[schapplm]

Those id's are provider specific... maybe that isn't a great key to use -- or maybe we could do a secondary match on name if we don't find a match on imdbnumber. I had a look at the JSON RPC code, and although the kodi db tracks where the id came from, json rpc doesn't expose it. My initial though was that perhaps we should include that along with the show id.

We could probably submit a patch to expose it if we can't think of a better way to correct this in the addon.

I think for correcting these scraper related issues, the WatchedList addon is not the right place. I would also favor for fixing the json output or adding keys. The main problem seems to me, like you mentioned, that different ids are stored in the same field in the database. The match on name would also pose some issues: e.g. names for shows/movies scraped in different languages (I guess many users have English media and media in another language).

Did we always use the json rpc or did we used to go right to the db files with sqlite?

From what I remember, it was always json. This was promoted very early as the cleanest access on the database (and also independent of the sqlite/mysql implementation of the video db).

When there is no IMDB no. and debug logging is on, errors to the effect of "Movie xy doesn't have an IMDB no., please try re-scraping" are logged. [...]
This also seems to affect the WatchedList add-on. All moves which have been scraped with the TMDB scraper exhibit the behaviour described in the beginning (log item). Maybe someone more tech-savvy than me could make a suggestion to the maintainer of the scraper how to rectifiy this situation.

I always wondered, where that came from. But this perfectly makes sense. As for the tv shows I also support this to be done on scraper/database/json-API-level.