Improve huge selection handling

[saskliutas]

Closes #1398

Updated usePropertyDataProviderWithUnifiedSelection to avoid loading instance keys when selection is huge and exceeds limit.
Updated Selectables.load to avoid freezing main thread when loading huge selection.

[changeset-bot]

🦋 Changeset detected

Latest commit: 767980a

The changes in this PR will be included in the next version bump.

This PR includes changesets to release 2 packages

Name	Type
@itwin/presentation-components	Patch
@itwin/unified-selection	Patch

Not sure what this means? Click here to learn what changesets are.

Click here if you're a maintainer who wants to add another changeset to this PR

[github-actions]

Unified selection benchmark

Benchmark suite	Current: 767980a	Previous: 701b08c	Deviation	Status
compute selection for 50k elements	386 ms	401 ms	-3.74%	〰️
compute selection for 50k elements (P95 of main thread blocks)	31 ms	32 ms	-3.13%	〰️
compute parent selection for 50k elements	345 ms	357 ms	-3.36%	〰️
compute parent selection for 50k elements (P95 of main thread blocks)	31 ms	35 ms	-11.43%	〰️
compute top ancestor selection for 50k elements	559 ms	561 ms	-0.36%	〰️
compute top ancestor selection for 50k elements (P95 of main thread blocks)	0 ms	0 ms	0%	🟰
compute category selection for 50k elements	101 ms	104 ms	-2.88%	〰️
compute category selection for 50k elements (P95 of main thread blocks)	0 ms	0 ms	0%	🟰
compute model selection for 50k elements	85 ms	84 ms	1.19%	〰️
compute model selection for 50k elements (P95 of main thread blocks)	0 ms	0 ms	0%	🟰
compute functional selection for 50k 3D elements	427 ms	443 ms	-3.61%	〰️
compute functional selection for 50k 3D elements (P95 of main thread blocks)	34 ms	31 ms	9.68%	〰️
compute parent functional selection for 50k 3D elements	443 ms	464 ms	-4.53%	〰️
compute parent functional selection for 50k 3D elements (P95 of main thread blocks)	35 ms	35 ms	0%	🟰
compute top ancestor functional selection for 50k 3D elements	1173 ms	1191 ms	-1.51%	〰️
compute top ancestor functional selection for 50k 3D elements (P95 of main thread blocks)	0 ms	0 ms	0%	🟰
compute functional selection for 50k 2D elements	2988 ms	3037 ms	-1.61%	〰️
compute functional selection for 50k 2D elements (P95 of main thread blocks)	0 ms	0 ms	0%	🟰
compute parent functional selection for 50k 2D elements	2983 ms	2933 ms	1.70%	〰️
compute parent functional selection for 50k 2D elements (P95 of main thread blocks)	0 ms	0 ms	0%	🟰
compute top ancestor functional selection for 50k 2D elements	3047 ms	3072 ms	-0.81%	〰️
compute top ancestor functional selection for 50k 2D elements (P95 of main thread blocks)	0 ms	0 ms	0%	🟰
hilite 50k elements	1266 ms	1192 ms	6.21%	〰️
hilite 50k elements (P95 of main thread blocks)	55 ms	52 ms	5.77%	〰️
hilite 50k group elements	240 ms	236 ms	1.69%	〰️
hilite 50k group elements (P95 of main thread blocks)	31 ms	31 ms	0%	🟰
hilite 1k subjects	46999 ms	48341 ms	-2.78%	〰️
hilite 1k subjects (P95 of main thread blocks)	26 ms	28 ms	-7.14%	〰️
hilite 50k subcategories	310 ms	315 ms	-1.59%	〰️
hilite 50k subcategories (P95 of main thread blocks)	39 ms	39 ms	0%	🟰
hilite 50k functional 3D elements	31446 ms	29150 ms	7.88%	〰️
hilite 50k functional 3D elements (P95 of main thread blocks)	41 ms	41 ms	0%	🟰
hilite 50k functional 2D elements	6678 ms	6329 ms	5.51%	〰️
hilite 50k functional 2D elements (P95 of main thread blocks)	43 ms	41 ms	4.88%	〰️

This comment was automatically generated by workflow using github-action-benchmark.

[Copilot]

Pull request overview

This PR addresses performance issues when unified selection contains a very large number of elements by avoiding unnecessary instance-key materialization and by yielding during large instance-key loads to reduce main-thread blocking.

Changes:

• Updated usePropertyDataProviderWithUnifiedSelection to short-circuit selection handling when selection size exceeds the configured limit (avoid creating/loading a full KeySet).
• Updated Selectables.load to periodically yield while iterating large instance-key sets to reduce UI freezing.
• Added changesets for @itwin/unified-selection and @itwin/presentation-components.

Reviewed changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 2 comments.

File	Description
packages/unified-selection/src/unified-selection/Selectable.ts	Adds main-thread yielding while iterating large instanceKeys selections during Selectables.load.
packages/components/src/presentation-components/propertygrid/UseUnifiedSelection.tsx	Avoids materializing KeySet from storage when selection is over limit; propagates either KeySet or size.
.changeset/hip-spoons-tease.md	Changeset for @itwin/unified-selection performance improvement.
.changeset/common-frogs-dig.md	Changeset for @itwin/presentation-components selection-limit optimization.