The parser methods used to try to handle Strings, as well as valid Rack
inputs. This led to some unnecessary complexity, giving rise to errors
like those seen in #56. This refactors to handle only IO-like Rack Input
streams.
Most of the work here is just to reorganize the tests.
There is an odd case where RDFSource#update feeds the IO stream before
trying to parse a graph. For now, we aggressively #rewind the stream
in #parse_graph to this problem. This might be unduely costly, and
there is probably a better fix in the long run.
The later commits introduce Rack::Lint and deal with a few issues found by running the existing tests and the W3C suite with those in place.
The parser methods used to try to handle Strings, as well as valid Rack inputs. This led to some unnecessary complexity, giving rise to errors like those seen in #56. This refactors to handle only IO-like Rack Input streams.
Most of the work here is just to reorganize the tests.
There is an odd case where
RDFSource#update
feeds the IO stream before trying to parse a graph. For now, we aggressively#rewind
the stream in#parse_graph
to this problem. This might be unduely costly, and there is probably a better fix in the long run.The later commits introduce
Rack::Lint
and deal with a few issues found by running the existing tests and the W3C suite with those in place.