Introduction, and separating data from implementation

Hi all,

Nice to meet you - I'm Eric Mill, a developer at the Sunlight Foundation. I
work on a few things here relevant to this project, including our Congress
API <http://sunlightlabs.github.com/congress/> and on the
github.com/unitedstates project (along with several non-Sunlighters) that
gathers the data around members of Congress.

I was just having a conversation with James off-list that he wisely
suggested would be good to have on-list. :) My concern was that the data
schema was a little too intermingled with implementation details.

Namely, I was offput by the spec suggesting specific MongoDB collection
names, and discouraging the use of specific out-of-spec fields like "_type"
because of how they might interfere with common patterns in MongoDB object
document mappers (ODMs). I use these ODMs myself and understand the
concern, but I also believe that as a new data schema attempting to get buy
in from people and organizations with vastly differing tech stacks, it'd be
smaller and less intimidating to make it clear the minimum requirements
necessary to publish data that complies with Popolo as a data spec.

People interested in building higher level implementations - like
interoperable Rails engines, Django apps, CPAN modules, whatever - would
probably also want to agree on some best practices there, but those to me
are separate from how to publish bulk data whose schema agrees with each
other.

James, like the good man he is, has already taken my concerns to heart and said
he plans <https://github.com/opennorth/popolo-standard/issues/24> to update
the spec to make the separation more clear. Still, if anyone else has
opinions on this sort of thing, we should talk about them!

-- Eric

-- 
Developer | sunlightfoundation.com