From the announcement:
"This release should fix the final bugs we accidently got into 3.23.34
and a long security bug that has been in MySQL a long time!
The main fixed bugs are that UPDATE didn't always use keys when
updating on something not based on a primary key and that 'affected
rows' wasn't returned to the client if the mysqld server wasn't
compiled with support for transactions."
From the announce mail:
"The 3.23.35 mainly fixes a critical bug in 3.23.34 with ORDER BY. We
don't know how the ORDER BY bug slipped through our testing suite or
how fatal it's really is, but as we have got a couple of reports about
core dumps regarding this, we recommend 3.23.34 users to
upgrade as soon as possible. Sorry for the inconvenience."
This type allows SQL transactions (COMMIT, ROLLBACK).
Note: BerkeleyDB is build as part of mysql323-server since mysql
needs a special version of it. Thus we can't depend on
databases/db3 at the moment.
mysql323-client installs libmysqlclient.so.10 (instead of
libmysqlclient.so.6).
mysql323-{client,server} are production quality, now (according to
MySQL AB at least).
and mysql322-server.
- mit-pthread support was dropped.
- libmysqlclient.so.9 was renamed to libmysqlclient.so.6 (like the
one coming with MySQL-3.22.x) so ports depending on MySQL having
a chance of using 3.23.x if installed.
mysql322-(client|server) and mysql323-(client|server) can't be
installed at the same time, of course.
- use --enable-assembler for ${ARCH}==i386
Requested by: Martin Nilsson <martin@filex.se>
There is an annoying bug in gcc 2.7.2.*: If -pipe is used for .s files,
gcc will "hang" forever, because the assembler is called with '-' and
noone is sending data.
As a workaround use .S (capital s) files. Then cpp will be called prior
to as and thus a pipe can be build.
Set MIT_PTHREADS to use the supplied mit-pthreads.
This is a workaround for PR ports/12824.
- replace one occurance of gcc with ${CC}
- don't use CFLAGS twice somewhere in mit-pthreads compilation
(There is still one place left where CFLAGS are used twice.)
This upgrade contains - as far as I can see - no functional change
in regard to the port. But I'm tired of receiving requests or patches
for an upgrade. People are focused on version numbers too much. *sigh* ;-)
