Custom fields schema

[Myk Melez]

Sean McAfee wrote:

>Is this enough data to put this issue to rest?  Pretty please?
>  
>

Not really.  First, it doesn't measure the impact of indexes.  Second, 
it doesn't include the cost of looking up field IDs.  Third, it only 
measures search performance, but most queries retrieve bug data for 
specific bugs.  Fourth, it doesn't take what bbaetz says into account.  
And fifth, I never claimed bugs table columns were more performant, only 
that they were performant enough, which they are, considering the 
"custom fields" already in use as standard colums.

But the primary reason it doesn't put the issue to rest is that sparse 
fields go into separate tables under the fields-as-columns proposal as 
well, so that proposal gets any benefit to separate tables.

-myk