This removes all the code for inserting, updating, or deleting through
CaQL context. All the code surrounding the updates are also converted to
use regular SysCache lookups or systable scans.

Note that this leaves the support for deleting rows with
caql_getcount("DELETE ...") alone. For now. This commit is all about
removing the functions that modify the current row of a caql scan.

There was only one caller of that facility. Replace it with direct
index_beginscan/getnext calls.

Since there is a function specifically for this lookup, might as well use
it.

We surely want to have a syscache for any lookups done at execution, for
speed. This makes it more clear that there indeed is a syscache for these.
(And I'm trying to remove CaQL altogether, anyway)

The CaQL lookup in assign_func_result_transient_type(Oid) was in fact
wrong before. It called caql_endscan(), and continued to access the tuple
after that. Fix that while we're at it.

The ExecuteDoStmt() function was backported from PostgreSQL 9.0, but it
was put before the AlterFunctionNameSpace() function, even though it's
after it in the upstream. Move it to where it is in the upstream, to
reduce merge conflicts in the future.

Not sure why the handling of language attribute was changed in GPDB. Let's
revert it back to the way it's done in the upstream. Also change the
handling of "prodataaccess" to follow the same pattern.

I tried to change all of these in commit 6d7f09f3, but missed a few.

This removes all utilities which are known to be deprecated from
gpMgmt and it also clears out any command line switches marked as
deprecated along with code marked deprecated. It also removes the
leftovers from previously removed utilities such as gpsuspend and
gpcheck_hostdump etc and fixes mentions of renamed utilities.

There should be no functional change from this commit as the only
removed utils have been marked deprecated for some time.

Not sure why we had these twice in the tinc suite. In any case, these
are straightforward DDL tests, so they can well live in the main test
suite.

We have the same tests in src/test/regress/bugbuster/sql/spi.sql already.

I don't see any references to "phone_book", by grepping the tincrepo
directory. So seems to be unused.

Moreover, we have this same data in src/test/regress/input/qp_regexp.source.

This is mostly the same as the README on previous directory level.

These are straightforward "run SQL, check expected output" style tests,
and only take a few seconds to run. Let's get them into the normal
regression suite.

We have all of these same tests in
src/test/regress/bugbuster/sql/schema_topology.sql.

We already have the exact same tests in the normal regression suite, as
src/test/regress/sql/DML_over_joins.sql.

We have new, improved versions of these in src/test/regress. It seems
that TINC will happily use those versions, if we just remove these
old copies from ext.

These are all direct ReleaseSysCache() calls in the upstream.

Move GPDB-specific keywords to a separate section.

The return value was unused, and the 'spin' and 'putInFreeList' argument
were alwasy passed as 'true', which gave the upstream behaviour.

I think these were left out by accident when we cherry-picked the upstream
patches to add location information to various structs.

This reverts commit ce3b3675.

I found these with "callcatcher", written by Caolán McNamara. Many thanks
for the tool! See https://www.skynet.ie/~caolan/Packages/callcatcher.html

Add -H option to disable update gpcrondump history table.
Option '-h' will not be deleted in this release, but it will be deprecated in a future release.

Update behave tests and unit tests.
Modify gpcrondump help file.

Authors: Chumki Roy, Marbin Tan, Nikhil Kak

After commit c63f1b5d, RowExclusiveLock lock on pg_class
won't be held till the end of transaction. Update the expected output file accordingly.

SnapshotNow MUST be used during insert flow to fetch the latest EOF value from
aoseg table. Earlier usage of ActiveSnapshot in AO insert flow caused data
corruption as potentially could read incorrect or stale EOF value from
aoseg. The specific scenario this happens when entry from AppendOnlyHash table
gets evicted out. It invalidates (zeros out) the latestWriteXid and hence
check for usedByConcurrentTransaction cannot be performed. So, any transaction
later inserting data to same AO table, if it has aquired the snapshot and has
latestWriteXid listed in its in-progress distributed transaction list, gets
the same segfile to write. Based on DTM visibility rules using ActiveSnapshot
current transaction will not read the EOF value from previous inserted
transaction as its listed in its in-progress list and hence will overwrite the
data but SnapshotNow will see the latest and make sure to append.