Extra consistency checks on category codes and URLs

[citizenlab]

Extra consistency checks on category codes and URLs

[darkk]

Check №2 is a bit tricky for cis.csv. Should cis.csv be treated in a same way as global.csv for corresponding countries? What definition of CIS should it use? E.g. should it include Georgia?

[hellais]

At OONI we don't actually use cis.csv at all and that country list has not been updated in a pretty long while. I would go to the extent of suggesting we remove it or move it to another directory.

[sneft]

For point 1, I have no doubt there are a number of these inconsistencies. We tried to fix these as we encountered them but haven't ever made a systematic effort to clean them.

For point 2, I agree that if a URL is present on the global list it should not be on a local list. Our old testing system flagged when you attempted to upload a local list with a URL duplicated in the global list. Our logic was that the global and local lists are meant to be run as a single unit, so we wanted to avoid duplication. I know we had some cases where this was inconvenient (e.g. wanting to test a very narrow sample in a bandwidth-limited place) but to my knowledge OONI is flexible enough to better accommodate custom lists for special circumstances.

(This requirement does add a small burden of labour on list compilers, as in my experience the average person compiling a local list will reasonably (and often appropriately) add certain URLs that are duplicated in the global list. Perhaps this is just a matter of good documentation and instructions to list compilers.)