Re: libpq5 8.3 breaks 8.2 compatibility with encodings

Peter Eisentraut <peter_e@gmx.net> writes:
> Perhaps we should leave a gap where JOHAB used to be and add it at the end.
> Otherwise we may have to do a soname exercise.

The question to me is whether the encoding numbers used by libpq
should be considered anything other than black-box numbers.  initdb
is arguably assuming more than it ought to about what they mean.

            regards, tom lane