c302 issues to write

c302-issues.txt

- Pass in reader objects rather than names of readers for for data_reader
  arguments

    Allows to configure readers in any way that's needed prior to use (e.g.,
    the spreadsheet location, PyOW context). There doesn't seem to be any
    reason that users of the data reader either need to initialize or resolve
    the readers differently for a given name, so it's better not to expect them
    to.

- Remove assumptions of spreadsheet reader being used

    In method names in particular, but possibly subtler distinctions as well.

- Replace unqualified relative imports with explicit relative imports

    In Python 3, such imports do not work any longer.

- Update documentation and code to run `c302/c302_*.py` as python `-m c302.c302_*`

    Relative imports break in possibly unexpected ways when you run a script
    from inside of its package directory rather than let python do it for you
    with `-m`

- Make a proper unit test suite

    There's a test.sh script (which also doesn't work with the python 3
    updates) that does not have clear pass/fail which is unhelpful for
    diagnosing errors.

- [owupd0] Update OpenWormReader to use the latest version of PyOpenWorm

    As a first approximation, can just hard-code
    <http://openworm.org/entities/bio#worm0>
    as the context to query against (it's sort of the 'top-level' context
    produced by insert_worm.py into the OpenWormData repo). Will be modified in
    owupd1 to its final form.

- Compare running times of NeuroML file generation with OpenWormReader vs
  Spreadsheet reader 
  blocked by: [owupd0]
    
    OpenWormReader will undoubtedly take longer. Must determine whether the
    difference is substatial enough to merit doing query of the PyOW database
    at a different phase (i.e., with a `pow translate`). This had been a
    limiting factor to inclusion in c302/CElegansNeuroML previously.

    This can be done prior to the full 

- Create a c302 DataTranslator and DataSources 
  blocked by: openworm/PyOpenWorm#364
  blocks: [owupd1]

    PyOpenWorm has a notion of DataTranslators that describe how data get from
    one form to another and each of those forms is called a DataSource. A
    DataSource should be created to describe the input to c302 and a DataSource
    should be created to describe the output NeuroML files. The DataTranslator
    should have a `translate` method which runs c302 with the first DataSource
    type and outputs the second. The first DataSource should have a Translation
    linking it back to the PyOpenWorm insert_worm DataSource (TBD:
    openworm/PyOpenWorm#364).

- [owupd1] Update OpenWormReader to use the latest version of PyOpenWorm

    Should accept a DataWithEvidenceDataSource identifier for the source of
    neuron network and muscle data, etc.