Suggested Patch for faster the creation of a FullTextLucene index when there's lots of existing rows.

[Angel Leon]

The attached patch is a suggestion to the H2 developer.

We decided to use the Lucene features in our project, currently FullTextLucene.java will trigger the adding and indexing of each document per row (indexAccess.writer.commit()), this is way too slow if you have existing data.

This patch, instead gives a boolean parameter to the insert() method, so that you can decide wether or not you want this indexing happening.

Once you're done, you can invoke a new method we've added called commitIndex() [you can rename that to whatever makes more sense to you if this name is not proper]

Basically, it doesn't make sense to trigger lucene's indexing mechanism for each existing row, at least not in our UI driven app.

If this is not the right place to submit this patch, please let me know, or if you know the developer it'd be great if he/she would get this email.

Cheers from the FrostWire Team.

[Matt Bishop]

Thomas can this be bounded by the current transaction?

>  FullTextLucene.java.patch
> 3KViewDownload

[Thomas Mueller]

Hi,

Thanks a lot for the patch! I will commit it.

Thomas can this be bounded by the current transaction?

Sorry I don't understand.

Regards,

Thomas