Document Haystack indexing behavior

[murphyke]

I wanted to document indexing behavior, especially regarding the indexable attribute of DataField and DataConcept objects, but I'm not sure where to do that: create new DataField and DataConcept pages in the gh-pages API documentation?

The behavior is as follows:

• The metadata for all published, unarchived concepts are always indexed. The concept metadata include name, plural name, description, keywords, category name, category description, category keywords, and, if the category has a parent category, the parent category name, parent category description, and parent category keywords. The indexing does not recurse all the way up the category hierarchy.
• If a concept is indexable, then the metadata for the concept's fields are also indexed with the concept. The field metadata include name, plural name, description, keywords, model name, and, if the field has a category, the same category metadata as for concept (although fields do not usually have a category). Tip: a public, unarchived concept should normally be indexable.
• If a concept is indexable and a constituent field is indexable, then the data values (labels) are also indexed with the concept. Tip: if data values for a field could meaningfully be queried as a way of locating an associated concept, then the field should be indexable. Typically, numeric, date or time fields should NOT be indexable.

Aside: I'm wondering if the DataFieldIndex class defined in search_indexes.py is needed; it may just cause redundant indexing (because DataConceptIndex drives down into constituent fields anyway)?

[bruth]

I invited you to GitBook: https://www.gitbook.com/book/dbhi/harvest/details which is backed by the new repo I created here: https://github.com/chop-dbhi/harvest-book. There is an web client and a OS X client to add and edit pages of the "book". I want to consolidate all the documentation for Harvest here.