SMALL KEY CACHE

The server crashed on division by zero because the key cache was not
initialized and the block length was 0 which was used in a division.

The fix was to not allow CACHE INDEX if the key cache was not initiallized.
Thus never try LOAD INDEX INTO CACHE for an uninitialized key cache.

Also added some windows files/directories to .bzrignore.

The buffer was simply too small.
In 5.5 and trunk, the size is 311 + 31,
in 5.1 and below, the size is 331


client/sql_string.cc:
  Increase buffer size in String::set(double, ...)
include/m_string.h:
  Increase FLOATING_POINT_BUFFER
mysql-test/r/type_float.result:
  New test cases.
mysql-test/t/type_float.test:
  New test cases.
sql/sql_string.cc:
  Increase buffer size in String::set(double, ...)
sql/unireg.h:
  Move definition of FLOATING_POINT_BUFFER

Update copyright comment in innochecksum.

patch so that it can later be compared with patchs with expanded symlinks 

The new --secure-file-priv checks dereference any symlinks in the paths and compare the 
resolved paths.
Thus the 5.0 test suite must do as the 5.1 and up and avoid using symlinks.

The query was re-written *after* we had tagged it with NON_AGG_FIELD_USED.
Remove the flag before continuing.


mysql-test/r/explain.result:
  Update test case for Bug#48295.
mysql-test/r/subselect.result:
  New test case.
mysql-test/t/explain.test:
  Update test case for Bug#48295.
mysql-test/t/subselect.test:
  New test case.
sql/item.cc:
  Use accessor functions for non_agg_field_used/agg_func_used.
sql/item_subselect.cc:
  Remove non_agg_field_used when we rewrite query '1 < some (...)' => '1 < max(...)'
sql/item_sum.cc:
  Use accessor functions for non_agg_field_used/agg_func_used.
sql/mysql_priv.h:
  Remove unused #defines.
sql/sql_lex.cc:
  Initialize new member variables.
sql/sql_lex.h:
  Replace full_group_by_flag with two boolean flags,
  and itroduce accessors for manipulating them.
sql/sql_select.cc:
  Use accessor functions for non_agg_field_used/agg_func_used.

  USING '..' ON WINDOWS

Backport of the fix to 5.0 (to be null-merged to 5.1).
Moved the test into the main test suite. 
Made mysql-test-run.pl to not use symlinks for sdtdata as the symlinks
are now properly recognized by secure_file_priv.
Made sure the paths in load_file(), LOAD DATA and SELECT .. INTO OUTFILE 
that are checked against secure_file_priv in a correct way similarly to 5.1 
by the extended is_secure_file_path() backport before the comparison.
Added an extensive test with all the variants of upper/lower case, 
slash/backslash and case sensitivity.
Added few comments to the code.

The 5.0 fix. Removed unreferenced files and the directory 
that has them.

             DEFINITION OF ANY ROUTINE.

This follow-up patch removes SHOW PROCEDURE CODE from the test
case as this command is only available on debug versions of the
server and therefore caused the test to fail on release builds.

             DEFINITION OF ANY ROUTINE.

The problem was that having the SELECT privilege any column of the
mysql.proc table by mistake allowed the user to see the definition
of all routines (using SHOW CREATE PROCEDURE/FUNCTION and SHOW
PROCEDURE/FUNCTION CODE).

This patch fixes the problem by making sure that those commands
are only allowed if the user has the SELECT privilege on the
mysql.proc table itself.

Test case added to sp-security.test.

Didn't build on Solaris.

The LGPL license is used in some legacy code, and to
adhere to current licensing polity, we remove those
files that are no longer used, and reorganize the
remaining LGPL code so it will be GPL licensed from
now on.

Note:  This patch only removed LGPL licensed files
       in MySQL 5.0, and is the first of a set of
       patches to remove LGPL from all trees.
       (See Bug# 11840513 for details)



include/my_compare.h:
  Mostly code moved in from my_handler
include/my_global.h:
  AIX-only code.   Function used to be in my_port.c
  Inlining instead.
libmysql/Makefile.shared:
  my_gethostbyname and my_port is removed
myisam/mi_check.c:
  ha_find_null is moved from my_handler and made static.

Reverse DNS lookup of "localhost" returns "broadcasthost" on Snow Leopard (Mac), and NULL on most others.
Simply ignore the output, as this is not an essential part of UDF testing.