Backport from 4.0: Fix a major bug in InnoDB query estimator for queries of type SELECT ... WHERE col < x and SELECT ... WHERE col > x; MySQL could pick a table scan though the result set was only a few rows in a big table


innobase/btr/btr0cur.c:
  Backport from 4.0: Fix a major bug in InnoDB query estimator for queries of type SELECT ... WHERE col < x and SELECT ... WHERE col > x; MySQL could pick a table scan though the result set was only a few rows in a big table

   now maintained in a separate "mysqldoc" BK source tree) - it will be
   replaced with the correct manual.texi when building the official
   source distribution
 - fixed automake error about reservedwords.texi by moving it to EXTRA_DIST
   instead


Docs/Makefile.am:
   - moved reservedwords.texi from info_TEXINFOS to EXTRA_DIST since it is
     not a full-featured texi file and automake would complain about
     "missing @setfilename"
Docs/manual.texi:
   - replaced manual.texi with a dummy placeholder file (manual.texi is
     now maintained in a separate "mysqldoc" BK source tree) - it will be
     replaced with the correct manual.texi when building the official
     source distribution

   so it's part of the distribution (manual.texi includes this file)
   It will be replaced with the correct one from the mysqldoc tree before
   building the distribution
 - removed generation and inclusion of MIRRORS file, since the mirror list
   is no longer part of manual.texi anyway
 - replaced YFLAGS with AM_YFLAGS in sql/Makefile.am to make automake happy
 - Redirect standard error when checking for dpkg-architecture in ltconfig
   to avoid error message showing during configuration


Docs/Makefile.am:
   - added reservedwords.texi (which gets included by manual.texi)
   - removed MIRRORS (mirror list is not part of the manual anymore)
Docs/reservedwords.texi:
   - Updated wording a bit after adding it to the tree
Makefile.am:
   - removed MIRRORS from EXTRA_DIST (it's not part of the manual anymore)
ltconfig:
   - Redirect standard error when checking for dpkg-architecture
     to avoid error message showing during configuration (thanks to
     John Wythe for the patch)
sql/Makefile.am:
   - replaced YFLAGS with AM_YFLAGS to make automake happy

into sinisa.nasamreza.org:/mnt/work/mysql

 - tagged ChangeSet 1.1315 as "mysql-3.23.55"


configure.in:
   - bump up version number 3.23.55 -> 3.23.56

 - use "gnutar" on Mac OS X to build the binary distribution


Build-tools/Do-compile:
   - merged relevevant changes from 4.0 (Options no-mysqltest, no-benchmark,
     log_timestamp function)
scripts/make_binary_distribution.sh:
   - make sure to use "gnutar" (on Mac OS X) to pack up the binary distribution
     (Mac OS X "tar" seems to be buggy - the resulting tarballs appear to be
     corrupted, e.g. when opening them with Midnight Commander. GNU tar seems
     to extract them correctly, though)

   twice, if --enabable-thread-safe-client is enabled, which broke RPM
   building)


Makefile.am:
   - reverted previous patch and removed libmysql_r from SUBDIRS again
     (it will automatically be added when --enable-thread-safe-client is
     used)

into sinisa.nasamreza.org:/mnt/work/mysql


sql/log.cc:
  Auto merged

  The problem yesterday in ERROR_LOCK_VIOLATION was not InnoDB Hot Backup, but some file system backup tool: add to file writes 100 retries with 1 second waits


innobase/os/os0file.c:
  The problem yesterday in ERROR_LOCK_VIOLATION was not InnoDB Hot Backup, but some file system backup tool: add to file writes 100 retries with 1 second waits

  Backport from 4.0: convert TL_READ_NO_INSERT to TL_READ to allow concurrent inserts to the table in INSERT INTO ... SELECT ... FROM table


sql/ha_innobase.cc:
  Backport from 4.0: convert TL_READ_NO_INSERT to TL_READ to allow concurrent inserts to the table in INSERT INTO ... SELECT ... FROM table

  Fix an assertion about ERROR_LOCK_VIOLATION 33 in file write if InnoDB Hot Backup is run concurrently with mysqld in Windows


innobase/os/os0file.c:
  Fix an assertion about ERROR_LOCK_VIOLATION 33 in file write if InnoDB Hot Backup is run concurrently with mysqld in Windows

This fixes a possible core dump problem in SHOW PROCESSLIST


sql/slave.cc:
  Only set thd->query to 0 if LOCK_thread_count is hold
sql/sql_db.cc:
  Only set thd->query to 0 if LOCK_thread_count is hold
  Also first set query_length, then query
sql/sql_parse.cc:
  Indentation changes

into hundin.mysql.fi:/home/heikki/mysql

  Backport from 4.0: fix bug in adaptive hash index search


innobase/btr/btr0sea.c:
  Backport from 4.0: fix bug in adaptive hash index search

(like FLUSH TABLES).
(Code is already in 4.0)


sql/log.cc:
  Fixed bug with the --slow-log when logging an administrator command
  (like FLUSH TABLES).

into sinisa.nasamreza.org:/mnt/work/mysql

  Backport from 4.0: fix the BLOB hang if the index tree is of height 1


innobase/row/row0purge.c:
  Backport from 4.0: fix the BLOB hang if the index tree is of height 1
innobase/btr/btr0btr.c:
  Backport from 4.0: fix the BLOB hang if the index tree is of height 1
innobase/include/btr0btr.h:
  Backport from 4.0: fix the BLOB hang if the index tree is of height 1

  Backport from 4.0: fix possible adaprive hash index latch hang


sql/ha_innobase.cc:
  Backport from 4.0: fix possible adaprive hash index latch hang

sql/net_pkg.cc:
  Fixed wrong packed length for packet > 16M.
  (Not used in 3.23)

into sinisa.nasamreza.org:/mnt/work/mysql

  Backport bugfix from 4.0: combined log file size >= 2 GB could cause log to be written in wrong place
btr0pcur.c:
  Backport bugfix from 4.0: index cursor restoration could theoretically fail


innobase/btr/btr0pcur.c:
  Backport bugfix from 4.0: index cursor restoration could theoretically fail
innobase/log/log0log.c:
  Backport bugfix from 4.0: combined log file size >= 2 GB could cause log to be written in wrong place

packet is issued.



BitKeeper/deleted/.del-net.c~ef21d6402bb882f9:
  Delete: libmysql/net.c
sql/sql_select.cc:
  reversing a fix, but we shall yet see about this one ...

(To make the current tests results work on windows)

This fixed a bug in GROUP BY on a BLOB column with NULL values.


myisam/mi_create.c:
  Fixed max_key_length when using UNIQUE keys.
myisam/mi_unique.c:
  Simple optimization
  Make different CRC for keys with null and empty strings.
mysql-test/r/group_by.result:
  Updated results
mysql-test/t/group_by.test:
  Test of bug

   (There are probably some more to fix)


libmysql/Makefile.shared:
   - clean up the vio_* and ctype-* symlinks that are created by link_source
     (thanks to Christian Hammers from Debian for the patch)
strings/Makefile.am:
   - remove ctype_autonf.c on "make distclean" (it's generated by "configure")

into mysql.com:/space/my/mysql-3.23-build

Safety fix to not cause conflicts with extended packages in 4.0.


sql/mysqld.cc:
  Safety fix to not allow one to specify too big max_alloc_packet
sql/net_serv.cc:
  Safety fix to not cause conflicts with extended packages in 4.0.
  (Should not cause any notable changes for 3.23 servers)