The MYSQL_AUDIT_NOTIFY_CONNECTION_CONNECT() call moved to the login_connection()
  function. So that it'll be invoked in any thread handling mode.

        If the SET PASSWORD query doesn't have the password string,
        the parsing of it can fail. It manifested first in MySQL 5.6 as
        it started to hide password lines sent to the plugins.
        Fixed by checking for that case.

   Symbols like TAB or NEWLINE should be escaped, which was
        forgotten in one place.

Check that leaf table list is really built before storing it.

Server may crash if sanity checks of COLUMN_GET() fail.

COLUMN_GET() description generator expects parent CAST item, which may not have
been created due to failure of sanity checks. Then further attempt to report
an error may crash the server.

Fixed COLUMN_GET() description generator to handle such case.

Factory timezone is supposed "For companies who don't want to put time zone
specification in their installation procedures. When users run date, they'll get
the message. Also useful for the "comp.sources" version."

This "message" is exposed as timezone abbreviation, which is supposed to be
short and thus may cause generated INSERT statements to fail.

Do not attempt to load Factory timezone.

Fixed that ALTER VIEW ALGORITHM=UNDEFINED behaved as if algorithm was not
specified.

update tests

When truncating a table's partition, we also need to invalidate the
query cache for it.

treated by git as text files.

Marking all MyISAM files as binary in .gitattributes: (*.frm, *.MYD, *.MYI)

MDEV-8267 Add /*old*/ comment into I_S.COLUMN_TYPE for old DECIMAL

MDEV-7906: InnoDB: Failing assertion: prebuilt->sql_stat_start || trx->state == 1 on concurrent multi-table update

Analysis: Problem is that SQL-layer calls handler API after storage
engine has already returned error state. InnoDB does internal
rollback when it notices transaction error (e.g. lock wait timeout,
deadlock, etc.) and after this transaction is not naturally in
correct state to continue.

Fix: Do not continue fetch operations if transaction is not started.

change the test case to be deterministic
(while still sufficient to test a bug)

jessie has newer automake so build-depends could not be satisfied.

refresh build-depends, remove automake, libtool,
doxygen, texlive-latex-base, ghostscript.

create a separate list of views to repair, and repair them
in a separate loop.

remove the code that checks for correct options for
for CHECK/REPAIR VIEW. Rewrite the grammar for the parser
to check that. This changes error messages as

-ERROR 42000: You have an error ... near '' at line 1
+ERROR 42000: You have an error ... near 'quick' at line 1

don't include #include directives into .pp files

Fixes MDEV-7658: MDEV-7026 fix reintroduces MDEV-6615 on AArch64

This is an addendum to the fix for MDEV-7026. The ARM memory model is
similar to that of PowerPC and thus needs the same semantics with
respect to memory barriers. That is, os_atomic_test_and_set_*_release()
must be a store with a release barrier followed by a full
barrier. Unlike x86 using __sync_lock_test_and_set() which is
implemented as “exclusive load with acquire barriers + exclusive store”
is insufficient in contexts where os_atomic_test_and_set_*_release()
macros are used.

Do not use format function attribute for sql_print_xxx() family of
functions as they use a MariaDB-specific extension of printf instead
of one provided by the system.

On Windows, the files get checked out with CRLF thanks
to .gitattributes.

on REPAIR don't do table-specific stuff for views
(because even if the view has a temp table opened for it,
it's not opened all the way down the engine. In particular,
Aria crashes in maria_status() because MARIA_HA* info - that is
table->table->file->file - is NULL)

initializers.

c99 style for assigning structure members