"Move innobase_release_stat_resources(trx) outside the 'if' in
ha_innobase::external_lock(). That would add more safety that whatever
MySQL does at a query end, there would be no risk of a hang on the btr
search latch."

Also call innobase_release_temporary_latches() in the beginning of
ha_innobase::close().

Approved by:	Heikki

ERROR_LOCK_VIOLATION is encountered during file operation.
This is caused by backup software, so InnoDB should retry while the backup
software is done with the file.

Approved by:	Heikki

this variable settable. Since the pluggable engine interface currently
doesn't provide a usable mechanism, we will add the latter functionality
once it's available.

of UNDO slots in the rollback segment. This is a partial fix since the
MySQL error code requested to properly report the error condition back
to the client has not yet materialized. Currently we have #ifdef'd the
error code translation in ha_innodb.cc. This will have to be changed
as and when MySQl add the new requested code or an equivalent code
that we can then use.

Given the above, currently we will get the old behaviour, not the "fixed"
and intended behaviour.

defs exceed the max row size.

The fix returns a more appropriate error message. Add a test case to
innodb.test and expected output to innodb.result.

Approved by:	Marko

UNIV_IBUF_DEBUG.

innobase_start_or_create_for_mysql(): Note that crash recovery is broken
when UNIV_IBUF_DEBUG is defined.

ibuf_counts[]: Make this a two-dimensional array.  No need to allocate
anything from the heap.  Eliminate ibuf_counts_inited, as the array
will be zero-filled by the runtime environment.

ibuf_count_check(): New function, to print out an explanation before
assertion failure.

(Bug #23313, Bug #21404)

ha_innobase::update_create_info(): New function, to report
the auto_increment_value.

to set this flag immediately after open(2)ing. This way an error caused by
O_DIRECT not being supported can easily be ignored.

* Add support for skipping the OS caching on Solaris by calling directio()
instead of fcntl().

Approved by:	Heikki

into a generic function which is called from both os_file_handle_error() and
os_file_handle_error_no_exit()

Approved by:	Marko

innodb_commit_concurrency was implemented.

MySQL calls external_lock at the beginning and end of a statement
when it is not calling start_stmt or commit or rollback.  Thus,
statement boundaries can be (and are already) detected without
monitoring thd->query_id.

The function innobase_commit() seemingly lacks the call to
innobase_release_stat_resources(), which should be called at
the end of every SQL statement.  The call was replaced by
equivalent statements by Vadim Tkachenko when he implemented
innodb_commit_concurrency in MySQL 5.0:

http://mysql.bkbits.net:8080/mysql-5.0/?PAGE=patch&REV=1.1886.70.1

referenced from other modules.

ChangeSet
  2007/03/20 10:22:15-04:00 iggy@recycle.(none) 
  Post Merge Fix.

storage/innobase/CMakeLists.txt
  2007/03/20 10:22:13-04:00 iggy@recycle.(none) +2 -2
  Post Merge Fix.

ChangeSet
  2007/02/14 22:06:41-08:00 igor@olga.mysql.com 
  Fixed bug #25971: indexes on text columns were ignored when ref accesses
  were evaluated.
  According to the new rules for string comparison partial indexes on text
  columns can be used in the same cases when partial indexes on varchar
  columns can be used.

mysql-test/r/innodb.result
  2007/02/14 22:06:39-08:00 igor@olga.mysql.com +1 -1
  Adjusted results after the fix for bug #25971.

ChangeSet
  2007/02/15 15:39:03+01:00 guilhem@gbichot3.local 
  Fix for BUG#25507 "multi-row insert delayed + auto increment causes
  duplicate key entries on slave" (two concurrrent connections doing
  multi-row INSERT DELAYED to insert into an auto_increment column,
  caused replication slave to stop with "duplicate key error" (and
  binlog was wrong)), and BUG#26116 "If multi-row INSERT
  DELAYED has errors, statement-based binlogging breaks" (the binlog
  was not accounting for all rows inserted, or slave could stop).
  The fix is that: if (statement-based) binlogging is on, a multi-row
  INSERT DELAYED is silently converted to a non-delayed INSERT.
  Note: it is not possible to test BUG#25507 in 5.0 (requires mysqlslap),
  so it is tested only in the changeset for 5.1. However, BUG#26116
  is tested here, and the fix for BUG#25507 is the same code change.

mysql-test/r/innodb-replace.result
  2007/02/15 15:39:01+01:00 guilhem@gbichot3.local +2 -2
  result update

mysql-test/t/innodb-replace.test
  2007/02/15 15:39:01+01:00 guilhem@gbichot3.local +2 -2
  now that multi-row delayed inserts are converted to normal inserts
  if the statement-based binlog is enabled,
  no error is issued even if this engine does not support INSERT DELAYED,
  as the insert does not go through the INSERT DELAYED code.
  To preserve the goal of this test, we change the statements to single-
  row inserts.

the innodb_gis test case.

ChangeSet
  2007/02/19 13:57:06+03:00 kaa@polly.local
  Bug#18743: Several test cases fails if "classic" configuration in 5.0
  The problem happened because those tests were using "cp932" and "ucs2"
  without checking whether these character sets are available.
  This fix moves test parts to make character set specific parts be
  tested only if they are:
  - some parts were moved to "ctype_ucs.test" and "ctype_cp932.test"
  - some parts were moved to the newly added tests "innodb-ucs2.test",
  "mysqlbinglog-cp932.test" and "sp-ucs2.test"

mysql-test/t/innodb.test
  2007/02/19 13:57:02+03:00 kaa@polly.local +0 -222
  Moved ucs2-specific test cases to innodb-ucs2.test

ChangeSet@1.2409.1.83  2007-03-06 10:36:15-07:00  tsmith@hindu.god
Bug #26598: Create variable to allow turning off of statistic gathering
on metadata commands

Add innodb_stats_on_metadata option, which enables gathering
index statistics when processing metadata commands such as
SHOW TABLE STATUS.  Default behavior of the server does not
change (this option is enabled by default).

os_file_handle_error().

Approved by:	Heikki

innobase_query_caching_of_table_permitted(): Make static.

ha_innobase::register_query_cache_table(): Move the function
definition from ha_innodb.h to ha_innodb.cc.  Add comments.

ha_innobase::change_active_index(): Do not call current_thd unless
UNIV_DEBUG is defined.

source distribution of MySQL.

Approved by:	heikki

debug assertion failure ut_ad(mutex_own(&(dict_sys->mutex))) in
dict_table_get_on_id() after the rollback following crash recovery.

and remove the constants TRX_USER and TRX_PURGE.

and never tested.

Remove the unused constants HA_INNOBASE_ROWS_IN_TABLE and
HA_INNOBASE_RANGE_COUNT.  Declare innobase_active_counter static.

The bug could be reproduced as follows:

Define a table so that the first column of the clustered index is
a VARCHAR or a UTF-8 CHAR in a collation where sequences of bytes
of differing length are considered equivalent.

Insert and delete a record.  Before the delete-marked record is
purged, insert another record whose first column is of different
length but equivalent to the first record.  Under certain conditions,
the insertion can be incorrectly performed as update-in-place.

Likewise, an operation that could be done as update-in-place can
unnecessarily be performed as delete and insert, but that would not
cause corruption but merely degraded performance.

consistent read set its own snapshot

regardless of start->undo_no.  Otherwise, aborted searches may show
up as genuine deadlocks.  This mistake was made in r1330.

rewind lock_latest_err_file and display the two transactions at the
point of aborting the search.  (Bug #25494)

The results are not affected.

ChangeSet@1.2456.1.3, 2007-02-27 20:06:37+02:00, monty@mysql.com +2 -0
  Fix (last) compiler warnings

  storage/innobase/pars/lexyy.c@1.23, 2007-02-27 20:06:36+02:00, monty@mysql.com +1 -1
    Fix compiler warnings (fix is also in pars0lex.l)

ChangeSet@1.2452, 2007-02-23 13:13:55+02:00, monty@mysql.com +177 -0
  Fixed compiler warnings
  ...
  Fixed compiler warnings detected on windows64