Is the Lucene 2.9 TokenStream API faster than the old one?
I'm looking into upgrading from 2.4 to 2.9 and noticed all the contrived code that handles attributes. Just wondering if anyone has any opinions on whether this will change with his .9 and that will change when 3.0 is gone.
I am confused about how creating attributes by reflecting and binding attributes on a map can be as efficient as simply having fields for each of the attribute types.
Is it worth upgrading to 2.9 today or just waiting for 3.0 when everything is complete. It doesn't matter to me for updating today.
source to share
"When it's complete" is actually not an accurate summary of Lucene 3.0 vs. 2.9. Version 2.9 contains all the same API updates and changes as 3.0.
All new features were added in 2.4, the release was numbered 2.9 (to make it clear it was a "special" version), and 3.0 just comes along and removes two things that "hold things": namely, it will remove all deprecated methods and classes (many of which hang for a long time) and the Java 1.4 compatibility requirement will be dropped (so Java class classes, generics, etc. will be introduced as needed).
Basically, instead of taking a huge leap in the clipping of APIs and functions in one version (which means people who needed functions had to change all their old legacy calls), functions were added first, so people can take advantage of the change and then focus on removing the use of deprecated APIs, knowing that 3.0 will basically be a drop-in replacement - just cleaned up, no new real features.
source to share