Re: Performance improvements using writer.delete vs reader.delete

Heh. I suppose I'll defer to your judgment. In my mind, the
simple 
system to make is to just buffer the adds, buffer the
deletes - later 
apply the adds, apply the deletes (or the reverse). I am
sure something 
in Solr would have a more sophisticated process, but my
guess was about 
what the new Lucene user would throw together. But again,
I'll defer to 
your judgment anytime.

- Mark

Mike Klaas wrote:
>
> On 3-Aug-07, at 3:27 AM, Mark Miller wrote:
>
>> Also, IndexWriter probably buffers better than you
would. If you 
>> buffer a delete with IndexWriter and then add a
document that would 
>> be removed by that delete right after, when the
buffered deletes are 
>> flushed, your latest doc will not be removed. Its
unlikely your own 
>> buffer system would work so well.
>
> Is it?  ISTM that any sane buffering system would
account for that case.
>
> -Mike
>
>
------------------------------------------------------------
---------
> To unsubscribe, e-mail: java-user-unsubscribelucene.apache.org
> For additional commands, e-mail: java-user-helplucene.apache.org
>
>

------------------------------------------------------------
---------
To unsubscribe, e-mail: java-user-unsubscribelucene.apache.org
For additional commands, e-mail: java-user-helplucene.apache.org