summaryrefslogtreecommitdiff
path: root/ext/mysqlnd/mysqlnd.c
Commit message (Collapse)AuthorAgeFilesLines
...
* | constifyAndrey Hristov2012-10-121-1/+1
| |
* | if ssl has been switched on transmit clear-text passwordAndrey Hristov2012-10-031-0/+3
| |
* | fix login failure with empty password. This is 5.5 specific - new codeAndrey Hristov2012-09-281-2/+2
| |
* | change public key setting to PERDIR and finish code that allowsandrey2012-09-251-0/+1
| | | | | | | | to set the key file programatically per mysqli_options()
* | remove old unicode code, that is was compiled out by using a macroandrey2012-09-251-16/+0
| |
* | Add SHA256 authentication support - password hashing to mysqlndandrey2012-09-251-5/+6
| | | | | | | | Automatic switchover to SSL with plain-text password is not part of this
* | plug a leak when the server ask the client for a auth protocol, thatandrey2012-08-221-8/+9
| | | | | | | | | | the client doesn't understand. This is 5.5 specific bug, as a result of a refactoring in 5.5
* | fix valgrind warningandrey2012-07-091-1/+1
| |
* | mergeandrey2012-06-291-0/+7
| |
* | close the underlying stream as early as possible and so notify theandrey2012-06-011-7/+6
| | | | | | | | NET layer
* | close the stream whenever possible, no need to keep it open altough theandrey2012-06-011-0/+5
| | | | | | | | connection is marked as unusable
* | more refactoring stuff. Reduced code duplication in the authenticationandrey2012-05-311-216/+170
| | | | | | | | code
* | small refactoring - factour out code from connection establishment.andrey2012-05-301-101/+153
| | | | | | | | | | Add a define for the default authentication protocol - get away from the hardcoded magic value
* | refactor MYSQLND_NET, split it two parts for easy resharingandrey2012-04-301-23/+26
|/
* Fix foldingJohannes Schlüter2012-04-221-1/+1
|
* - Year++Felipe Pena2012-01-011-1/+1
|
* Add a middleman which can intercept the calls and can reroute themAndrey Hristov2011-11-021-3/+28
|
* - shift a bit of code aroundAndrey Hristov2011-11-021-13/+13
| | | | | - when cloning use the methods of the original object
* fix trace log messagesAndrey Hristov2011-11-021-2/+2
|
* Now really split the method tables and have differentAndrey Hristov2011-10-311-308/+324
| | | | | | | methods for the handle and the data object. Add auxiliary functions to work with the new methods. Add possibility to clone a connection object - shadow copy.
* Split struct MYSQLND in struct MYSQLND and struct MYSQLD_CONN_DATA.Andrey Hristov2011-10-311-91/+116
| | | | | A step in the direction of keeping internal data private
* Move from directly referencing an aggregated structure to using aAndrey Hristov2011-10-251-121/+132
| | | | | | pointer to a structure. The structure is still aggregated but we add a level of indirection for possible plugins to overwrite the storage
* Add hooks for __call like functionality and also allow toAndrey Hristov2011-10-251-239/+428
| | | | | | handle calls that spread over two internal functions. Let's call it a local TX.
* move plugin specific stuff to one file. create object factoryAndrey Hristov2011-10-241-45/+5
| | | | | for similar objects
* shift code around to two new files - mysqlnd_driver.c Andrey Hristov2011-10-211-113/+18
| | | | | | and mysqlnd_ext_plugin.c (mysqlnd extension plugin)
* api_ext to reverse_apiAndrey Hristov2011-10-191-38/+3
|
* MYSQLND_NET refactored not to use any direct reference to MYSQLND (the ↵Andrey Hristov2011-10-101-12/+14
| | | | | | | connection) QC will need an ifdef-layer to compile with this
* - Fix foldingJohannes Schlüter2011-09-251-1/+1
|
* - fix build (macros mess)Pierre Joye2011-09-171-1/+1
|
* - fix TS buildPierre Joye2011-09-171-1/+1
|
* Fixed Windows buildDmitry Stogov2011-09-161-1/+1
|
* - Fix TSRM buildJohannes Schlüter2011-09-011-2/+2
|
* - Register extensions using mysqlnd (mysql, myslqi, pdo_mysql) with mysqlndJohannes Schlüter2011-08-311-0/+38
|
* fix valgrind warnings in debug buildsAndrey Hristov2011-08-261-1/+1
|
* relax constraintAndrey Hristov2011-08-101-1/+1
|
* Add mysqli_error_list() that returns an array with errors. Typically onlyAndrey Hristov2011-08-041-0/+28
| | | | | | | | one and just one for libmysql. mysqlnd can return generate more than one error during its work and with mysqli_error() only the last error is being reported. In the array returned by mysqli_error_list() / $mysqli->error_list, all errors will be found. The list is reset when the next command is executed
* longer error messages, don't cut themAndrey Hristov2011-07-271-2/+2
|
* free data in case of failure in the callee not in the callerAndrey Hristov2011-07-051-5/+3
|
* reuse codeAndrey Hristov2011-05-311-16/+11
|
* fix valgrind warningAndrey Hristov2011-05-041-1/+2
|
* add a proxy for sppintf and vspprintfAndrey Hristov2011-03-181-18/+18
|
* less trace infoAndrey Hristov2011-03-181-22/+2
|
* move from macros to functions, which can be overwritten, if neededAndrey Hristov2011-02-251-1/+41
|
* Don't lose this data anymore. It worked without the fix,Andrey Hristov2011-01-311-2/+3
| | | | | but it with it is just correct (trunk only).
* clear error before doing anything, because we can getAndrey Hristov2011-01-281-0/+3
| | | | | | | | | | trapped by previous error : php -r '$link = mysqli_init(); var_dump($link->real_connect("127.0.0.1", "unknown", "unknown")); var_dump($link->real_connect("127.0.0.1", "known", "password"));' shows the problem. The second connect did not work, as it should be.
* small cleanupsAndrey Hristov2011-01-241-2/+2
|
* Handle MySQL 5.5 authentication features.Andrey Hristov2011-01-241-59/+198
| | | | | | | | Authentication protocol can be changed, a new raw packet is introduced, which includes only the "encrypted" data for the auth plugin, sent after change protocol (0xFE) is sent to the client.
* fix ZTS buildAndrey Hristov2011-01-191-1/+1
|
* Update on the mysqlnd auth plugins, changed the pluginAndrey Hristov2011-01-141-11/+14
| | | | | | | | structure and where they load themselves. 0xFE packets (method switch) needs to be done, additional wire-level packet to be created to be parsed and the packet doesn't fit the PACKET_OK structure anymore.
* Change things to allow passing of the password lengthAndrey Hristov2011-01-141-4/+9
| | | | | | | to mysqlnd. This is needed as a password might include a \0 and thus we need to be binary safe.