[hibernate-dev] HSearch ideas

[Hardy Ferentschik]

Hi,

I was reviewing parts of the HSearch code over the last few days and  
wanted to run a few ideas past you.

- Indexing process
What about exposing lucene indexing paramters like mergeFactor,  
maxMergeDocs and minMergeDocs? You could allow the
user to add these parameters to the hibernate configuration file and then  
pick them up within the Workspace/SearchFactory class?

- Filter
Now that HSearch allows to pass down a Sort object to the Lucene Searcher,  
wouldn't it make sense to also allow Filter instances to be used, eg  
DateFilter and QueryFilter? Similar to sorting you could just add a setter  
for the Filter instance to FullTextQuery.

- Daemon timer for FSMasterDirectoryProvider and FSSlaveDirectoryProvider
I reviewed the current implemention of these two classes, because I had  
problems shutting down Tomcat when using HSearch in a Master/Slave  
configuration. Since the Timer instances in these classes are not started  
as daemons the timer threads seem to hang around when attempting a  
shutdown. However, as far as I can see it should be save to make them  
daemon timers without changing the implementation.

- Keyword substitution
I noticed that almost all classes start with '//$Id: $', but none of them  
had the svn:keywords property set. Shouldn't keyword substitution be  
turned on?

-- Hardy