db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Daniel John Debrunner <...@apache.org>
Subject Re: Collation implementation WAS Re: Should COLLATION attribute related code go in BasicDatabase?
Date Wed, 21 Mar 2007 13:53:38 GMT
Mamta Satoor wrote:
> Dan, why would we need to override the StringDataValue 
> getValue(RuleBasedCollator collation) method in CollatorSQLChar? I think 
> it will be sufficient to implement this method in SQLChar with following 
> pseudocode
> StringDataValue getValue(RuleBasedCollator collation)
> {
>    if (collation == null)
>    {
>      // Just need UCS_BASIC collation, so just return myself.
>      return this;
>    }
>     CollatorSQLChar s = new CollatorSQLChar();
>     s.setValue(this);
>      s.setCollator(collation);
>     return s;
> }
> I see this method being called when store is trying to construct a 
> template row of DVDs based on the format ids and collation type that 
> store is aware of. Since the formatid for character types will always 
> correspond to SQLChar (the base class)(for simplicity, I am only talking 
> in terms of SQL type CHAR. Same logic applies for other SQL 
> character types) , the only thing that store might want is to use the 
> super class CollatorSQLChar as DVD if the collation type is 
> 1(TERRITORY_BASED). When do you see the need for Derby code wanting to 
> go from CollatorSQLChar to SQLChar?

1) When needing to collate using USC_BASIC.

2) If CollatorSQLChar does not override that getValue(RuleBasedCollator 
collation) then the class would not be honoring the api contract, which 
will eventually lead to bugs as other new code assumes the class is 
honoring the contract of the api.


View raw message