------- Comment #2 from Curatica at gmail dot com 2009-12-25 02:04 -------
Please, understand that for me this is just a disinterested, academic
discussion: no offense. I am not sure that I agree with the theory.
The standard (8.5.1) states that:
T x = a;
is a "copy-initialization" but does not infer that this has anything to do with
the copy constructor. If I "fix" the source by adding "const" to the copy c-tor
argument:
Base( const Base& b ) { /*...*/ }
the code compiles indeed but there is no evidence whatsoever that the copy
c-tor is being invoked when the code
Base b2 = 5;
is executed (as one could see, the copy c-tor prints a specific text to stdout,
which does not appear in this case). Furthermore, if I take out the copy c-tor
altogether, the code compiles and executes correctly. In either case, it seems
clear to me that
Base( int a ) { /*...*/ }
is invoked. This is in accordance to "12.3.1 Conversion by constructor" in the
standard [class.conv.ctor].
My guess is that different people interpreted the standard differently: the
machine code is generated correctly in all cases but the compilation error
which depends on the "constatness" or lack thereof of the copy c-tor argument
is bogus.
--
Curatica at gmail dot com changed:
What |Removed |Added
----------------------------------------------------------------------------
Status|RESOLVED |UNCONFIRMED
Resolution|INVALID |
http://gcc.gnu.org/bugzilla/show_bug.cgi?id=42470
