(I'm back on-line, I think...)
I committed several changes to the development source tree this morning
(~15 hours ago). The changes affect the following areas:
1) There is an 8-byte integer data type. It needs some testing and
possibly configuration help on most of the supported platforms. Works on
i686/Linux and should work on Alpha.
2) pg_dump now surrounds table and column names with double-quotes, to
preserve case and funny characters through a dump/reload operation. Hope
this is OK with you Bruce; let me know... btw, last time I tested this
code (two weeks ago?) it was still slightly off of a perfect dump/reload
of the regression tests. The test I am doing is to dump the regression
test database, then reload that into a new database, then dump the new
database. The resulting pg_dump output should be the same as the dump of
the original database.
3) some docs sources have been updated.
4) some additional regression tests have been defined to cover the
HAVING clause and the int8 data type.
5) automatic data type conversion now happens in every place it needs
to, I think. The last changes are to get source columns to match target
columns in simple INSERT/FROM statements.
Except for the "random" and "resjunk" regression tests, things look good
on my development machine; I've done a build and regression test
directly from the CVS source tree after these changes with no other
errors noted.
- Tom