Re: Does C++ allow you to define explicit casts?

Ben> What I don't want is to be able to pass one of them to a Ben> function expecting the other -- but it *should* be possible Ben> with an explicit cast, e.g. (Bytecount) x. Problem is, the Ben> "cast operator" that can be defined on a class seems to Ben> define implicit casts, not explicit. Is there any way to do Ben> what I want? I think this just wins as long as you make "just enough" effort. Make user-defined conversions between ints and *counts, back and forth, but _not_ between Bytecount and Charcount. This program does what you want under g++ 2.95.4: ------------------------------- cut ------------------------------- typedef long int EMACSINT; // I don't think these can be typedefs of instances of a common template // class (they'd be the same class, then). Too lazy to try. But you could // get the effect you want with a macro, of course. class Charcount { EMACSINT data; public: Charcount (EMACSINT i) { data = i; } operator EMACSINT () { return data; } }; class Bytecount { EMACSINT data; public: Bytecount (EMACSINT i) { data = i; } operator EMACSINT () { return data; } }; extern void f (Bytecount b); int main () { Charcount c = 1; f (Bytecount (c)); // wins, but ... f (c); // *loses* } ------------------------------- cut ------------------------------- The second call to f() loses because the interpretation must be f (Bytecount (Charcount::operator int (c))); which is rejected because both user-defined conversions are implicit. The error is test.cc:22: conversion from `Charcount' to non-scalar type `Bytecount' requested There is an arbitrary restriction to at most one *implicit* user-defined conversion in a conversion chain. The first call f (Bytecount (c)); however, *wins* because now only the call to Charcount::operator int is implicit. Note, I don't have anything more recent than about 10 years old at hand, either (Stroustrup and Ellis, 1e, and it's at home, to boot, so everything above is ISTR). So you should check with a more recent version of the standard (eg, Buchholz [2002] :^) to be sure. Not to mention the compiler ;-) -- Institute of Policy and Planning Sciences http://turnbull.sk.tsukuba.ac.jp University of Tsukuba Tennodai 1-1-1 Tsukuba 305-8573 JAPAN Don't ask how you can "do" free software business; ask what your business can "do for" free software.