* master: (39 commits)
Add __debugInfo() to UPGRADING.
fix TS build
Update NEWS
Update NEWS
Update NEWS
Small tidy ups and raise visibility of GitHub PR process
Bug #41631: Observe socket read timeouts in SSL streams
wrap int8_t and int16_t with #ifdef to avoid possible clashes
- Updated to version 2014.6 (2014f)
Removed Countable::count() change info from UPGRADE.INTERNALS too
NEWS and UPGRADING for intdiv()
Revert "Add SO_REUSEPORT + SO_BROADCAST support via socket stream context option"
Fixed skip case for intdiv 64-bit test
Use callback structure
Add EXPECTF
Fix handling of multi-result sets with PS...used to clean not only the result set but the whole PS.
5.5.17 now
5.4.33-dev now
Add SO_REUSEPORT + SO_BROADCAST support via socket stream context option
Add SO_REUSEPORT + SO_BROADCAST support via socket stream context option
...
Conflicts:
ext/fileinfo/libmagic/softmagic.c
main/streams/xp_socket.c
implies more memory copy. The old method is still available and can be used.
It stays as default. Choosing the method is through a flag to mysqli_query()/mysqli_real_query()
New mode can be forced with an INI setting, for all extensions that support this mode
(ext/mysql and mysqli, because PDO due to it's architecture can't support it)
The setting is mysqlnd.fetch_data_copy=[0|1]
Bug#62046 mysqli@mysqlnd can't iterate over stored sets after call to
mysqli_stmt_reset().
Stmt's state was changed to STMT_PREPARED and leads the driver to think
that the stmt was just prepared. Because result set exist also for only
prepared but not executed sets, the check for stmt->result isn't any good.
There is a result set, because as response for COM_PREPARE the server sends
metadata and metadata is part of a result set. After COM_EXECUTE the server
sends metadata again, which could be different and in this case more correct.
Bug #61411 PDO Segfaults with PERSISTENT == TRUE && EMULATE_PREPARES == FALSE
Wrong allocation, that doesn't follow the scheme of using stmt->persistent
was the root cause of the problem and the crash at free.