Factorie postagger is not using sentence segmentation

[dirkgr]

Somewhere deep in the bowels of Factorie, it says that the postagger might work better when sentence information is available. When you run an all-Factorie workflow (rather than using the nlpstack wrappers in sequence), it uses the sentence annotations. To be as safe as possible, and as true to an all-Factorie workflow as possible, we should do the same.

[schmmd]

What sentence information does it use? What sentence information don't we provide to Factorie?