On Mon, Jan 3, 2011 at 8:26 PM, Yonik Seeley <yo...@lucidimagination.com> wrote: > On Mon, Jan 3, 2011 at 2:03 PM, Simon Willnauer > <simon.willna...@googlemail.com> wrote: >> While we are on it, would it make sense to move omitTfAP into the >> Index enum. It always felt odd that you can omit norms using the enum >> but use a setter to omit TF & Pos. > > I think the attempted move to type safety / enums is what added the > most complexity to Field, and I think it's a mistake to go further > down the path of combinatorial explosion. I don't care if new enum > values are made, just please don't remove the setters.
In 4.0 we have all the freedom to clean this up and make it extendible, consistent and easy to use. It might be practical to use a bitset for the "features" should be enabled / disabled in the indexer per field. This is going to be important once we make a codec consume the entire document and boolean flags need to be transported down to the codec. Yet, I am not a big fan of enums really since extending them is pretty hard though. On the other hand a setter is really just sugar on top of some representation and in that case really an inconsistency. One more evidence that we should fix Field and its whole hierarchy. simon > > -Yonik > http://www.lucidimagination.com > --------------------------------------------------------------------- To unsubscribe, e-mail: dev-unsubscr...@lucene.apache.org For additional commands, e-mail: dev-h...@lucene.apache.org
