Backport from upstream with this commit:

    commit bc8036fc
    Author: Tom Lane <tgl@sss.pgh.pa.us>
    Date:   Fri May 11 17:57:14 2007 +0000

        Support arrays of composite types, including the rowtypes of regular tables
        and views (but not system catalogs, nor sequences or toast tables).  Get rid
        of the hardwired convention that a type's array type is named exactly "_type",
        instead using a new column pg_type.typarray to provide the linkage.  (It still
        will be named "_type", though, except in odd corner cases such as
        maximum-length type names.)

        Along the way, make tracking of owner and schema dependencies for types more
        uniform: a type directly created by the user has these dependencies, while a
        table rowtype or auto-generated array type does not have them, but depends on
        its parent object instead.

        David Fetter, Andrew Dunstan, Tom Lane

Signed-off-by: NGeorge Caragea <gcaragea@pivotal.io>

* Fix stat check for older behave test:
    Used to rely on gp_restore to do the stat update, but is now on
    python.
* Fix the tests to indiviudally check the statistics for each
partition table as the top level table does not gather statistics from its
subpartitions.
* Add specific gpdbrestore -T stat update check
* Add specific gpdbrestore -S stat update check

Add unit tests for generating the correct tables to restore.

Authors: Marbin Tan & Chris Hajas

When doing a restore of a specific table or a specific schema (-T or -S)
with a full backup, it will update all statistics for all AO/CO tables; it should
update only the specified table/schema.

Moved gp_update_ao_master_stats from the C side to the python code.
gpdbrestore -T and -S will now update AO/CO stats through the python code

Authors: Marbin Tan & Chris Hajas

file.

Authors: Marbin Tan, Chris Hajas and Pencheng Tang

This brings the code in convert_IN_to_join() back the way it is in the
upstream, making merging easier.

And some other cosmetic cleanup to match upstream better.

This makes merging easier.

catullus.pl is no longer used to generate pg_type.h, since commit 30c6ae3e.
Update the instructions accordinly. The description of how pg_proc.h works
was slightly out-of-date as well, updated that too.

When we use error log, if the master finds a badly formatted data row, it
increments the rejected row count and then sends the row to the segment so that
it can be stored in the error log file. On the segments, the row gets parsed
again and the segment increments the reject count again.  For reporting the
total number of rejected rows to the user we sum up the rejected row count from
the master and the segments. We need to ignore the count from the master
because it will be included in the reject count from the segments.

Also includes additional cleanup and typo fix of variable name by Heikki
Linnakangas.

Most of these test additions are inspired from Pivotal's internal
testing and needed to be added to the open source installcheck to
give the community more test coverage on AO/CO tables.  This commit
mostly adds extra coverage for indexes and partition tables.

based on patch by Daniel Gustafsson

We only have a few extra types in GPDB, so it doesn't seem worth maintaining
the catullus mechanism for updating pg_type.h. When a new column is added
in the upstream, we'll just fix the GPDB-added lines manually. There are no
GPDB-added columns in pg_type.h, so we don't need the GPDB_COLUMN_DEFAULT
tricks we used in pg_proc.h.

Catullus.pl still needs type information to map the type names in
pg_proc_gp.sql to type OIDs. It used to get that information from the
pg_type.sql file, but I changed it to read the DATA lines in pg_type.h
directly.

Also revert unnecessary whitespace and comment changes against upstream, to
make merge easier.

remove_unused_initplans() didn't notice if there were references to a
subplan in the PRECEDING or FOLLOWING clause of a window function, because
plan/expression walkers didn't visit the PRECEDING/FOLLOWING clauses. Fix
the walkers.

We already had a test case for this in the regression suite, in
gp_optimizer, but instead of fixing the bug we had memorized the internal
ERROR message in the expected output. Oopsie.

The code to remove unnecessary InitPlans assumed that an InitPlan can only
return a single Param, which is wrong. Fix the code to handle multiple
Params.

Bug report and test case by CK Tan.

ExecCopySlotHeapTupleTo() is not used anywhere (anymore) so remove.

GPMapreduce isn't threaded so -pthreads is a noop causing annoying
warning messages in clang.

Moves the Python and Perl version tests in under ignore blocks to
make the test less fragile. Testing for specific versions is not
ideal, if this is to be kept as a real test it should be rewritten
to look for interpreters satisfying minimum version requirements
rather then exact (although the value of that can be argued as
well). Keep the version output in the test though since it can be
good information when debugging where access is limited to the
regression.diffs file. Also fix a trivial typo in a comment while
in there.

This adds a new switch to autoconf, --enable-mapreduce, to connect
the GPMapreduce extension to the build. If set to on the mapreduce
extension in gpAux/extensions will be built and installed.  As a
side effect this will properly clean and distclean for gpfdist in
gpAux/extensions as well. Autoconf checks for libyaml since it's
a depedency for GPMapreduce.

This is the first stab at gpmapreduce, once in we can start moving
gpmapreduce back into src/bin and move the mapreduce test suite in
under gpAux/extensions.

Commits backported from upstream are listed below.

	commit 87f2ad13
	Author: Tom Lane <tgl@sss.pgh.pa.us>
	Date:   Sun Mar 27 12:51:04 2011 -0400

	    Fix plpgsql to release SPI plans when a function or DO block is freed.

	    This fixes the gripe I made a few months ago about DO blocks getting
	    slower with repeated use.  At least, it fixes it for the case where
	    the DO block isn't aborted by an error.  We could try running
	    plpgsql_free_function_memory() even during error exit, but that seems
	    a bit scary since it makes a lot of presumptions about the data
	    structures being in good shape.  It's probably reasonable to assume
	    that repeated failures of DO blocks isn't a performance-critical case.

	commit 42b2907d
	Author: Tom Lane <tgl@sss.pgh.pa.us>
	Date:   Sun Nov 29 03:02:27 2009 +0000

	    Add support for anonymous code blocks (DO blocks) to PL/Perl.

	    Joshua Tolley, reviewed by Brendan Jurd and Tim Bunce

	commit 9048b731
	Author: Tom Lane <tgl@sss.pgh.pa.us>
	Date:   Tue Sep 22 23:43:43 2009 +0000

	    Implement the DO statement to support execution of PL code without having
	    to create a function for it.

	    Procedural languages now have an additional entry point, namely a function
	    to execute an inline code block.  This seemed a better design than trying
	    to hide the transient-ness of the code from the PL.  As of this patch, only
	    plpgsql has an inline handler, but probably people will soon write handlers
	    for the other standard PLs.

	    In passing, remove the long-dead LANCOMPILER option of CREATE LANGUAGE.

	    Petr Jelinek

Update to avoid misunderstanding. `--with-perl` means building Perl modules (PL/Perl) instead of perl runtime, `--with-python` means building Python modules (PL/Python) instead of python runtime.

The trgm2int() function was backported from PostgreSQL 8.4, but never used.
And the TRGMINT() macro was removed as part of that backport, without
changing the callers.

This makes the code to compile, and the regression tests to pass, except
for the fact that GPDB doesn't support GIN, which produces an error.

Fixes issue #491, reported by @digoal.

It doesn't seem to be needed anymore. Window and Motion nodes have a
suitable targetlist nowadays, so no need to reach out to their child node
for one.