Was in close (the virtual machine could have been detached.
  modified:   storage/connect/jdbconn.cpp

- in DOMNODELIST::DropItem
  if (Listp == NULL || Listp->length <= n)
    return true;
is wrong, should be:
  if (Listp == NULL || Listp->length < n)
    return true;
- Crash in discovery with libxml2 in XMLColumns because:
            if (!tdp->Usedom)    // nl was destroyed
              vp->nl = vp->pn->GetChildElements(g);
is executed with vp->pn uninitialized. Fixed by adding:
          vp->pn = node;
line 264.
-In discovery with libxml2 some columns are not found.
Because list was not recovered properly, nodes being modified and not reallocated.
Fixed lines 214 and 277.
  modified:   storage/connect/domdoc.cpp
  modified:   storage/connect/tabxml.cpp

Add support for zipped table files
  modified:   storage/connect/domdoc.cpp
  modified:   storage/connect/domdoc.h
  modified:   storage/connect/filamap.cpp
  modified:   storage/connect/filamap.h
  modified:   storage/connect/filamzip.cpp
  modified:   storage/connect/filamzip.h
  modified:   storage/connect/ha_connect.cc
  modified:   storage/connect/libdoc.cpp
  modified:   storage/connect/plgdbutl.cpp
  modified:   storage/connect/plgxml.cpp
  modified:   storage/connect/plgxml.h
  modified:   storage/connect/tabdos.cpp
  modified:   storage/connect/tabdos.h
  modified:   storage/connect/tabfmt.cpp
  modified:   storage/connect/tabjson.cpp
  modified:   storage/connect/tabxml.cpp

Essentially revert MDEV-6759, which addressed a double free of memory
by removing the freeing altogether, introducing the memory leaks.
No double free was observed when running the test suite -DWITH_ASAN.

Replace some mem_heap_free(foreign->heap) with dict_foreign_free(foreign)
so that the calls can be located and instrumented more easily when needed.

  Enable using multiple zip files
  modified:   storage/connect/filamzip.cpp
  modified:   storage/connect/ha_connect.cc
  modified:   storage/connect/ha_connect.h
  modified:   storage/connect/mycat.h
  modified:   storage/connect/tabdos.cpp
  modified:   storage/connect/tabdos.h
  modified:   storage/connect/tabfmt.cpp
  modified:   storage/connect/tabjson.cpp

  A first experimental and limited implementation.
  modified:   storage/connect/CMakeLists.txt
  modified:   storage/connect/filamap.cpp
  new file:   storage/connect/filamzip.cpp
  new file:   storage/connect/filamzip.h
  modified:   storage/connect/ha_connect.cc
  new file:   storage/connect/ioapi.c
  new file:   storage/connect/ioapi.h
  modified:   storage/connect/mycat.cc
  modified:   storage/connect/plgdbsem.h
  modified:   storage/connect/plgdbutl.cpp
  modified:   storage/connect/tabdos.cpp
  modified:   storage/connect/tabdos.h
  modified:   storage/connect/tabfmt.cpp
  modified:   storage/connect/tabfmt.h
  modified:   storage/connect/tabjson.cpp
  modified:   storage/connect/tabjson.h
  new file:   storage/connect/tabzip.cpp
  new file:   storage/connect/tabzip.h
  new file:   storage/connect/unzip.c
  new file:   storage/connect/unzip.h
  new file:   storage/connect/zip.c

Role names with trailing whitespaces are truncated in length as of
956e92d9 to fix MDEV-8609. The problem
is that the code that creates role mappings expects the string to be null
terminated.

Add the null terminator to account for that as well. In the future
the rest of the code can be cleaned up to never assume c style strings
but only LEX_STRINGS.

MDEV-10713: signal 11 error on multi-table update - crash in handler::increment_statistics or in make_select or assertion failure pfs_thread == ((PFS_thread*) pthread_getspecific((THR_PFS)))

Different fix. Don't allow Item_func_sp to be evaluated unless
all tables are prelocked.

Extend the test case to make sure Item_func_sp::val_str is called
(the table must have at least one row for that).

Revert "MDEV-10713: signal 11 error on multi-table update - crash in handler::increment_statistics or in make_select or assertion failure pfs_thread == ((PFS_thread*) pthread_getspecific((THR_PFS)))"

This reverts commit 035a5ac6.

Two minor problems and one regression:
1. caching the value in str_result. Other Item methods may use it,
   destroying the cache. See, for example, Item::save_in_field, where
   str_result is moved to use a local buffer (this failed main.grant)
2. Item_func_conv_charset::safe is now set too late, it's initialized
   only in val_str() but checked before that, this failed many tests
   in optimized builds.

to fix 1 - use tmp_result instead of str_result, to fix 2, use
the else branch in the Item_func_conv_charset constructor to set
safe purely from charset properties.

But this introduces a regression, constant strings can no longer be
converted, say, from utf8 to latin1 (because 'safe' will be false).
This fails few tests too. There is no way to fix it without reverting
the commit and converting constants, as before, in the constructor.

MDEV-11162 Assertion `num_records == m_idx_array.size()' failed in Filesort_buffer::alloc_sort_buffer(uint, uint)

Updating result for the group_by_innodb.test

MDEV-11162 Assertion `num_records == m_idx_array.size()' failed in Filesort_buffer::alloc_sort_buffer(uint, uint)

When JOIN::destroy() is called for a JOIN object that has
   - join->tmp_join != NULL
   - also has join->table[0]->sort

then the latter was not cleaned up.
This could cause a memory leak and/or asserts in the subsequent queries.

Fixed by adding a cleanup call.

truncate unused IO_CACHE backing store files in binlog_cache_data
to release the disk space they were occupying

        Rpl_filter::parse_filter_rule() made NULL-safe.

Update mysqldump man page

check for VIEW/DERIVED fields

MDEV-10713: signal 11 error on multi-table update - crash in handler::increment_statistics or in make_select or assertion failure pfs_thread == ((PFS_thread*) pthread_getspecific((THR_PFS)))

Move expression execution out of Item constructor.

Exclude untouched in prepare phese subqueries from the select/unit tree
because they became unreachable by execution.

The problem was that null_value was not set to "false" on a well-formed row.
If an ill-formed row was followed by a well-forned row, null_value remained
"true" in the call of Item::send() for the well-formed row.

Check for readline before checking for curses headers, because
MYSQL_CHECK_READLINE fails when curses is not found, but
CHECK_INCLUDE_FILES simply remembers the fact and continues.  So if
there's no curses, MYSQL_CHECK_READLINE will abort, the user will then
installs curses and continue the build. Thus, CHECK_INCLUDE_HEADERS
will remember that there is no curses, but other checks from
MYSQL_CHECK_READLINE will remember that curses are there. It will
result in inconsistent HAVE_xxx defines.

be consistent and don't include the table name into the error message,
no other CREATE TABLE error does it.

(the crash happened, because thd->lex->query_tables was NULL)

Don't assume that a word of n bytes can match a word of
at most n * charset->mbmaxlen bytes, always go for the worst.

  Fix includes launchpad fix plus more to cover writing BIN tables.
  modified:   storage/connect/tabfix.cpp
  modified:   storage/connect/value.cpp
  modified:   storage/connect/value.h

- Typo: Change the name of filamzip to filamgz to prepare future ZIP tables.
  modified:   storage/connect/CMakeLists.txt
  added:      storage/connect/filamgz.cpp
  added:      storage/connect/filamgz.h
  deleted:    storage/connect/filamzip.cpp
  deleted:    storage/connect/filamzip.h
  modified:   storage/connect/plgdbsem.h
  modified:   storage/connect/reldef.cpp
  modified:   storage/connect/tabdos.cpp
  modified:   storage/connect/tabdos.h
  modified:   storage/connect/tabfix.cpp
  modified:   storage/connect/tabfmt.cpp
  modified:   storage/connect/tabjson.cpp

Due to the collation used on the roles_mapping_hash, key comparison
would work in a case-insensitive manner. This is incorrect from the
roles mapping perspective. Make use of a case-sensitive collation for that hash,
the same one used for the acl_roles hash.

The function Item_func_isnull::update_used_tables() must
handle the case when the predicate is over not nullable
column in a special way.
This is actually a bug of MariaDB 5.3/5.5, but it's probably
hard to demonstrate that it can cause problems there.

Test intentionally crashes the server, thus corrupted pages possible.