On Tue, May 17, 2005 at 09:12:38AM -0700, Joe Buck wrote: > On Tue, May 17, 2005 at 12:00:59PM -0400, Paul Koning wrote: > > I'm upgrading to V4.0.0 and struggling with some code that's seriously > > into templates. One puzzling error is this one: > > > > keyed_obj.hh:159: error: no matching function for call to > > 'CxnIndex::CxnIndex(CxnIndex)' > > Indeces.hh:150: note: candidates are: CxnIndex::CxnIndex(CxnIndex&) > > Indeces.hh:145: note: ... and some more > > > > It's not entirely clear to me why something very much like a copy > > constructor is being invoked in the first place. But the bigger > > puzzle is: why isn't the copy constructor being matched? I would have > > thought that any call with a T argument should match a function or > > method with an &T argument in its declaration... > > CxnIndex::CxnIndex(CxnIndex&) is not a general copy constructor. Notice > the non-const reference. The error message is saying that it's trying > to pass a temporary to a copy constructor. You would need > > CxnIndex::CxnIndex(const CxnIndex&) > > to match that. As for why you need a copy constructor, we'd have to > see the code for that, but this list isn't really the right place for > helping you debug this problem further.
One GCC-related point is possibly worth mentioning: GCC is now much stricter about enforcing the rule that a copy constructor must be accessible even if the copy will be elided. See the last entry of http://gcc.gnu.org/bugs.html#nonbugs_cxx jon
