Skip to content

feat: make cluster jewels in the item crafter - #35

Merged
evrydayjunglist merged 2 commits into
mainfrom
feat/crafter-cluster-themes
Nov 13, 2026
Merged

evrydayjunglist merged 2 commits into
mainfrom
feat/crafter-cluster-themes

Conversation

@evrydayjunglist

@evrydayjunglist evrydayjunglist commented Nov 13, 2026 •

Copy link
Copy Markdown
Collaborator

What this changes

A player reported: a cluster jewel made with the item crafter is created, but it adds no passive when socketed, while one from /cluster_jewel works.

A cluster jewel is its enchantments, which no implicit list or pool names, so the crafter had nothing to put on the item. LocalExile's crafter catalog now lists them (LocalExile/LocalExile#948, its Agents/docs/dev/item-crafter.md, "Cluster jewels"), and this is Waypoint following it.

  • Choosing a cluster jewel base puts its enchantments on the item: "Adds N Passive Skills", which also holds the theme (what the small passives grant), and its jewel sockets. They are sent to /item_give in the enchant list.
  • The theme is a list of the size's themes by name ("Small passives grant"). The game's own value is a number, which it words itself, so Waypoint words it from the theme's stats: "Added Small Passive Skills grant: 12% increased Fire Damage".
  • The passive count has a slider with the base's range (8 to 12 on a Large Cluster Jewel), and the socket enchantment a list of its socket counts.
  • The theme decides the jewel's affixes: Prefixes and Suffixes list that theme's notables ("1 Added Passive Skill is …") with the modifiers every theme has.
  • An enchantment's card says "Enchantment", the preview lists enchantments first as the game does, and Add modifiers has a kind "Enchantments" on a base that has some, to put one back.
  • Import takes a copied cluster jewel's passive count, socket count and theme from its enchantment lines, and finds its notables among that theme's affixes.

docs/launcher.md is updated, and four interface texts are new, with Russian and Simplified Chinese (not reviewed by a native speaker).

Needs LocalExile/LocalExile#948

Against a Gateway that does not list any of this, nothing changes: a cluster jewel has no enchantments in the crafter. A server with that pull request still gives such a jewel a theme, a passive count and its sockets when it is given, so the report itself is fixed by the server change alone; this adds the choice.

Checks

  • dotnet build Waypoint.slnx and Waypoint.Linux.slnx: 0 warnings
  • dotnet test Waypoint.slnx: 465 passed (one new, in CrafterCatalogTests)
  • tests/Launcher.Checks: 208 passed
  • tools/missing-translations.py: clean
  • Merges cleanly with main 4d8b8f1 (which changed docs/launcher.md nearby; both edits are in the merged text)
  • In the running app, against a release folder with a server built from feat(cluster-jewels): real themes on cluster jewels, and the crafter lists them LocalExile#948 and the 3.29.3d client in game:
    • Large Cluster Jewel started with Axe and Sword Damage, 8 passives and 2 sockets.
    • Picked Fire Damage, added Prismatic Heart from Prefixes, Give item: "Added Dragon Shard Large Cluster Jewel"; saved with theme value 13, 8 passives, the two-socket enchantment and the notable, as /cluster_jewel makes from the same inputs.
    • The socket list switched the jewel to one socket; a Small Cluster Jewel showed "Adds 2 Passive Skills" and its life theme.

Not checked: the import window in the running app (a unit test covers it), the adapter's tests/*.Tests.ps1 (nothing under scripts/ changed), Linux. Give item was run with the Waypoint Tools mod 1.0.2, whose loaded.json had to be put right by hand because Waypoint was started while the game ran (fixed in the mod since, LocalExile/LocalExile-Mods#39).

Also here: a test that failed by run order

Second commit, a87128d, tests only. ClientExtractionFailureTests.Other_extraction_failures_keep_their_own_error asked for exactly two 3.29.3c executables. ClientLocator keeps every build a LocalExile's list named for the whole test run, and a test in ClientUpdateTests names a made-up 3.29.3c hash, so the count was three whenever that test ran first. It failed this pull request's first build run and main's own; the two classes run alone failed six runs of six.

The test now checks Waypoint's own rows by their hashes (they come first). Nothing under src/ changed. After the change the two classes passed six runs of six and the whole suite three of three (465 each).

Reported: a cluster jewel made with the crafter is created, but adds no
passive when socketed, while one from /cluster_jewel works.

A cluster jewel is its enchantments, which no implicit list or pool
names, so the crafter had nothing to put on the item. LocalExile's
crafter catalog now lists them (its Agents/docs/dev/item-crafter.md,
"Cluster jewels"), and this is Waypoint following it:

- Choosing a cluster jewel base puts its enchantments on the item:
  "Adds N Passive Skills", which also holds the theme (what the small
  passives grant), and its jewel sockets. They are sent to /item_give in
  the enchant list.
- The theme is a list of the size's themes by name: the game's own value
  is a number, which it words itself, so Waypoint words it from the
  theme's stats ("Added Small Passive Skills grant: 12% increased Fire
  Damage"). The passive count has a slider with the base's range, and the
  socket enchantment a list of its socket counts.
- The theme decides the jewel's affixes: Prefixes and Suffixes list that
  theme's notables with the modifiers every theme has.
- An enchantment's card says "Enchantment", the preview lists
  enchantments first, and Add modifiers has a kind "Enchantments" on a
  base that has some, to put one back.
- An imported cluster jewel takes its passive count, socket count and
  theme from its enchantment lines, and its notables are found among
  that theme's affixes.

Against a Gateway that does not list any of this, nothing changes.
Other_extraction_failures_keep_their_own_error asked for exactly two 3.29.3c executables. ClientLocator keeps every build a LocalExile's list named for the whole run, and a test in ClientUpdateTests names a made-up 3.29.3c hash, so the count was three whenever that test ran first: the two classes alone failed six runs of six.

The test now checks Waypoint's own rows, which come first, by their hashes.
@evrydayjunglist
evrydayjunglist merged commit cc8d446 into main Nov 13, 2026
2 checks passed
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

1 participant