and use the same error message for CREATE and open table time

That is, when the precision of DEFAULT NOW() is less than
the precision of the column, do not convert it to unireg_check,
use the new approach where DEFAULT is tryly an expression.

* revert part of the db7edfed that moved calculations from
  fix_fields to val_str for Item_func_sysconst and descendants
* mark session state dependent functions in check_vcol_func_processor()
* re-run fix_fields for all such functions for every statement
* fix CURRENT_USER/CURRENT_ROLE not to use Name_resolution_context
  (that is allocated on the stack in unpack_vcol_info_from_frm())

Note that NOW(), CURDATE(), etc use lazy initialization and do *not*
force fix_fields to be re-run. The rule is:
* lazy initialization is *not* allowed, if it changes metadata (so,
   e.g. DAYNAME() cannot use it)
* lazy initialization is *preferrable* if it has side effects (e.g.
  NOW() sets thd->time_zone_used=1, so it's better to do it when
  the value of NOW is actually needed, not when NOW is simply prepared)

into "fix" and "check" parts

in a case of a specially crafted invalid frm

as in

create table t1 (a int);
create view v1 as select default(a) as NOT_UPDATABLE from t1;

Call profiling.restart() and profiling.reset() only if profiling is e…

in other case we will get:

../alex/dev/server/sql/sql_class.cc: In member function ‘void THD::free_connection()’:
../server/sql/sql_class.cc:1664:3: error: ‘profiling’ was not declared in this scope
   profiling.restart();                          // Reset profiling
      ^~~~~~~~~

../server/sql/sql_class.cc: In member function ‘void THD::reset_for_reuse()’:
../server/sql/sql_class.cc:1689:3: error: ‘profiling’ was not declared in this scope
  profiling.reset();
       ^~~~~~~~~

errors.

This reverts commit 7b89b9f5.

Based on contributions by Daniel Black and Otto Kekäläinen.

.. logged under row binlog format

In the early stages of ALTER TABLE execution, the implementation
checks whether its a NOOP (alter_info->flags == 0), and if so,
it returns after logging the command to binary log. The logging,
however, was done unconditionally.

Fixed by skipping the logging for temporary tables when under row
based replication.

.. share->last_version' failed in myisam/mi_open.c:67: test_if_reopen

During the RENAME operation since the renamed temporary table is also
opened and added to myisam_open_list/maria_open_list, resetting the
last_version at the end of operation (HA_EXTRA_PREPARE_FOR_RENAME)
will cause an assertion failure when a subsequent query tries to open
an additional temporary table instance and thus attempts to reuse it
from the open table list.

This commit fixes the issue by skipping flush/close operations executed
toward the end of ALTER for temporary tables. It also enables a shortcut
for simple ALTERs (like rename, disable/enable keys) on temporary
tables.

As safety checks, added some assertions at code points that should not
be hit for temporary tables.

Signed-off-by: Daniel Black <daniel.black@au.ibm.com>

This is never hit in practice, because the wrapper code catches it,
but fix the zero length exit to restore all the non volatile registers.

From: https://github.com/antonblanchard/crc32-vpmsum/commit/aaf0ac48a401abfaa0ad2d520be755e66c1e2e95Signed-off-by: Anton Blanchard <anton@samba.org>

These are different from the existing crc32 functions which
where really crc32c.

Move crc32-vpmsum to extra.
Compile static crc32-vpmsum instead of adding sources directly.
Make use of crc32-vpmsum via my_checksum().

Based on contribution by Daniel Black.

Table before collecting engine independent statistics now is reopened in read mode,
InnoDB allow write operations in this case.

The memory alloc-ed initially for wsrep_sst_auth to store the
value specified in config was lost as the global variable was
reset while in process of masking it and thus, could never be
reclaimed on shutdown.

Make the slave SQL thread always output to the error log the message "Slave
SQL thread exiting, replication stopped in ..." whenever it previously
outputted "Slave SQL thread initialized, starting replication ...".

Before this patch, it was somewhat inconsistent in which cases the message
would be output and in which not, depending on the exact time and cause of
the condition that caused the SQL thread to stop.

             buildbot)

Updated test results according to MDEV-6720.

Reduce number of network send() calls for batch update requests

Adding Converter_double_to_longlong and reusing it in:
1. Field_longlong::store(double nr)
2. Field_double::val_int()
3. Item::val_int_from_real()
4. Item_dyncol_get::val_int()
As a good side efferct, now overflow in conversion in the mentioned
val_xxx() methods return exactly the same warning.

include the dropped object type

- 12d75e61 - new thread stack size
- 99e48cb1 - warning text changed
- 6c173324 - default values for BLOBs
- c87e002b - don't return a negative zero

this is useless now, flags are recalculated on load anyway.
But storing flags on disk means we cannot easily change (add,
remove, or renumber) them in the new MariaDB version.

because table->map is set much later.
Use check_vcol_func_processor() to detect fields too.