I think the plugin/program logic needs to recognize the year associated with a series in a filmography is not the same as the year associated with the series...
Thinking about this some more, I remind myself the issue is somewhat larger than this. The database design is based on the idea of visible vs. invisible movie records. If a movie is part of one's collection, it's visible. If it only exists because the movie is in a person's filmography, it contains only the information necessary for that purpose, and it's invisible. Theoretically, any one movie in the database should be in one state or another. If an invisible movie is added to one's collection, the existing invisible record should be made visible. In other words, there should be no duplicates.
In practice, however, this is not what's happening. I've already pointed out the problem with series records and years. Another problematic situation is where one is using the option to include only "main page" actors (the top 15, in credits order). BTW, I find this option essential for restricting the number people in the database to a manageable level. When people records are updated, their full filmography is added. For actors, that filmography is likely to include movies in the collection for which the person is not one of the top 15 credited. Exactly what happens in this case, I'm not sure. Based on what I see in my database, it's either of (1) the actor is added to the existing movie record, even though they are not one of the top 15; or (2) a duplicate invisible record is created to record the movie detail as presented in the filmography. In case (1), updating the movie will remove the actor from the movie credits, and the movie is removed from their filmography—a "no win" situation where neither a people or a movie update will fix the problem. In case (2), it seems the next time this happens with a different actor, the same duplicate invisible record is not necessarily used to record the filmography data.
I thought it would be interesting to see what impact this situation has had on my database. So I attempted to view my "invisible" movie records. Unfortunately, my tired old computer has difficulty handling a list with 270,000 (!!) items. I understand that number is going to be large, but I have to wonder if it's
that large because of this issue. I was able to examine one interesting case before my computer choked.
The X Files series had 48 duplicate invisible records. The visible record showed the top three actors (for reasons I don't understand, these are the only ones on the IMDb main page) followed by six unknown actors who had appeared in one to three episodes. Updating the record, as expected, removed these six actors.
I wasn't able to see if movies are similarly affected, but I suspect they are. When I look at filmographies, it's very common to see movies that I know are in my collection, but which are not highlighted as such. Ctrl-clicking such a link confirms it to be an invisible duplicate. This, of course, is expected. What I'm unsure of is whether there are multiple invisible duplicates—as there are for series.
I'm not sure what the solution is. I suspect it necessarily involves a significant change to the database structure, along with how the program handles these relationships. In other words, something to consider for the next major version, not a bug fix. Perhaps movie records for filmographies need to be completely distinct from "visible" records, and the two kept in sync in terms of credits.
