← Back to team overview

maria-developers team mailing list archive

Re: Doubt in bin logging for CREATE VIEW

 

Hi Sriram.

To clarify, the missing view features
(union, restrict/cascade, subquery)
are major projects outside the scope of
your GSoC project, so don't bother
spending time on it.

It would take months just
to test changes that large.

Just admire how much is
not done even after 20 years. :)

Thanks, James.

--------------------------------------------
On Thu, 6/19/14, sriram patil <spsrirampatil@xxxxxxxxx> wrote:

 Subject: Re: [Maria-developers] Doubt in bin logging for CREATE VIEW
 To: "James Briggs" <james.briggs@xxxxxxxxx>
 Cc: "maria-developers@xxxxxxxxxxxxxxxxxxx" <maria-developers@xxxxxxxxxxxxxxxxxxx>
 Date: Thursday, June 19, 2014, 2:46 AM
 
 Hi James,
 Well, I liked the
 good news part. :)
 I will share the test results with you too. And yeah, will
 look into the create view comments.
 
 
 Thanks,Sriram
 
 On Thu,
 Jun 19, 2014 at 2:48 PM, James Briggs <james.briggs@xxxxxxxxx>
 wrote:
 
 
 Hi Sriram.
 
 
 
 The good news ... nice catch, you earned your GSoC
 salary!
 
 
 
 The mysql_register_view() has a dozen
 
 different ways to fail:
 
 
 
 - RAM: my_error(ER_OUT_OF_RESOURCES, MYF(0));
 
 - disk: sql_create_definition_file()
 
 - name collision
 
 - etc.
 
 
 
 So the create view statement can
 
 definitely fail on the master,
 
 and should not replicate when that happens.
 
 
 
 Note that warnings can also be set, so
 
 investigate if that affects replication:
 
 
 
 push_warning(thd,
 
 Sql_condition::WARN_LEVEL_WARN,
 
 ER_WARN_VIEW_MERGE,
 
 ER(ER_WARN_VIEW_MERGE));
 
 
 
 The bad news ... I read over the create view source,
 
 and there's tons of work that hasn't been finished
 yet
 
 (just read the comments about union and restrict/
 
 cascade for starters.)
 
 
 
 Another future project!
 
 
 
 Thanks, James.
 
 
 
 
 
 
 


References