🚀 go-pugleaf

Stuart Maclean <S.D.Maclean@ecs.soton.ac.uk> schrieb mir
zur gleichen Anfrage in comp.lang.c++:

: In article <40cm1r$oej@fbi-news.informatik.uni-dortmund.de>,
: warda@veronica.informatik.uni-dortmund.de (Armin M. Warda) wrote:

:  >  Hello,
:  >
:  >is there a possibility to find out within a method who 
:  >called the method?
:  >
:  >I could explicitly give the 'this'-pointer as parameter
:  >when calling a method:
 
 
: Isn't it neater when you dont have to explicitly pass this around 8-)
 
: Unfortunately, if YOU don't supply the info, the runtime system will have
: to do more work to get the caller object. What with looking at the stack
: and stuff, you're going to need to know your architecture/compiler quite
: well.
 
: Here's what I do to identify the calling object on the SPARC, compiler: GCC
 
: .text
:    .align 4
:    .global _caller
: _caller:
:    mov %g1, %o1      ! %o1 is scratch, so OK to save %g1 there
: 
:    restore           ! back to caller's register window (cached stack frame)
: 
:    mov %i0, %g1      ! GCC passes the 'this' pointer in %o0, which becomes
:                      ! %i0 in the new window. Grab it into %g1
: 
:    save              ! back to the original window
: 
:    mov %g1, %o0      ! return value (ie the calling object) returned in %o0
: 
:    retl              ! return to caller. Note this function is a LEAF !!!
: 
:    mov %o1, %g1      ! restore clobbered %g1.
:                      ! This is the delay slot instruction.
: 
: ! end of _caller
 
 
: Note that this ASSUMES that there was indeed an object address in %i0 of
: the previous stack frame.  In other words,  we assume that the calling
: function was indeed a non-static member function, not an ordinary function
: or static member function.
 
: Then use as such
 
: class Foo {
: public:
:    void Bar();
: };
: 
: void Foo::Bar()
: {
:    if ( caller() == this ) {
:       this is an intra-object call....
: }
 
 
: NOTE : As per usual when comparing object equality by way of address
: equality, beware of multiple inheritance. For a discussion, see Meyers,
: item 17, page 57.
: But who ever uses MI anyway ;-)

: Hope this helps. Any comments welcomed.
 
: Later,
: Stuart

A.
-- 
-------------------------------------------------------------------------------
Armin M.Warda   <http://ls4-www.informatik.uni-dortmund.de/QM/MA/warda/pi.html>
Universitaet Dortmund,  Fachbereich Informatik, Lehrstuhl IV,  D-44221 Dortmund