Names v. field access

Tue Sep 30 23:30:00 GMT 2003

>>>>> "Fernando" == Fernando Lozano <fernando@lozano.eti.br> writes:

Fernando> This is risky. The market recognizes Sun implementation as
Fernando> "the" java language, and having a different behaviour may
Fernando> create portability problems, and no one will acknowledge Sun
Fernando> is wrong. At least make this an option.
In this case it is really unlikely that any code relies on this bug.
And, if it does, the fix is trivial and obvious.
Furthermore, jikes rejects this code, so there is precedent out there.
>From my perspective, the language isn't like the libraries. The
language has a specification which can actually be used to describe
the language. The libraries don't really have this; with the
libraries we're forced into bug compatibility. I'd prefer us to be
more strict about the language.
If, in some situation, bug-compatibility is warranted, then let's add
an option. But let's default to the equivalent of -pedantic.
Tom