This is a very good topic. It appears that there are to schools of thought. 1) The addition of outside inputs (rock dust for example) is ok as long as it is a short term application to solve long term problems. So its okay if you use inputs to jump start a system, perhaps applying rock dust for 3 -5 years but it should not be a long term solution 2) No outside inputs are needed, simply planting a diverse polyculture will take care of nutrient needs for all plants.
In Sepp Holzer's new book he talks about how plants exchange nutrients via regular/natural cycles of root death and mycellium which can transfer these nutrients considerable distances in exchange for sugar (carbohydrates) from its host plant. It is not only nitrogen from nitrogen fixing plants which can be exchanged but all nutrients can be exchanged. For example it is possible to plant comfrey next to a fruit tree. The comfrey has a deep root which seeks out potassium that the tree may not be able to reach. When the comfrey roots die as they do every day (plants roots grow and die back cyclically every day, this is healthy and normal) the mycellium in the soil pick up the now free potassium in the soil and exchange it with the fruit tree in exchange for carbon.
The second method is the more natural of the two but some people are concerned (rightly so) that if a nutrient such as selenium is not present in the soil or sub soil it can not be exchanged not matter what plants you have growing.
I would think in most situations you can get everything you need with polyculture but some rare places may be completely missing some necessary elements. In this case importing things like rock dust is justified.
Hope this helps.
catch it before it slithers away! Oh wait, it's a tiny ad: