This is a modifiction of my previous patch after receiving feedback. This is a better way to fix the problem. With this patch, data directory and index directory will use only forward slashes (/) when on Windows.
mysqldump.c:
Removed fixPaths routine. Was improper fix for bug #6660
sql_show.cc:
Changed append_directory to convert backslashes to foward slashes when on Windows.
sql/sql_show.cc:
Changed append_directory to convert backslashes to foward slashes when on Windows.
client/mysqldump.c:
Removed fixPaths routine. Was improper fix for bug #6660
character set conversion of a string constant.
mysql-test/r/ctype_utf8.result:
Adding test
mysql-test/t/ctype_utf8.test:
Addign test
sql/item_strfunc.cc:
Remove previous fix: it was not 100% correct.
A non const_item() can return a constant String.
sql/sql_string.h:
A new method to cut Alloced_length to str_length for
string constants, to avoid reusing them as a buffer
in things like CONCAT().
ndb/src/ndbapi/NdbEventOperationImpl.cpp:
Proactive compile fix (in case we ever enable "EVENT_DEBUG" on IRIX):
Two identical declarations on loop level will collide at function level.
sql/filesort.cc:
A fix (bug #8799: Killed filesorts can fail inited==RND assertion in ha_rnd_end).
Should call ha_rnd_end() only if ha_rnd_init() was called.
"mysql-pro-gpl-cert" packages as well (thanks to Kent for the suggestion)
Build-tools/Do-solaris-pkg:
- fixed package name matching regex to be satisfied with "mysql-pro-gpl-cert" packages as well
(thanks to Kent for the suggestion)
in the 4.1.10a RPMs for SLES9/RHEL3:
- small improvement: use LSB functions to display startup success and failure, if available.
Fall back to more primitive builtin functions otherwise.
- joined two pieces of code performing the same functionality into one "wait_for_pid" function
- added a "reload" function (LSB requirement)
support-files/mysql.server.sh:
- small improvement: use LSB functions to display startup success and failure, if available.
Fall back to builtin functions otherwise.
- joined two pieces of code performing the same functionality into one "wait_for_pid" function
- added a "reload" function (LSB requirement)
Add diagnostics to track why ut_a(block->state == BUF_BLOCK_FILE_PAGE) failed in buf_flush_ready_for_replace() for a user
innobase/buf/buf0flu.c:
Add diagnostics to track why ut_a(block->state == BUF_BLOCK_FILE_PAGE) failed in buf_flush_ready_for_replace() for a user
Fix path that needs no parameterization.
man/mysql.1.in:
Fix path that needs no parameterization.
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
This really should not happen on Windows and part of the problem not fixed here is why show create table includes data directory when being run on Windows. However, this patch fixes the bug in mysqldump.c
mysqldump.c:
Added fixPaths function to convert \ to / in data directory and index directory entries only on Windows
client/mysqldump.c:
Added fixPaths function to convert \ to / in data directory and index directory entries only on Windows
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
- Moved static variables defined inside of function to file scope to avoid this linking problem on FC3
sql/ha_berkeley.cc:
Moved list of bdb extension to file scope
sql/ha_ndbcluster.cc:
Moved list of ndb extesions to file scope
BitKeeper/etc/logging_ok:
Logging to logging@openlogging.org accepted
dict_load_table(): Refuse to open ROW_FORMAT=COMPACT tables
of MySQL 5.0.3 and later.
innobase/dict/dict0load.c:
dict_load_table(): Refuse to open ROW_FORMAT=COMPACT tables
of MySQL 5.0.3 and later.
Added a test case for bug #9017.
item_cmpfunc.h:
A wrong not_null_tables method for Item_cond_xor
caused a conversion of a left join into an inner join
that was not valid.
sql/item_cmpfunc.h:
A wrong not_null_tables method for Item_cond_xor
caused a conversion of a left join into an inner join
that was not valid.
mysql-test/t/join_outer.test:
Added a test case for bug #9017.
mysql-test/r/join_outer.result:
Added a test case for bug #9017.
mysql-test/lib/mtr_report.pl:
Keep the Perl version in sync with the shell script so that 'Do-compile' will call
PS and ES tests even after some previous test failures.
(See previous change to 'mysql-test-run.sh' and 'Do-compile'.)
Many changes to improve error handling, and fix problem of multiple packaging processes trampling each other (32 & 64bit at the same time)
Build-tools/Do-solaris-pkg:
Many changes to improve error handling, and fix problem of multiple packaging processes trampling each other (32 & 64bit at the same time)
Item_func_equal::not_null_tables() always return 0.
mysql-test/r/join_outer.result:
Testcase for BUG#8711
mysql-test/t/join_outer.test:
Testcase for BUG#8711
sql/item_cmpfunc.h:
Fix for BUG#8711: "<=>" may have true value for NULL arguments, i.e. it can accept NULL-complemented
table rows. This differs from assumptions made in Item_func::not_null_tables(), so add
Item_func_equal::not_null_tables().
Item_func_equal::not_null_tables_cache value doesn't make sense now, but we still let it be
calculated (and ignore it)
adapt the Windows build files accordingly.
VC++Files/tests/mysql_client_test.dsp:
As the directory structure on Windows now resembles that on Unix (no more
"tests/tests", by the correction in "make_win_src_distribution.sh"), the location
of "mysql_client_test.*" has changed. Adapt the build tools accordingly.
VC++Files/tests/mysql_client_test_ia64.dsp:
As the directory structure on Windows now resembles that on Unix (no more
"tests/tests", by the correction in "make_win_src_distribution.sh"), the location
of "mysql_client_test.*" has changed. Adapt the build tools accordingly.
scripts/make_win_src_distribution.sh:
1) The semantics of "cp -R src targ" depend on the (non)existence of "targ".
By simply doing "$CP -R $i $BASE/$i", the directory hierarchy below $BASE differed
from the source for each "$i" that already existed as a target directory.
This is now prevented by checking the existence and adapting the command.
2) Comment where "$BASE" (the target directory) comes into existence.
3) Introduce an "abort" function as a start point for more sanity checks, use it.
ndb - Reenable MAX_OPEN_FILES to enable LCP in some situations
ndb/include/mgmapi/mgmapi_config_parameters.h:
Reenable MAX_OPEN_FILES to enable LCP in some situations
ndb/src/kernel/blocks/ndbfs/Ndbfs.cpp:
Reenable MAX_OPEN_FILES to enable LCP in some situations
ndb/src/mgmsrv/ConfigInfo.cpp:
Reenable MAX_OPEN_FILES to enable LCP in some situations
Better ibbackup message
Add a message explaining why we do a 'crash recovery' after an ibbackup restore; suggested by Tim Smith
innobase/log/log0recv.c:
Better ibbackup message
The reported problems were due to two completely unrelated omissions.
1) The file sort procedure didn't correctly create the sort key in
make_sortkey when the sortkey was an unsigned integer.
2) The name resolution procedure for column references inside a HAVING
clause did not propagate the unsigned_flag of the resolved references.
This patch corrects both problems.
mysql-test/r/select.result:
Added test result for BUG#7425.
mysql-test/t/select.test:
Added test for BUG#7425.
sql/filesort.cc:
Take into account whether 'item' represents a signed or an unsigned integer.
sql/item.cc:
Once an Item_ref is resolved, propagate the unsigned_flag to the resolved item.
