This is the mail archive of the email@example.com mailing list for the libstdc++ project. See the libstdc++ home page for more information.
According to Jason Merrill: > >>>>> Alexandre Oliva <firstname.lastname@example.org> writes: > > > Within the scope of S<T>, `a', `S::a' and `S<T>::a' are all the same > > thing > > Nope. S::a and S<T>::a are the same, but they are dependent, and `a' is > not. g++ doesn't currently implement that consistently. Wouldn't you agree that it should? As Stroustrup points out, generating warnings or errors for *potential* inconsistencies is usually the wrong thing. Rather, only when *actual* inconsistencies arise -- in this case, at instantiation -- should the potential error be flagged as an actual error, or else accepted if it's not wrong. -- Chip Salzenberg - a.k.a. - <email@example.com> "There -- we made them swerve slightly!" //MST3K