PostgreSQL v7.0 - Release Candidate 2

In order to allow for testing of those changes implemented since RC1 was
put out last week, we've just created a Release Candidate 2. The only
change(s) between this Release Candidate and the full, Final Release, is
the doc updates that Thomas Lockhart is currently working on ...

This release is once more available as:

ftp://ftp.postgresql.org/pub/postgresql-7.0RC2.tar.gz

Or multi-parts, as:

ftp://ftp.postgresql.org/pub/postgresql-7.0RC2.base.tar.gz
ftp://ftp.postgresql.org/pub/postgresql-7.0RC2.support.tar.gz
ftp://ftp.postgresql.org/pub/postgresql-7.0RC2.test.tar.gz
ftp://ftp.postgresql.org/pub/postgresql-7.0RC2.docs.tar.gz

Marc G. Fournier ICQ#7615664 IRC Nick: Scrappy
Systems Administrator @ hub.org
primary: scrappy(at)hub(dot)org secondary: scrappy(at){freebsd|postgresql}.org

---------- Forwarded message ----------
Date: Sat, 29 Apr 2000 14:08:51 -0400 (EDT)
From: PostgreSQL Main <pgsql(at)mail(dot)hub(dot)org>
To: scrappy(at)mail(dot)hub(dot)org
Subject: PostgreSQL v7.0 - Release Candidate 2

- More libpgeasy update of connectdb() parameter ordering
- Update pgeasy examples
- Aix patches from Zeugswetter Andrea
- Fix libpq example return values
- List of available pg_options items was a tad out-of-date.
- Update COPY manual page for \N.
- Fix \N mention in manual
- Update obsolete info in CREATE INDEX ref page.
- Clean up description of SET CONSTRAINTS.
- change reindex ERROR/NOTICE message
- Check that user-specified opclass in CREATE INDEX corresponds to operators
that will actually work on the column datatype.
- Produce an appropriate error message when opclass is not supported by
specified index access method. Clean up wording of some existing error
messages, too.
- Correct oversight in hashjoin cost estimation: nodeHash sizes its hash
table for an average of NTUP_PER_BUCKET tuples/bucket, but cost_hashjoin
was assuming a target load of one tuple/bucket. This was causing a
noticeable underestimate of hashjoin costs.
- Tweak outUnique to include uniqColIdx[] field in the printout. This does
not cause any compatibility problems because stored rules don't contain
plan nodes --- in fact, we don't even have a readfunc for Unique nodes.
- Repair problem noted by Elphick: make_rels_by_joins failed to handle
cases where joinclauses were present but some joins have to be made
by cartesian-product join anyway.
- Repair coredump seen when a view refers to an inheritance group
(SELECT FROM table*).
- On HPUX, shl_load should be called with options BIND_IMMEDIATE rather
than BIND_DEFERRED.
- -D switch to postmaster should override any PGDATA environment variable
that might be hanging about.
- Except_Intersect_Rewrite() failed to ignore resjunk targetlist entries,
- Correct error in rewriter that caused SELECT count(*) FROM view
to give wrong results:
- Also, make 'modified' flag be local to ApplyRetrieveRule:
- Setting statistic options from SET PG_OPTIONS caused a backend crash
because StatFp never got set in that case.
- 7.0 buffer manager can support different backends running with different
fsync settings, so the -F option no longer needs to be treated as secure.
- Clean up const-vs-not-const compiler warning in MULTIBYTE code.
- initdb didn't always remove temp file
- Update pg_ctl so that it does not redirect outputs from postmaster
to a temp file.
- Fix \h to not go past array bounds
- Fix still more static-declaration-vs-nonstatic-definition glitches.
- Our test to see if we had permission to install into Perl5 install area
always failed if Perl makefile's INSTALLSITELIB variable was specified
in terms of another variable. Fix by adding an echo-installdir target
to the Perl makefile, which the upper-level Makefile can invoke.
- Allow libpq++ compile failure to stop entire compile.
- Make ECPGraise's str parameter const to suppress warnings from gcc
and errors from pickier compilers.
- Clean up ecpg test files.
- Remove DriverClass.java. It is generated by the compile.
- Update libpgeasy readme.
- Change libpgeasy to take dbname at end like all other interfaces.
- Remove broken tracing code (which would be dangerous if it did work...)
- Make c++ examples return 0 from main().
- Update example: PgLargeObject constructor now takes a conninfo string,
not a bare database name.
- Security fix for plperl.
- plpgsql RAISE statement was careless about the possibility of a NULL
field value being displayed;
- pltcl didn't work well at all when Tcl had been built with a different
compiler than the one selected to build Postgres with.
- Remove -qhalt=w flag from aix42.
- Finally figured out that HP's cpp won't do ANSI preprocessing constructs
unless you feed it -Aa or -Ae switch.
- Add a regress test case for SELECT count(*) FROM view, so that we'll
know if that case ever breaks again...