Item tooltip comparison and unique "Sort by" ignore skills granted by the candidate item
メンテナーはふだん 8 日以内に返信
まだ誰も着手していません。
評価
調査の方向性
Start with the granted-skill handling in src/Modules/CalcSetup.lua, especially the sections around lines 846, 1465–1520, and 1566–1578; then trace the CALCULATOR-mode comparison paths in Calcs.lua, ItemDBControl.lua, and ItemsTab.lua. Check how candidate item skills and removed item groups are represented in non-MAIN calculations. Done means item comparison tooltips and unique-item sorting include skills granted by the candidate, while replacement comparisons also handle the prior granted skill correctly.
索引モデルが issue の本文から書いたものです。
説明
What happens
Comparisons for an item that is not equipped (the unique/rare database "Sort by" stat sort and the item hover tooltip "Equipping this item will give you") never include a skill that the candidate item itself grants. The skill only starts to count after the item is actually equipped.
Example (derived from the code, not yet checked in the UI):
- Make a build whose main skill deals chaos damage with spells or attacks, and equip any one-handed sword.
- Items tab > Uniques, filter to One Handed Swords, hover United in Dream (Grants Level 25 Envy Skill). Or sort the list by Combined DPS.
- Equip United in Dream.
Expected: the tooltip/sort value predicts the DPS you get after equipping, including the added chaos damage from Envy.
Actual: the tooltip/sort value is computed without Envy. After equipping, DPS is higher than predicted, and items that grant skills (auras, buffs, triggered skills) sort lower than they should.
This is the other half of #8093 (that issue is about an equipped item's granted skill still being counted when the candidate replaces it). Both come from the same code path.
Cause
Socket groups for item- and tree-granted skills are only created or removed when env.mode == "MAIN":
- https://github.com/PathOfBuildingCommunity/PathOfBuilding/blob/a69391d6daaaea1aaf8054668a094aad88dd7eaa/src/Modules/CalcSetup.lua#L1468
- https://github.com/PathOfBuildingCommunity/PathOfBuilding/blob/a69391d6daaaea1aaf8054668a094aad88dd7eaa/src/Modules/CalcSetup.lua#L1480-L1520 (create or update the group for each entry in
env.grantedSkills) - https://github.com/PathOfBuildingCommunity/PathOfBuilding/blob/a69391d6daaaea1aaf8054668a094aad88dd7eaa/src/Modules/CalcSetup.lua#L1566-L1578 (remove groups whose item is gone)
The item comparisons run in "CALCULATOR" mode with a repSlotName/repItem override:
- https://github.com/PathOfBuildingCommunity/PathOfBuilding/blob/a69391d6daaaea1aaf8054668a094aad88dd7eaa/src/Modules/Calcs.lua#L100-L101
- https://github.com/PathOfBuildingCommunity/PathOfBuilding/blob/a69391d6daaaea1aaf8054668a094aad88dd7eaa/src/Classes/ItemDBControl.lua#L258-L263
- https://github.com/PathOfBuildingCommunity/PathOfBuilding/blob/a69391d6daaaea1aaf8054668a094aad88dd7eaa/src/Classes/ItemsTab.lua#L2578-L2579
In that mode env.grantedSkills does include the candidate's skill (CalcSetup.lua#L846, #L1465), but no socket group is made for it. Skill building only walks build.skillsTab.socketGroupList, so the candidate's skill never becomes an active skill in the comparison.
Possible fix
Work out the granted-skill groups for the env being calculated in every mode. In non-MAIN modes, build them into a per-env list instead of skipping them, and skip existing granted groups whose source item is not in env.player.itemList. MAIN mode keeps writing them to socketGroupList as it does now. That would also fix #8093.
- 主要言語
- Lua
- スター
- 5.5k
- フォーク
- 2.4k
- 平均マージ
- 6日 14時間
- マージ済み PR（30日）
- 13
環境構築
- Dockerfile または Docker Compose ファイルあり
- プルリクエストのテンプレートあり
- コントリビューションガイドを読む
はじめの一歩
- issue を最後まで読み、次にプロジェクトのコントリビューションガイドを読みます。
- 着手することを issue にコメントします — 二人が同じ作業をするのを防げます。
- リポジトリをフォークし、ブランチを切って変更します。
- issue 番号を参照したプルリクエストを送ります。
PathOfBuildingCommunity/PathOfBuilding のほかの issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 75/100
PathOfBuildingCommunity/PathOfBuilding#10431 ·
メンテナーはふだん 8 日以内に返信
-
難易度 2/5 1〜3時間 初心者へのやさしさ 79/100
PathOfBuildingCommunity/PathOfBuilding#10428 ·
メンテナーはふだん 8 日以内に返信
-
enhancement
難易度 2/5 1〜3時間 初心者へのやさしさ 60/100
PathOfBuildingCommunity/PathOfBuilding#10420 ·
メンテナーはふだん 8 日以内に返信
-
enhancement
難易度 2/5 1〜3時間 初心者へのやさしさ 65/100
PathOfBuildingCommunity/PathOfBuilding#10395 ·
メンテナーはふだん 8 日以内に返信
-
bug: accuracy
難易度 2/5 1〜3時間 初心者へのやさしさ 68/100
PathOfBuildingCommunity/PathOfBuilding#9911 ·
メンテナーはふだん 8 日以内に返信
PathOfBuildingCommunity/PathOfBuilding の issue をすべて見る
似ている issue
-
難易度 2/5 1〜3時間 初心者へのやさしさ 72/100
メンテナーはふだん 5 日以内に返信
-
enhancement priority::low upstream
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
メンテナーはふだん 1 日以内に返信
-
bug reviewed-by-AI
難易度 2/5 1〜3時間 初心者へのやさしさ 82/100
olimorris/codecompanion.nvim#3508 ·
メンテナーはふだん 1 日以内に返信
-
The Balanced badge stays on a card after it changes to an enhancement that isn't rebalanced対応中かも @Jr-JoshS が今日担当しました。 オープン
難易度 2/5 1〜3時間 初心者へのやさしさ 71/100
Balatro-Multiplayer/BalatroMultiplayerPvP#640 ·
メンテナーはふだん 1 日以内に返信
-
Babies count as citizens for housing and the lost-fort check, though a baby uses no bed or seatオープン
難易度 2/5 1〜3時間 初心者へのやさしさ 78/100
Layr-Labs/arena-fortress#141 ·
メンテナーはふだん 1 日以内に返信