Prefer preserving caller-owned routes during rerip

[ShiboSoftwareDev]

Root cause

The autorouter's stable-initial-assignment subclass restores fanout routes after a whole-graph rerip, but tiny-hypergraph's partial-rip optimizer had no ownership hook. It therefore reopened caller-owned fanout routes during quality optimization even though ordinary routes crossed the same hot regions.

In the SRJ29 sample003 reproduction this changed 18 preloaded fanout sections across six partial-rip rounds. Pipeline 9 later received reconstructed through_obstacle traces and failed exact repair.

Fix

Add one explicit getRouteIdsPreferredForPreservation() hook shared by both rerip modes:

• quality partial rerips retain preferred routes and reopen other routes in the hot region;
• completion selective rerips first search for another blocker owner;
• a preferred route remains eligible when it is the only blocker, so this does not make routes immutable or disable reripping.

Default behavior is unchanged because the base hook returns an empty set. The autorouter's existing StableInitialAssignments subclass opts in for its preloaded route IDs.

Result on the real reproduction

srj29 sample003, Pipeline 10, 1x effort:

	Before	After
Completion	failed	100%
Relaxed DRC	failed	100%
Runtime	69.7s to failure	101.7s solved
Vias	n/a	39

Tests

• bun test tests/outside-in-partial-rip-tiny-hypergraph-solver.test.ts tests/selective-rerip-tiny-hyper-graph-solver.test.ts
• Real PCB reproduction is tracked in the stacked capacity-autorouter reproduction PR.

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
tiny-hypergraph	Ready	Preview	Aug 16, 2026 3:21pm

[tscircuitbot]

Benchmark This PR

Run benchmarks by commenting on this PR:

/benchmark [benchmark.sh args...]
/benchmark --dataset <hg07|srj18> --limit <n> --solver <core|poly> --families <default|default+deep|all> --concurrency <n|auto>

Everything after /benchmark is forwarded directly to ./benchmark.sh.
Every command runs current main and the PR head sequentially on the same Blacksmith machine, then reports their results side by side.

Examples:

• /benchmark -> all committed completed SRJ18 Pipeline7 cases, core solver
• /benchmark --limit 4 -> first four committed Pipeline7 cases
• /benchmark --dataset srj18 --concurrency auto -> srj18 samples with runner CPU-count concurrency
• /benchmark --sample sample003 -> one committed SRJ18 Pipeline7 case
• /benchmark --dataset hg07 --limit 20 --solver poly -> first 20 hg07 samples with the poly solver
• /benchmark --dataset hg07 --limit 40 --families default+deep -> first 40 hg07 samples with deeper candidate families

Any PR whose title contains [BENCHMARK TEST] will automatically run the benchmark workflow on PR updates.