A question on indexing StoredFields (using the compressioncodec):
I have a flat file storing in sequential order contents I intend to add to
stored fields. I keep a list of lengths in memory:
(byte,byte),(byte,byte,byte)... <--- file
[2,3,...] <-- memory
where each element correspond to the storedField for a doc.
doc1: 2 bytes
doc2: 3 bytes
Currently, I would need to read them into a BytesRef allocating either
bunch of small byte or load this entire file into a rather large byte.
Do you think it makes sense to change the StoredField api to read from a
more generic version of BytesRef, that it can point either to file or