Conversation
mcagnion
marked this pull request as draft
May 10, 2026 07:07
mcagnion
changed the title
mcagnion
marked this pull request as ready for review
May 12, 2026 21:40
mcagnion
force-pushed
the
feature/weighted-score
branch
from
July 26, 2026 19:31
359f900 to
0c331b0
Compare
mcagnion
marked this pull request as draft
July 26, 2026 19:44
mcagnion
force-pushed
the
feature/weighted-score
branch
from
July 28, 2026 09:19
7a269ae to
ddeca82
Compare
mcagnion
force-pushed
the
feature/weighted-score
branch
3 times, most recently
from
August 6, 2026 10:43
c48337f to
4042a77
Compare
mcagnion
force-pushed
the
feature/weighted-score
branch
from
August 15, 2026 15:55
074c6f6 to
c7fe989
Compare
mcagnion
marked this pull request as ready for review
August 21, 2026 09:07
mcagnion
force-pushed
the
feature/weighted-score
branch
from
August 26, 2026 18:17
70a1a53 to
d68cdf3
Compare
Use one set of weights and one ratio calculation for passive-tree and item rankings. Keep weight editing in Trade Query and expose the score through the shared power-stat list. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Cover default and custom weights, ratio edge cases, Trade Query scoring, and passive-tree integration. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Propagate Weighted Score through the Anoint sort list and calculate each notable instead of leaving every score at zero. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Weighted Score is calculated rather than stored in a calc output field. Expose it through the shared power-stat callback so other consumers can request the score without duplicating the calculation. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Loading WeightedScore inside getValue created a new module table on every call. Load it once at module scope for tight loops such as Radius Jewel Finder calculations. Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
Timeless Jewel fallback weights previously read a nonexistent output field and collapsed every node to the same result. Use the Weighted Score callback and guard a zero baseline. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Fallback dropdown entries copied only the stat, transform, and label, dropping the callback needed by Weighted Score. Carry getValue into each entry. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>
Use Full DPS when the active weights require it and reuse the cached baseline for Weighted Score calculations.
Use one dropdown action across Tree, Item DB, and Anoint selectors. Keeping the action inside the menu leaves the League filter available in the narrow Item DB pane.
Restore the shared accessor, transforms, and Full DPS fallback expected by Trade Query.
Avoid generating a Minion Weighted Score entry and hide the score from modifier menus that can only read stored output fields.
Keep negative scores and request Full DPS only when the active weights need it.
Calculate candidates with the matching Full DPS context and retain the baseline output for Timeless Jewel fallback scoring.
Evaluate contextual metrics against the displayed item baseline so Weighted Score and Full DPS rank candidates correctly.
Forward the callback when Reset reopens the editor so the next Save invalidates caller caches.
Evaluate crafted prefix and suffix candidates through the contextual stat API so Weighted Score uses its baseline and Full DPS requirements instead of tying every modifier at zero.
Keep crafted affix, Compare Power, and Timeless fallback scoring aligned with saved weights.
Remove redundant type and action flags, then route existing consumers through the shared power-stat accessors while preserving their score contracts.
Name the weighted ratio accumulator by its actual result and document why the synthetic power-stat value shares the trade-query scale across consumers.
Keep the Trade Query lookup in the shared WeightedScore module while preserving each consumer's local invalidation behavior.
Preserve all 39 weighted-score scenarios while grouping related cases into 19 labelled contracts and removing a redundant Power Builder run.
Keep the generic allocated-node filter and its regression coverage on the dedicated Power Report bugfix branch.
Align stat, output, and selection names with their actual data shapes, and document the shared editor callback and score scaling invariants.
Rank contextual Anoint and crafted-affix candidates against the displayed item while preserving scalar Anoint deltas against the no-anoint baseline.
Share local fixtures and bound the Power Builder smoke to reduce review cost. Replace the Timeless editor-only check with real Weighted Score fallback generation.
Keep the allocated-node filter and its tests on the dedicated Power Report branch.
Use divergent direct and fallback ratios so the contract fails if FullDPS branch selection regresses.
Add the current-build case to the existing exact-score test and remove the weaker non-zero check.
Carry the existing Weighted Score path across the upstream Add Implicit extraction. Use the displayed item as the baseline and rerank the list after weight edits.
mcagnion
force-pushed
the
feature/weighted-score
branch
from
September 27, 2026 12:23
da4e8a3 to
53ece6f
Compare
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
1 participant
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description of the problem being solved:
The Item DB list, Anoint Notable picker, Tree heatmap, and Power Report all rank candidates against a single chosen stat: Life, EHP, Total DPS, Armour, etc. When a build cares about a combination of stats (e.g., 100% Full DPS, 50% EHP, 50% Life Regen, 30% Movement Speed) sorting by a single stat can only approximate that combination.
This PR adds a
Weighted Scoreoption across the shared stat-based ranking surfaces, including the Item DB, Anoint Notable picker, Tree and Compare Power Reports, item modifier and crafted-affix selectors, and Timeless Jewel fallback weighting. The weights reuse the Trade Query'sStat Weight Multipliers, so configured trade weights produce the same ranking everywhere.An
Edit Weights...entry opens the existing Stat Weight Multipliers popup. Saving immediately refreshes surfaces that rank existing candidates; the Timeless Jewel finder preserves its explicitGenerateaction after weights are changed.Steps taken to verify a working solution:
Items tab: Unique DB list sorted by Weighted Score
Inline "Edit Weights..." entry in the Sort dropdown
Anoint picker: Notable list sorted by Weighted Score
Power Report sorted by Weighted Score: