To me, it's all the same thing. I have a repository of reference material, and I have a repository of tasks to accomplish, but most of the tasks require the reference material. So having it all in one heavily linked and searchable pool seems to obvious thing to do, with a queue of "pages" that come to hand, or get put back into the queue.

Separating them just gives me two unrelated pools that I then have to cross-correlate.