neath@solar-1.stars.flab.Fujitsu.JUNET (05/03/89)
A recent posting to this group discussed the use of typedef to specify a type
that is a `pointer to a member function of class X'. Can this be extended to
include a reference to a member function? If not, why not? Specifically, the
following code fragment results in a "internal <<cfront 1.2.1 2/16/87>> error:
bus error" message:
class A {
int a;
public:
A(int n) {a = n;};
void print() {cout << a;};
};
typedef void A::MEMF;
MEMF* p1 = A:print; // This is accepted by cfront
MEMF& p2 = *A:print; // This results in bus error!
What I am really trying to do is see if there is a mechanism that can be used
to implement features similar to BEFORE, AFTER, and AROUND methods found in
Lisp. If I can get access to the pointer maintaining the link to the member
function of a class, I can change this to point to some function implementing
the appropriate method behavior. This functionality greatly facilitates code
reuse, since it becomes much easier to take an existing class and make a slight
modification to it without having to derive a new class from it. Any comments?
Regards
Martin Neath <neath@dsg.ti.com>
------------------------------------------------------------------------
DISCLAIMER: As always, the opinions expressed above are strictly my own
and do not reflect those of my employer, Texas Instruments.
------------------------------------------------------------------------
--
Regards,
Martinmat@mole-end.UUCP (Mark A Terribile) (05/05/89)
> typedef void A::MEMF; > MEMF* p1 = A:print; // This is accepted by cfront > MEMF& p2 = *A:print; // This results in bus error! Eh? I'm surprised that cfront accepts the first line. Can you explain to me how this is supposed to be a valid statement in the language? > ... If I can get access to the pointer maintaining the link to the member > function of a class, I can change this to point to some function implementing > the appropriate method behavior. This functionality greatly facilitates code > reuse, since it becomes much easier to take an existing class and make a > slight modification to it without having to derive a new class from it. > Any comments? Yeah. If you have source code, why patch the object? COBOL has the ALTER statement: A: GO TO B. . . . ALTER A TO C. Rewrites the goto at A as ``GOTO C'' . The ALTER statement is generally regarded as one of the best ways to rapidly make code maintainance hopeless. I can't see any difference between the ALTER statement and what you are trying to do. In both cases, you are subverting the source-code-to-executable translation by making changes at execution time. What's wrong with deriving classes? How does patching in at execution time facilitate code reuse better than using the source code mechanisms that are provided explicitly for the purpose? -- (This man's opinions are his own.) From mole-end Mark Terribile