Signed-off-by: Christian Pointner <equinox@spreadspace.org>

…ydumper#35)

…a cluster. (mydumper#33)

This change leverage the MarkDown format in the file

Following the improvement in PR mydumper#59

MariaDB 10.2 splitted server/client variables and now
MYSQL_SERVER_VERSION is not available. We now take the
version string from the client header.

Patch provided by by Brian Evans (grknight@gentoo.org):
https://bugs.gentoo.org/635176

Upstream issue:
https://jira.mariadb.org/browse/MDEV-13773

… line (mydumper#15)

* adding the ability to use --ask-password or -a in order to be prompted to use a password rather than passing it through the command line

* Adjusting getPassword to use getpass() so the password isn't displayed to the user

* forgot to include the option for password prompting in myloader.c as well

* Forgot to include getPassword.c in the cmakelist in order to get it to work in myloader

* removing new lines

Removing new lines as requested

The patch tries to not use the server version string for the version
output.
MySQL has added LIBMYSQL_VERSION to mysql_version.h
MariaDB has added MARIADB_VERSION_NUMBER_STR via an include and no
longer declares MYSQL_SERVER_VERSION for the client.

…mydumper#75)

just by filename which mean the **CURRENT WORK DIR**, but it should
be the file in data directory passed by -d

Commit 37aa066 broke compilation under
at least Debian 9 and CentOS 6, as MYSQL_TYPE_JSON is not defined in the
MySQL versions they ship.

* New option to specify tables to skip via a file.

The --regex option works well but cannot scale to hundreds of thousands
of distinct table names when a large number of tables must not be backed
up.

This new --omit-from-file option allows specifying a file containing a
list of table names (in database.table format, one per line) to skip
during backup.

* Document the new --omit-from-file option.

…er#99)

* add support for mysql 5.7 generated fields

* minor memory leak fixes

* Fix memleak in dump_table_data

We've noticed a high memory usage when there were alot of tables/databases.
The memory usage increased while dumping the tables.
Valgrind gave us the answer:
==24877== 24,192,604 (1,757,400 direct, 22,435,204 indirect) bytes in 73,225 blocks are definitely lost in loss record 175 of 185
==24877==    at 0x4C28C20: malloc (vg_replace_malloc.c:296)
==24877==    by 0x60267C9: g_malloc (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4200.1)
==24877==    by 0x603D62F: g_slice_alloc (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4200.1)
==24877==    by 0x60416C2: g_string_sized_new (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4200.1)
==24877==    by 0x41FCA0: dump_table_data (mydumper.c:2723)
==24877==    by 0x4207C7: dump_table_data_file (mydumper.c:2521)
==24877==    by 0x420D13: process_queue (mydumper.c:522)
==24877==    by 0x6047844: ??? (in /lib/x86_64-linux-gnu/libglib-2.0.so.0.4200.1)
==24877==    by 0x4E3D063: start_thread (pthread_create.c:309)

* Fix more memleaks

* TLS support

* DRY the connection code

* Add and use SSL verification options

* added option to disable ssl build

* remove ssl arguments when -DWITH_SSL=OFF

* vars from connection.c

* support for old MYSQL_SERVER_VERSION

Makes more sense

Removed session options from myloader that do not exist.