Boost C++ Libraries: Ticket #1301: copy-initialization from const-qualified object doesn't work

attachment set

Fri, 05 Oct 2007 16:58:02 GMT

attachment → optional_init_bug.cpp

a failed test

status changed

Fernando Cacciola — Fri, 02 Nov 2007 23:45:56 GMT

status new → assigned

I think it's a gcc bug. In any case I added the missing test case to see how other compilers handle it.

Sat, 03 Nov 2007 08:29:34 GMT

Replying to fcacciola:

I think it's a gcc bug. In any case I added the missing test case to see how other compilers handle it.

Thanks for adding a test into optional_test_ref.cpp. BTW, optional<int const&> orci2 = ci ; is a typo of optional<int const&> orci2(ci) ; ?

status changed; resolution set

Fernando Cacciola — Tue, 06 Nov 2007 22:07:48 GMT

status assigned → closed
resolution → wontfix

Replying to Shunsuke Sogame <pstade.mb@gmail.com>:

Replying to fcacciola:

I think it's a gcc bug. In any case I added the missing test case to see how other compilers handle it.

Thanks for adding a test into optional_test_ref.cpp.

:) BTW, I noticed that you have updated a test case only when I replied to the ticket, after I already added my own test case.

BTW, optional<int const&> orci2 = ci ; is a typo of optional<int const&> orci2(ci) ; ?

Oops, of course.

BTW. I confimed this is a gcc bug as this simple program shows it:

struct crw {

crw ( int const& _cref ) : cref(_cref) {}

int const& cref ;

} ;

int main( int, char* [] ) {

int const ci = 0 ; crw copy_init(ci); assert( &copy_init.cref == &ci ) ;

crw direct_init = ci ; assert( &direct_init.cref == &ci ) ; This one fails here

}

I'm marking the new failures as expected.