Core: Plando Items "rewrite"

[Silvris]

What is this fixing or adding?

This PR is the combination of three major changes to plando items.

• Moves plando items to the options api as a member of PerGameCommonOptions
• Changes the placement of items to use fill_restrictive, which means plando should be capable of avoiding placements that can never result in a playable world
• Plando now evaluates item and location groups.

How was this tested?

Manually via generating games with yamls using plando items. Opening as draft since any game using pre_fill will need thorough testing with this approach (KDL3 was already found to be bugged).

Ran unittests, but more unittests should likely be added.

[Exempt-Medic]

It seems like this removes the ability to have plando blocks without any locations listed. For example, so that specific items end up in specific worlds. Is the intended workaround to use Everywhere?

[Silvris]

It seems like this removes the ability to have plando blocks without any locations listed. For example, so that specific items end up in specific worlds. Is the intended workaround to use Everywhere?

I believe I missed this usecase (plando items guide doesn't mention it, and it's built into get_unfilled_locations_for_players so I couldn't tell it from the plando code).

[BadMagic100]

Disclaimer: I have not looked at the code yet

It might be worth considering and documenting/fixing possible negative interactions with #3032 if you haven't already

I'm also curious about how pre_fill is impacted as it may affect generic ER

[beauxq]

Some people use plando when they WANT to make something that will be seen as unbeatable according to the logic.
Will this still be possible with this change?

[Silvris]

Some people use plando when they WANT to make something that will be seen as unbeatable according to the logic. Will this still be possible with this change?

It might still be possible, but current methods (such as locking the item behind a location that logically requires it) will no longer work. It should be noted that this doesn't prevent impossible combinations (that usually end in fill error) such as filling your entire sphere 1 with junk.

[Silvris]

Disclaimer: I have not looked at the code yet

It might be worth considering and documenting/fixing possible negative interactions with #3032 if you haven't already

I'm also curious about how pre_fill is impacted as it may affect generic ER

This should be fine, as we use the actual instance of the item instead of creating a new item and then removing using that. It's definitely worth testing though.

It appears that any world that uses pre_fill to place items that are not in the itempool must also define get_pre_fill_items so that the state used for plando can account for those items.

[Silvris]

List of worlds that implement pre_fill and not get_pre_fill_items:

• Adventure (does not place progression in prefill)
• Blasphemous
• Castlevania 64 (does not place progression in prefill)
• Hylics 2
• KH2
• KDL3 (KDL3: Version 2.0.0 #3323)
• Pokemon Emerald
• Pokemon RB
• Shivers
• Super Metroid (does not place progression)

Worlds that implement get_pre_fill_items but do not actually include all pre_fill items

• A Link to the Past
• Raft (very funny)

I will need to check all of these later to confirm if this is the proper reason for some worlds having issues.

[NewSoupVi]

✅️

I do still wish upon a star every day that the actual PR that addresses the Witness thing gets merged one day

[Jouramie]

I only looked at the stardew test, LGTM.

[ScipioWright]

Approved for TUNIC, also I'm deleting that code anyway
#4908

[hatkirby]

Lingo changes look fine to me. I've opened a PR to remove the plando stuff entirely anyway (#4910).

[Exempt-Medic]

Did some more testing, not a ton, but enough. I'm fine with the behavioral changes this brings and definitely fine with all the bugs it fixes. ALttP changes are waiting on Berserker still

[Berserker66]

LttP is probably fine, like with many LttP changes I always have the suspicion that while it looks fine it breaks in some "hilarious" way months later, but we'll only find out if we send it.