Can you clarify if it's really helpful to split based on this versus just splitting on a single space character? My impression was that segmentAndStem converts extraneous whitespace to a single space, but I may be wrong

I feel like this would be more readable

Can you clarify in what scenarios viewer won't be set?

It doesn't make sense to have a required parameter after optional ones.

We can make the optional parameters required (but with an optional type), eg. cursor: ?string instead of cursor?: string. But perhaps we should just reorder so that the required parameter comes first?

The limit has been increased by 1 in the most recent diff update. It was my impression that we weren't planning on doing the "fetch 1 extra" strategy anymore, though I don't remember why that decision was made.

@inka, can you clarify if we're using the "fetch 1 extra" strategy still, and why the change was made?

Looks like we're using the "fetch 1 extra" strategy in D7651 now. So that answers the first question... just curious about the second question. Basically, I recall that we were initially doing this, and then stopped doing it, and now looks like we're doing it again... curious for context on that

Initially I used it somewhere else - in D7117, when I was fetching messages to be processed entirely on the server side. It didn't make sense there because the server had to do the same amount of work either way (run the query the same number of times).
Here I am calling this endpoint from the client, who benefits from knowing if the results have ended - they don't have to call the server again. The server runs the same number of queries as if it didn't fetch and ignore the additional entry, but the client doesn't have to call the server and get an empty result.
So this is a different situation than before. It makes the code on the client side more readable as well.

Here is what you wrote on that diff, that explains it:

We do something similar in message-fetchers.js (assume that if the result count is the same as the LIMIT, then there is more to fetch), but the difference there is that we have another variable we pass to the client that indicates if we think the result count is "exhaustive" or not. If we weren't passing that variable to the client, then there would be no point

Thanks for explaining!