trx->sess or trx_dummy_sess is non-NULL.

page_get_free_space_of_empty_noninline() by 2.  Until r1571, that function
did not return the same value as page_get_free_space_of_empty().

that was introduced in r1571.  Inlining is disabled in ha_innodb.cc.

When creating an index containing a too long record, InnoDB would
dereference a NULL pointer when trying to determine the maximum row length.

innodb_check_for_record_too_big_error(): Replace the dict_table_t*
parameter with a Boolean flag.  There is not always a dict_table_t object
when this function is called.

page_get_free_space_of_empty_noninline(): Move the definition and
declaration from row0mysql (!) to page0page.  Make the signature
identical with page_get_free_space_of_empty().

create_clustered_index_when_no_primary(): Add the parameter "comp".
Remove unnecessary casts.

Approved by:	Marko

The Unix user running mysql-test-run usually does not have any privileges
on the MySQL test database.

records in ROW_FORMAT=COMPACT is REC_N_NEW_EXTRA_BYTES + 8.  The
REC_N_NEW_EXTRA_BYTES was accidentally omitted in r1546.  This function
should never be called on those records, though.

Approved by:	Heikki

prebuilt->sql_stat_start.  In an ALTER TABLE statement in the innodb_gis
test, an ut_ad() assertion failed, because no IX lock had been acquired
on the table, because prebuilt->sql_stat_start was inadvertently reset
to FALSE, by this function.  This function was called via
ha_innobase::info() and mysql_prepare_alter_table().

ChangeSet@2007-05-10 12:59:39+03:00, monty@mysql.com 
  WL#3817: Simplify string / memory area types and make things more consistent
  (first part)
  
  The following type conversions was done:
  
  - Changed byte to uchar
  - Changed gptr to uchar*
  - Change my_string to char *
  - Change my_size_t to size_t
  - Change size_s to size_t
  
  Removed declaration of byte, gptr, my_string, my_size_t and size_s. 
[...]
ha_innodb.cc:
  Removed some old types
  Updated hash-get-key function arguments
  Added missing casts for alloc() and printf()
  Removed some not needed casts

ha_innodb.h:
  Removed some old types

do not incorrectly mark the column as fixed-length.  (Bug #28138)

ibuf_entry_build(): Instead of prefix_len, pass fixed_len to
dtype_new_store_for_order_and_null_size().  Add debug assertions.

btr_index_rec_validate(): Correct a comment about prefix indexes.

rec_get_converted_size_new(), rec_convert_dtuple_to_rec_new(): Add
debug assertions and comments.

dict_col_type_assert_equal(): New debug function.

normal processing of the query by MySQL instead of generating an error.

Reviewed by: Heikki

Approved by:	Heikki

was added in r1521.

broken in r1521.

In reality, the patch breaks the handling of prefix indexes of
variable-length columns in ROW_FORMAT=COMPACT.  Reverting the patch
is only a partial fix of Bug #28138.

thd_is_replication_slave_thread(), thd_has_edited_nontrans_tables():
Remove blank line between the function comment and the function
definition.  There should be exactly one line between the return
type and the function comment, and this line should be one of
'', 'static', 'UNIV_INLINE', and 'extern "C"'.

spotted by: Marko

Patch to allow monitor threads to stop before proceeding with normal shutdown. 
Also have a separate time counter for tablespace monitor.

reviewed by: Heikki

This makes it possible to see which expression was false by looking at the
error message.

Approved by:	Marko

tables heavier than ones that had not. This helps killing the "right"
transaction in case of a deadlock.

Approved by:	Heikki

choosing the lightest transaction to kill when a deadlock occurs.
This fixes Bug#21293 partially.

Approved by:	Heikki

thd_to_trx(ha_thd()), in order to avoid potential memory corruption.

row_prebuilt_t; currently prebuilt->read_just_key and
prebuilt->keep_other_fields_on_keyread.

whose name starts with srv_.  These variables are declared in the header
files that are covered by #include directives in ha_innodb.cc.

longlong to long long because MYSQL_SYSVAR_LONGLONG marco expects long long
type.

Also change
((ulint)innobase_buffer_pool_size) / 1024
to
(ulint)(innobase_buffer_pool_size / 1024)
and remove comment which is no longer true.

Provided that innobase_buffer_pool_size is always 64bits these statements
are equivalent if ulint is 64 bit (well it will screw up if
innobase_buffer_pool_size is negative). And if ulint is 32 bit the later
variant gives a little more chance that the value will fit.

Approved by:	Heikki

the common headers.  This may be related to WL#2936
(pluggable storage engines).

singleton object whose address is stored into innodb_hton_ptr.

and adapt some things.

Note that ha_innodb.cc depends on mysql_tmpfile() being declared in
<mysql/plugin.h>.  Until the function is declared there, you can
uncomment the buggy definition of mysql_tmpfile in ha_innodb.cc.

Remove storage/innobase/*/Makefile.am.  The whole compilation is driven by
storage/innobase/Makefile.am and storage/innobase/plug.in.

plug.in: Declare InnoDB as a dynamic plugin.

ha_innodb.h: Remove the declarations of many global variables.  The variables
are no longer directly referenced outside of storage/innobase.

trx_t: Add the field trx->duplicates.

trx_create(): Initialize the fields trx->active_trans and trx->duplicates.

innobase_query_is_update(): Remove.  Consult trx->duplicates instead.

Spotted by:	Marko

Suggested by:	Heikki

row_ins_check_foreign_constraint(), row_ins_scan_sec_index_for_duplicate():
Make use of the predicates page_rec_is_infimum() and page_rec_is_supremum().

server to enter InnoDB. This can be made further customizable by the
user if we introduce a new config parameter. This will wait until
config parameters can be easily added.

Approved by:	Marko