db-derby-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Øystein Grøvlen <Oystein.Grov...@Sun.COM>
Subject Re: Modified UTF-8 or UTF-16 for temporary Clobs?
Date Wed, 23 May 2007 17:39:22 GMT
Mike Matrigali wrote:
> Oystein Grovlen - Sun Norway wrote:
>> Mike Matrigali wrote:
>>  > What is the expected usage pattern for an update on a clob that uses
>>> these "temporary" clobs?  What is the usual input format, what is the
>>> usual output format?  Do you expect more than one update usually?  Does
>>> an update have to rewrite the end of the file on and shrink or expand of
>>> a middle of the clob?
>> In client/server, a single update may cause a sequence of updates 
>> against the embedded driver.  This is because the locator 
>> implementation  uses a stored procedure to perform the update, and 
>> Derby does not support Clob as procedure parameters.  To overcome 
>> that, the Blob is split into segments of max size for VARCHAR.  Hence, 
>> it is important at the overhead of positioning is not big in these 
>> cases.  One way to optimize for this would be to remember the end 
>> position from last write.
> How do these calls relate to calls to the store code for the update?  Is
> there still one call to store or one per segment?

The calls work on a temporary copy of the Blob. Store is affected when:

1) ResultSet.updateRow is a called and the Blob is a column in the 
result set.

2) PreparedStatement.execute is called and the Blob is a parameter.

At that point the Blob is streamed into Store the same way as before.


View raw message