Re: mysqlnd_qc (mysqlnd query cache) 1.0.1 with prepared statement support

On 13.07.2010, at 23:27, Ulf Wendel wrote: > Lukas Kahwe Smith schrieb: >> On 13.07.2010, at 21:16, Ulf Wendel wrote: >>> Lukas Kahwe Smith schrieb: >>>> On 13.07.2010, at 19:55, Ulf Wendel wrote: >>>>> Lukas Kahwe Smith schrieb: >>>>>> On 13.07.2010, at 19:26, Ulf Wendel wrote: >>>>>>> Benjamin Eberlei schrieb: >>>>>>>> Actually missing prepared statements and unbuffered queries were >>>>>>>> the >>>>>>>> only issues I had with the Query Cache so far. Can i compile the >>>>>>>> Mysqlnd >>>>>>>> + Query Cache PECL Trunk against 5.3.99-DEV only to test it? >>>>>>> ... remember that PDO emulates prepared statements by default. >>>>>> by default? since when? >>>>> Since when does it not? >>>>> >>>> good question. >>>> it certainly didnt in the beginning: >>>> >>>> http://netevil.org/uuid/444a6017-0548-2459-2943-44a601714d58 >>>> as you can see the feature was added later on and had to be enabled manually. >>> Well, a blog post... one would have to check the CVS log but I'm too lazy/tired... >>> To add one "I recall and in this blog...", emulation should have been the default when we >>> started adding mysqlnd support in mid 2008. Whenever... >>> >>> Question is if and when to disable the emulation default in favour of native PS. The >>> answer may be a bit like with ext/mysql, it may be: fear the installed user base. >> The documentation also only talks about emulation being used in case its not natively >> available. I have not looked at the source, but if indeed this has changed, then it ought to be >> documented and I am not sure I really like this change, especially in light of the rather crude and >> flaky parser used and the simplistic quoting logic used. > > You are not the first one to be surprised about the status quo and you are certainly an > insider. > > Yes, the PDO emulation is not the most sophisticated one. But I don't need to tell you > about the shortcomings of the MySQL server and the drawbacks of native prepared statements. Although > PDO_MYSQL will fallback to non-prepared statements, if the server hints that a statement cannot be > prepared this does not solve the issue with the MySQL (Server) Query Cache for many. Also, the extra > round trip to find out that a statement cannot be prepared is not desired. > > It is not an easy decision. Especially for me who always yells about abusing PS for security... > > How to move forward: improve documentation? ok moving the discussion to the pdo list. i guess first up we need to confirm if this is a general default or just the mysql default and when this all changed. regards, Lukas Kahwe Smith mls@pooteeweet.org