tnzmnjm
commented
4 months ago - Merging #49 into this issue
- Consider how we could record the information about what happened and return it in the results (in turtle format). That way NLnet can review the results and understand what went wrong when we tried to gather the information about the tests for a particular project (where a project is a line of data in the original TSV file).
Context
We are already discovering flaws with some of the project data e.g. where the URL for the project's repository doesn't map to a current repo. (See https://github.com/commercetest/nlnet/issues/3 for various examples of the issues.)
Currently we skip many of these entries since we cannot discern which repo to query for tests. It may be useful to track and report these flaws e.g. to the team at NLnet with the aim of improving the upstream data about the projects.
The reporting might as well be in RDF format to facilitate automated consumption of the information. Therefore the results need to be in three column format AFAIK. TBD what information we can usefully provide and how to encode the results e.g. when would a binary result be acceptable (e.g.
valid|invalid) vs. more nuanced results (e.g.accessed_ok|unrecognised_as_a_code_repo|HTTP Error{nnn} received)?