fix: oscillating net pair suspends only that pair, not the whole pass - #1110
Open
tbontb-iaq wants to merge 3 commits into
Open
fix: oscillating net pair suspends only that pair, not the whole pass#1110tbontb-iaq wants to merge 3 commits into
tbontb-iaq wants to merge 3 commits into
Conversation
segmentIntersectsRect declared any segment that is neither exactly vertical nor exactly horizontal collision-free. Real pin coordinates are not always exactly on the routing grid (e.g. x = 5.3999378 next to x = 5.4), so nearly axis-aligned segments sailed through chip bodies and were accepted by UnroutedTraceRecoverySolver as collision-free. Non-axis-aligned segments now run a slab (Liang-Barsky) clip against the rect. Axis-aligned fast paths keep their historical semantics: a real interior overlap counts, a segment that merely touches an edge or corner does not (tMax - tMin > eps for the slanted path). Fixes tscircuit#1099
TraceOverlapShiftSolver treated its A -> B -> A oscillation guard as a reason to stop the entire solve. The first oscillating net pair aborted the pass, leaving every other independent cross-net overlap unresolved. Suspending only the oscillating pair lets the remaining pairs be corrected. The minimized repro from issue tscircuit#1102 now resolves the RTS/DTR overlap (2 surviving cross-net overlaps -> 1, the genuinely oscillating GND/VBUS pair); board-1273 goes from 6 unresolved pairs down to 1 in 37 iterations (previously 6 pairs after 27). Fixes tscircuit#1102
|
The latest updates on your projects. Learn more about Vercel for GitHub.
|
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Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be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.
Fixes #1102
/attempt #1102Problem
TraceOverlapShiftSolver._step()treated its A -> B -> A oscillation guard as a reason to stop the entire solve (this.solved = true). The first oscillating net pair aborted the whole pass, leaving every other independent cross-net overlap silently unresolved.Fix
suspendedNetPairKeysand only that pair is suspended.findNextOverlapIssueskips suspended pairs, so independent pairs keep being corrected until no correctable overlap remains.Verification (minimized repro from the issue, 5 chips / 14 pins)
Before: solver stops at iteration 9 with 2 cross-net pairs still collinearly overlapping (
GND/VBUSoscillating +RTS/DTRindependent).After: 1 remains — only the genuinely oscillating
GND/VBUSpair, exactly the expected behavior.RTS/DTRis separated.On the full
board-1273fixture: unresolved cross-net pairs go from 6 down to 1 (the single oscillating pair), solved in 37 iterations; no traces added or dropped. The repro test's iteration-count assertion was updated accordingly and its SVG snapshot regenerated (routing visibly changed only in the regions where the previously-aborted pairs now get corrected).Full suite: 353 pass / 0 fail (357 tests, 4 skipped).