The static class with the "disable forced good item" field is gone. Now, certain tests that want to check for specific access progression can run a method that removes the forced good item and adds it back to the pool. Tests that don't care about this will collect the forced good item like normal. This should prevent the intermittent fill failures on complex doors unit tests, since the forced good item should provide enough locations to fill in.
The most serious issue this PR addresses is that entrances that use doors without items (a small subset of doors when door shuffle is on, but *every* door when door shuffle is off, which is the default) underestimate the requirements needed to use that entrance. The logic would calculate the panels needed to open the door, but would neglect to keep track of the rooms those panels were in, meaning that doors would be considered openable if you had the colors needed to solve a panel that's in a room you have no access to.
Another issue is that, previously, logic would always consider the "ANOTHER TRY" panel accessible for the purposes of the LEVEL 2 panel hunt. This could result in seeds where the player is expected to have exactly the correct number of solves to reach LEVEL 2, but in reality is short by one because ANOTHER TRY itself is not revealed until the panel hunt is complete. This change marks ANOTHER TRY as non-counting, because even though it is technically a counting panel in-game, it can never contribute to the LEVEL 2 panel hunt. This issue could also apply to THE MASTER, since it is the only other counting panel with special access rules, although it is much less likely. This change adds special handling for counting THE MASTER. These issues were possible to manifest whenever the LEVEL 2 panel hunt was enabled, which it is by default.
Smaller logic issues also fixed in this PR:
* The Orange Tower Basement MASTERY panel was marked as requiring the mastery doors to be opened, when it was actually possible to get it without them by using a painting to get into the room.
* The Pilgrim Room painting item was incorrectly being marked as a filler item, despite it being progression.
* There has been another update to the game that adds connections between areas that were previously not connected. These changes were additive, which is why they are not critical.
* The panel stacks in the rhyme room now require both colours on each panel.
Fixes an issue where the Silph Co 6F hostage check becomes unavailable if Giovanni has been defeated on 11F. This is due to the NPC having separate scripts depending on whether Giovanni was defeated. The code for the check has been moved to before the branch.
Changes the finished_game to new variable so now it only checks the game's memory and if it has sent the finished flag before
Fixed ag2 not requiring 1 of each black magic
Fix hitlist if you exclude summon level 7 and have summon levels option turned off
As discovered by this bug report https://discord.com/channels/731205301247803413/1182522267687731220 it's currently possible to accidentally have the starting weapon of a player overwritten by a triforce fragment if TriforceLocations is set to dungeons and StartingPosition is set to dangerous. This fix makes sure to remove the location of a placed starting weapon if said location is in a dungeon from the pool of possible locations that triforce fragments can be placed in this circumstance.
Two items on Route 120 are on the other side of a pond but were considered accessible in logic without Surf.
Creates a new separate region for these two items and adds a rule for being able to Surf to get to this region. Also adds the items to the existing surf test.
Basically, the function for "checking entrances both ways" only checked one way. This resulted in unreachable locations.
This affects Expert seeds with (non-remote doors and specific types of EP Shuffle), as well as seeds with non-remote doors + specific types of disabled panels + specific types of EP Shuffle.
Also includes two changes that makes spoiler logs nicer (not creating unnecessary events).
Currently when a room shuts down while clients are connected it instantly spins back up. This fixes that behaviour categorically.
I still don't know why or when this problem started, but it's certainly wreaking havok on prod.
`colorama` has type stubs when it didn't before
`ZillionDeltaPatch.hash` annotated type could be `None` but md5s doesn't allow `None`
type of `CollectionState.prog_items` changed
`WorldTestBase` moved
all of the following are related to this issue:
https://github.com/python/typing/discussions/1486
CommonContext for `command_processor` (is invalid without specifying immutable - but I don't need it anyway)
ZillionWorld options and settings (is invalid without specifying immutable - but I do need it)
Trevor L
CaitSith2
Fabian Dill
Star Rauchenberger
Alchav
Zach Parks
Silvris
Aaron Wagener
lordlou
PoryGone
Yussur Mustafa Oraji
Bryce Wilson
NewSoupVi
Rjosephson
JaredWeakStrike
Doug Hoskisson
t3hf1gm3nt
beauxq