Brain

=====

* added case restriction for field names (they now must be lowercase)

=====

* added CachedConnection class, which noticeably increases speed of reading operations for slow connections
* added getRemoveConflicts() for connection objects
* generalized connection classes structure, moved parts of logic to separate classes

=====

* fixed minor bug in read(), when it returned empty list when no fields were found
which match given masks
* search() uses 2 requests to DB engine instead of (number_of_conditions + 1)
* delete() uses much less requests to DB engine (approximately one request per affected table)
* read() uses less requests to DB engine (number_of_fields + 1)
* refactored modify() and insert() logic in order to reduce calls to database:
now using slightly more than one query per table affected
* conflicts checking logic is slightly more strict now - creating list/map
in place of value is considered a conflict too
* changed search() behavior a little - now NOT acts more logically

-----

* fixed fuzz tests work (it was broken after moving FakeConnection to tests in previous release)
* fixed several bugs with lists autocreation: when new lists are created during modify()
or insert(), their elements are filled with Nones
* fixed bug in read() logic - it did not raise error when reading from non-existent list element
* fuzz test: added tests for all parameters of read() and delete()
* fuzz test: added tests for remove_conflicts parameter in modify() and insert() actions
* made fuzz test return times for each type of request separately
* added non-atomic performance tests: time for functional tests and times for each request in fuzz test
* constructing test suites hierarchically

-----

* removed Marshaller/Unmarshaller hooks from xmlrpchelpers
* returning list [id, data, id, data, ...] instead of dictionary from dump()
* simplified xmlrpchelpers structure (since custom Marshaller is not needed anymore)
* using lists instead of tuples in search()
* added support for long search conditions
* added note about mandatory keyword parameters dictionary in XML RPC server help.
* not using 'brain.test' in tests (i.e, make brain and brain.test independent)

-----

* documentation was moved to test module and can be now fed to doctest.testfile()
* fixed several minor typos in documentation
* added examples with raised exceptions to documentation
* using set() when checking search() results in doctests
* fixed bug with non-created hierarchy during autovivification
* added policy for conflicts removal as connection parameter
* added ability to run doctests to test\run.py