Wednesday, November 25, 2009, 10:35:06 AM, you wrote:

RK> On 22.11.2009 17:05, Marnen Laibow-Koser wrote:
>> Ralph Shnelvar wrote:
>> [...]
>>> RK> ?  Btw, having columns with indexes in their name is considered bad
>>> RK> schema design because you either could normalize it into another 
>>> table
>>> RK> and join or provide more meaningful names instead of "x02" if it is 
>>> used
>>> RK> as a placeholder.  My 0.02EUR.
>>>
>>> RK> Kind regards
>>>
>>> First of all, I'm a noob.  I should have said that.
>>>
>>> Thus your solution
>>> (1) Makes sense.
>>> (2) Explains a lot about Ruby.
>>>
>>> For both I thank you.
>>>
>>>
>>> In terms of normalizing the database, I ran some timing tests and
>>> normalizing it slows me down a couple of orders of magnitude.  I'm
>>> willing to live with the lack of purity for the sake of performance.
>> 
>> A couple orders of magnitude?   This suggests to me that your 
>> normalization and/or queries are poorly done.  What is chewing up so 
>> much time, and what does the normalized schema look like?  It should be 
>> possible to find a solution that performs reasonably and doesn't have 20 
>> repeating fields.

RK> I second that.  Ralph, any information about the slowness?  Is it in the
RK> database or outside?

I _think_ it is, in part, coming from the logging that is being done.

Instead of one create ... one can have 21.