Chiseled Bookshelves keep comparator output after being emptied

Can confirm.

This line in the changelog suggests that this is intentional:

Comparators can detect the last book placed/removed

Similarily, chiseled bookshelves have a block state last_interaction_book_slot that also could mean that this is intentional.

However, this seems inconsistent with every other type of container in the game, and it's not immediately obvious to me why this behaviour would be desirable redstone-wise, as it means some actions with the bookshelf cannot be detected by a comparator (e.g. taking out the 6th book - the comparator output does not change).

My suspicion is that the original concept of the chiseled bookshelf might have allowed for players to take the book out of any arbitrary slot, in which case it would make sense to remember which was the last book interacted with. However, in its currently implementation as a stack / LIFO queue you're only able to take out the book that you put in last, in which case this behaviour does not seem too useful to me.

Yeah, I could see knowing the last slot interacted with being interesting if you could interact with specific slots and add/remove books from any position. I was hoping that was the case when it was first announced. It certainly makes for more interesting decoration options and puzzle possibilities that way.

As is, it seems really hard to be useful for anything Redstone related as the same action of adding/removing a book from the first and last slots results in the same result.

I think it'd work better if you could interact with individual slots to add or remove the specific books from any of the 6 specific positions. Then have Redstone outputs of 1-6 and 7-12 to represent book added and book removed respectively (or 1-2 be add/remove for slot 1, 3-4 be add/remove for slot 2, etc...). This way you could track a sequence of steps with some RSNOR latches, and AND gates to detect a sequence like 3, 8, 4 meaning book added to slot 3, book removed from slot 2, and book added to slot 4, and reset the latches on any other inputs. Similar to existing combination lock systems today that use item frames.

Pulling out and detecting add vs. remove would help in puzzle/escape room options as well, especially as needing to take a specific book out means that the specific books could contain the instructions needed for the next piece of the puzzle. Since the books could have different instructions, pulling out the specific one needed from the shelf as dictated by an earlier puzzle clue is an interesting addition.

If it remains that you have to put the books in and take them out in order, then it should probably work like all the other containers and just output the signal strength of how many items it contains.

Otherwise, if we can pull books out in any order based on which slot we're looking at, then it'd be cool to have a different system to detect more complex interactions. Default would probably be 0 for when it's empty, but I don't think once books are added it'd ever go back. As if you remove the last book, you want the specific signal strength to detect which slot that book was removed from and not that it's empty again. Unless the bookshelf outputs the specific signal strength and then after a few ticks goes back to zero if the shelf is empty. (Not sure if knowing the shelf is completely full with a strength of 13 in a similar fashion would be useful, but a possibility as well, but again after an initial delay of knowing the signal strength of the specific slot where a book was added.)

Why is this "confirmed"? They even said at Minecraft Live "you're actually going to pull a Redstone signal that's gonna tell you what is the last book that was interacted with." So this must be the intended behavior. 

Yes, the targetted intent is not how it is implemented. So from the perspective of how it is implemented now this is a bug that it keeps emitting. So there are two ways to fix it. Either remove the extra signal emitted when all books are removed. Or implement the targetted bookslot mechanic so that a specific book can be removed so that it will emit the signal strength representing that slot.

Allowing specific slots to be removed or placed with books would allow for more decoration variants as well. But I presume how the texture works as well wouldnt really allow for that kind of shift without doing all possible slot combinations as textures and make a mask/table lookup for it.

I see this been marked as solved, but the bookshelf has been changed so you can now choose which book you remove and that means the lack of reset on the signal is very much a bug still.

It means you can't detect when a book is removed at all, which makes the bookshelf use a lot more limited.

It's resolved as intended, therefore it is not a bug.