-linux-gnu
is a nice example for that:

An i686-pc-linux-gnu toolchain is not able to generate i386 compatible
code nor is the libc nor are the systems other libraries being used by
i686-pc-linux-gnu toolchains compatible to the i386. The excutable will
contain various instructions which are not available on i386s.

An example in gcc's sources where this becomes clearly exposed it
libstdc++'s atomicity.h: There exist i486 and i386 versions.
The i486 versions is incompatible to the i386.

Try using i686-pc-linux-gnu for generating executables for original
Intel i386 CPUs and you will experience what I said. The same applies to
various other architectures and OS. 

Ralf