compatible with s3.amazonaws.com and s3-us-east-1.amazonaws.com regions
catch any exceptions thrown by C++ functions
set min and max values of threadnum and chunksize

and will move development to GPDB source tree after this.

Commit 03201855 failed to update
the unit tests related to the validate_tablenames function, so this
commit alters the function slightly to make it more unit-testable
and makes the necessary unit test changes.

Previously, gpdbrestore had no check to determine whether a table to be
restored with the -T option existed in the dump, and the restore would
either fail during the ANALYZE step with an unclear message or would
succeed if --noanalyze was used. This commit adds a check to fail early
if a nonexistent table is passed to -T.

column compression tests perform bunch of alter type commands,
affecting other tests. Better to perform such opertions on dedicated database.

new conflixt appear:
/opt/gcc_infrastructure/bin/ld: warning: libxbsa64.so, needed by ../../../../src/bin/pg_dump/cdb/libgpbsa.so, not found (try using -rpath or -rpath-link)

This reverts commit b9d8b615.

gptransfer starts multiple threads in workerpool to transfer each source and destination
pair, to ensure the order specified by user, batch-size should be set to 1.

Partition table transfer uses dictionary as a map between source and destination pair,
which will not follow the order of insertion. Since gpdb uses python 2.6 and the
OrderedDict is only available from 2.7, so updated with a workaround using list.

Authors: Pengcheng Tang, Lawrence Hamel

When applying motion, a merge other than normal gather motion should
be added on the top node if it has sort list, this can make sure that
tuples are still in order after gathered to QD. Only checking if top
level parsetree has sort clauses may miss the implicit order constraint
in a view

The old value of rpath is an absolute path on build host, which is
not exist on target machine. Change it to relative path.
After this commit, LD_LIBRARY_PATH could be removed. Problem is that
we still have a lot of extra dependency for enterprise building, so
LD_LIBRARY_PATH is still kept.

This is first release and doesn't support write operation.
S3 has its own unittest with gtest

These tests have not been updated recently to accomadate the new
changes in gpdb, needed to fix it up and make sure gpfdist still works.

We need to make sure that fstream struct gets initialized or we may
pay the price of accidentally freeing a random address that's not
malloced.

This will make it so that we can build gpfdist with or without
OpenSSL when doing the configure.
* Print a message that SSL is not supported if built without OpenSSL

As gpfdist will now be part of the gpdb build by default, we may just
treat it as part of gpdb core.
* Integrate gpfdist configure file to the top level configure and
  changed necessary files to accomadate the changes.
* Modify travis.yml for gpfdist structural changes
* Update gpfdist.c to use GP_VERSION from pg_config.h

Fix warning issues from transform.c
* There were many prototype warnings that was being mentioned.
  Added a headerfile for missing prototype warnings, also included the
  structs into the headerfile.

Replace src/port/glob.c
* src/port/glob.c is dead code at the moment and can be safely replaced by
  the windows changes from gpfdist/port

Check for libyaml:
* disable transformations for gpfdist if not found.

schema name, and table name.

This commit fixes the quotation and escaption for the
whole backup and restore, and resolves related issues which
block us from supporting special chars.

List of remarkable changes:

Removed gpcrondump shell script.
Renamed gpcrondump.py to gpcrondump

Added -S option in gpdbrestore for schema level restore instead of using wildcard '*'.
Added few methods to parse and extract databases, schemas, tables, and others ddl from plain dump file
      to support special chars
Added some behave tests under gppylib/test/behave/mgmt_utils/

Fixed gpcrondump deleting user specified schema file
Fixed gpdbrestore not validating non existed tables correctly.
Fixed the pg_get_partition_def() and pg_get_partition_template_def() function to
      dump DDL statement properly with quotation and escaping.
Fixed ddboost restore and Post data restore not forming filter command correctly

Updated -x option in gpcrondump to not use comma separated specification.
Updated -T option in gpdbrestore to not use comma separated specification.
Updated the pre and post restore filter to parse special chars within schema and table names
Updated some unit tests, behave tests and the help docs

Authors: Pengcheng Tang, Chumki Roy

The jump table was missing some entries, and hence the code wouldn't have
worked correctly. Fortunately it was all dead code, as sendInitGpmonPkts()
only called itself, there were no other callers.

This fixes issue #256. Thanks to Craig Harris for the report!

The 1.622 updates in b3668c40 adds
a depedency on CHint without adding the required header file to
build without warnings/errors.

- This is a placeholder Concourse CI pipeline and task. See concourse.ci
  for details on the Concourse system.

1, In extreme cases where we are acquiring resource queue lock and
cannot find it in share memory, GPDB would raise PANIC, which is not
desirable;

2, Change one log level from LOG to DEBUG1, because that code path would
hit much frequently after previous resource queue commit;

Add support to the vagrantfiles for having additional synced_folders
locally without having to patch the Vagrantfile. Local folders are
configured in a YAML file in the same directory as the Vagrantfile
on the following format:

    synced_folder:
       - local: ../../../orca/gporca
         shared: /gporca
       - local: ../../../orca/gpos
         shared: /gpos

This also add comments to clarify the RTT computational logic

Need to check UDPIC_FLAGS_STOP flag when the ACK sequence equals to receivedAckSeq. Otherwise,
The receiver will keep polling ACK and the sender will keep sending ACK.

This reverts commit f3134604.

This reverts commit e929d4d0.

master before dispatching the command to segments.

gp_truncate_error_log() function first dispatches the permissions check to the
segments and prints out the results before performing the check on the master.
This makes it hard to test this function if the cluster is configured with a
different number of segments than the test since different number of segments
produce different number of lines in the output. Performing the check first on
the master will cause the transaction to error out immediately and will be
easier to test.

segments configured in the cluster.