* Gregory Stark (stark@enterprisedb.com) wrote:
> Hm, I suppose this is expected. I always hated the idea that libraries could
> introduce new symbols without an soname bump but obviously nobody's going to
> be interested in an soname bump today...
>
> ! /home/stark/src/local-HEAD/pgsql/src/test/regress/./tmp_check/install//usr/local/pgsql/bin/psql: symbol lookup
error:/home/stark/src/local-HEAD/pgsql/src/test/regress/./tmp_check/install//usr/local/pgsql/bin/psql: undefined
symbol:pg_valid_server_encoding_id
Just to clarify... You compiled a new program which uses the new
symbols against new headers (otherwise they would have complained during
compilation) and then tried to use that new program with an old library.
If so, then yes, that's expected to break.
Thanks, Stephen