pt ctor ordering/missing ctor

[Jason Merrill]

>>>>> Alexandre Oliva <oliva@dcc.unicamp.br> writes:

 >> But, we could use the point-of-instantiation (in the technical
 >> standardese sense) to at least partially order these things, if we
 >> thought that was worthwhile.  That would mirror the order I suggested
 >> for explicit instantiations.  If two things have the same point of
 >> instantiation, I don't see any basis for choosing one over the other.

 > The one referred first?

 > Should I try to design test cases for the testsuite to test these
 > ordering properties?

IMO, no.  I don't think the ordering is important.

Jason