gm72

Thought of creating a side-by-side comparison if this needs any further clarification:

The changes introduced with snapshot 22w07a have it so the visible bookshelf (rightmost in screenshot side-by-side.png) is inactive while the hidden one (second from left) is active.  Prior to that the behavior was opposite and intuitive.

A lot of thought and effort is being put into the Villager Trade Rebalance Experiment at the moment.  Closing this issue (again, see MC-71354) with "works as intended" is - with all due respect - a little disappointing.

The explanation "Random does what Random likes." is not very convincing.

Consider the enchanting table mechanics:  You will not end up with an enchanted piece of armor that has both e.g. Protection and Fire Protection, you will not end up with a pick axe that has both Fortune and Silk Touch, you will not end up with a sword that has both Sharpness and Smite, and so on.  I am NOT suggesting that this would make sense, let alone that this should be possible.  My point is that the developers were clearly able to implement a solution that prevents incompatible enchantments, that restricts Random's hand ever so slightly.

So it isn't inconceivable that librarians can be "trained" to not offer the same enchantment multiple times, be it the same or different level or price.  In the least, make it so that the more advanced trades can only be cheaper or higher-level enchantments than before.  Lower levels or higher prices don't make any sense.

Adding screenshot "intuitive.png" to illustrate my case:

Without a doubt, obstructions like the honey blocks should disable bookshelves in place of the yellow terracotta, and obstructions like the slime blocks should disable bookshelves in place of the dark green terracotta in the corners.  However, the sides of bookshelves in place of the lime green terracotta are visually obscured by the slime blocks so these bookshelves' effect of increasing the available enchantment levels should also be cancelled by the slime blocks (as is the case before snapshot 22w07a, and in release 1.18.1 and older) and not by the honey blocks (as of 22w07a, in 1.18.2 and likely still in 1.19).

Enchanting items is still 100% possible, so I'd agree that this is more of a change than it is a bug.  However, the process has (a) become less intuitive (you have to place a torch on the left red bookshelf in binary.png to block the brown bookshelf) and (b) requires more torches (prior to this a torch in the corners could block up to six bookshelves, now only three).  "Degrade" isn't the word I'd choose (where there's a will there's a way) but the experience certainly has not improved with this change.

A different way to describe the blocking behavior is that in v18.1 and before, the space taken up by a stack of two bookshelves was treated as one logical unit, i.e. a single obstruction either on the lower level (like the carpet in MC-2474) or on the upper level (like the trapdoors in MC-245857) would block bookshelves on both levels.  The fix for MC-245857 seems to have been to separate the two levels, i.e. blocking a bookshelf on the upper level now requires an obstruction on the upper level, while at the same time that doesn't affect the bottom bookshelves any longer.

What violine1101 is describing as "overcompensated" is that this same fix in 22w07a also changes the bookshelves that an obstruction is affecting.  Before, an obstruction directly to the side (East, North, South, West) of the enchantment table would only affect one column of bookshelves, while an obstruction diagonally would affect the three columns directly to the side of the obstruction and in the very corner of the setup.  As of 22w07a, an obstruction to the side blocks three columns of bookshelves, while an obstruction in the corners only affects one.  So if only this latter (inadvertent?) part of the change was reversed, it would help with my complaint (a) above and make enchanting more intuitive and familiar again.  Due to the independent levels it would still (b) require more torches than before but I think that's an acceptable compromise as it would leave the fix for MC-245857 in place.

Thank you for your time and expertise, violine1101!  As far as I can tell you haven't misinterpreted anything.

This is consistent with all the earlier observations.  In the original layout.png all 15 bookshelves are active because the torches could only block bookshelves in the very corners of the 5x5 deepslate brick square.  Similarly, in positions.png the torch doesn't have an effect because there is no bookshelf on the rightmost deepslate block, the snow blocks the three bookshelves in the bottom row on the right (hence the level 24 enchantment in the third slot), and the trapdoors block two bookshelves each in the top resp. bottom row on their side.

Added two more screenshots, binary.png and example.png.  The first one (binary.png) shows a possible layout if you're looking to systematically go through the available enchantments in v1.18.2 (color codes like for resistors):  A torch etc. where the brown glass is will disable the one "brown" bookshelf (represented by brown concrete), a torch for the red glass will disable the two red bookshelves, two torches for the yellow glass will disable the four yellow bookshelves, and four torches for the gray glass take care of the remaining eight gray bookshelves.  In example.png I'm using a regular torch to disable the top left bookshelf (brown in binary.png) and two soul torches to disable the four bottom bookshelves (yellow).  With a total of five bookshelves inactive the third slot offers a level 20 enchantment, down 5 x 2 = 10 from 30.

Tried the other way around, i.e. placing an obstruction and then vary the position of a bookshelf to see where it is blocked.  Unless I've lost my marbles the blocking works in a more horizontal than vertical fashion in v1.8.2.  Please see the screenshots lower.png and upper.png.  If an obstruction is placed where you see a glass block, any shelf in the area of the concrete blocks of the same color has no impact, i.e. doesn't increase the available enchantment levels.

Created a new world in v1.18.1 and used different ores to mark the floor, please see new-layout.png.  In v1.18.1 the available enchantment levels (= "AELs") without obstructions are 8/13/30.  No matter whether I'm using torches, snow or trapdoors (both on the ground and up high as in MC-245857, please see positions.png) as obstructions, the results are the same:  AELs 8/8/28 when there is any obstruction above the redstone ore (blocking the single bookshelf), and AELs 4/12/26 when placed above the emerald or diamond ore (blocking two bookshelves).  This is what I'm used to seeing, the AEL for the third slot to drop by two for each blocked bookshelf.

Next I upgraded the new world to v1.18.2.  Without any obstructions the AELs are 8/15/30.  No matter what is placed above the redstone ore, the AELs are still 8/15/30, i.e. nothing is blocking the single bookshelf.  With a torch, snow or a trapdoor directly on the emerald ore, the AELs are 6/11/24 (odd that the third slot is down from 30 by six instead of four, given that two bookshelves would appear to be blocked).  It gets even more interesting though:  With a trapdoor "high" above the emerald ore or any obstruction above the diamond ore the AELs are 5/17/26.

Something doesn't seem right, and I'm sticking with that story! 😉

Unless I'm misinterpreting what I'm seeing, MC-245857 is resolved because it was marked as a duplicate of MC-2474, and the latter is unresolved.  Do you know of any changes that have actively been made to fix MC-245857 and when those have been made?

Good point reg. the collisions.  I'll try my scenario with carpets/snow instead of torches after work.  However, I can already state that a torch straight next (i.e. not diagonally) to the enchantment table does reduce the available enchantment levels.

MC-245857 – and MC-2474 which it refers to – speak of 16 bookshelves and that fewer of them than the stated expectation are increasing the available enchantment levels.  The scenario here uses 15 bookshelves but more than expected are active.  I don't dispute that there may be a connection but I'm afraid I can't view this as intended behavior.  I would expect the carpet in MC-2474 and the trap doors in MC-245857 to block the bookshelves.  Just as I would expect my torches to block nine of them – but they do not, as of v1.18.2.

This seems to be related to the height map issue mentioned in the first comment (see MC-179952).  Placing the northwest sapling into a spot for which the debug screen shows good height map values allows the tree to grow, even if the other three saplings have bad height map values.  With a bad height map value for the NW sapling but good values for the other three the tree does not seem to grow.

Snapshot 20w18a restores the ability to optimize a world, and that process appears to correct the corrupt height map data.  Trees are growing again, thank you!