IMHO what Per wanted to point out was, that these functions - which might be perhaps called intrinsic as they are supplied together with the compiler and sometimes parts of the "standard library" - are sometimes implemented as macros in the "standard headers" (i.e. compiled together with the application, not compiled into the compilator's binary) - and this fact might indicate that in compilers which use such constructs, the "real inline assembler" might be safe.

An explicit assurance in the manual would be much better, though.

JW