Code coverage I think is going to be poor for this repo, so will be great to see. The changes coming up in #93 are large so coverage I think would make sense to start on after we merge #93 (or in the https://github.com/mapbox/mapnik-vector-tile/tree/mapnik-geometry branch).
A big reason why coverage is crap is the lack of a good harness in C++ for writing visual tests and/or tests that use external fixtures (like directory of expected geojson fixtures). So I think we should start on that. I still like the idea of this existing outside of mapnik.
The JSON input and output grammars in the mapnik-geometry branch are way nicer now (since they work on mapnik::geometry objects rather than vertex_converters (aka paths) so using them to read and serialize fixtures would be great.
@zerebubuth, @kevinkreiser and myself have written tests for Avecado which implicitly test m-v-t. We did some tests verifying PBF outputs and examining the Mapnik layers that m-v-t created.
Code coverage I think is going to be poor for this repo, so will be great to see. The changes coming up in #93 are large so coverage I think would make sense to start on after we merge #93 (or in the https://github.com/mapbox/mapnik-vector-tile/tree/mapnik-geometry branch).
A big reason why coverage is crap is the lack of a good harness in C++ for writing visual tests and/or tests that use external fixtures (like directory of expected geojson fixtures). So I think we should start on that. I still like the idea of this existing outside of mapnik.
Resources:
mapnik::geometry
objects rather than vertex_converters (aka paths) so using them to read and serialize fixtures would be great.