Support standard IPC-2581C NetShort declarations in DFM

[LK]

Summary

• Preserve standard IPC-2581C Set/NetShort metadata, including transformed locations and source step/layout occurrence.
• Require explicit copper net attribution. A NetShort identifies one physical contact; it does not give a netless graphic ownership or exempt a whole component or net pair.
• Validate declarations against final copper on the declared layer. The supported case is a two-net contact where one connected region is contained in the other, without significance filtering or geometric uncertainty beyond coincidence precision.
• Preserve other contacts, nearby gaps, third-net clearance, and unrelated drill checks. Ambiguous functional copper still fails closed with component, Set, and feature diagnostics.

Specification and limitations

The supplied IPC-2581C June 2020 draft, §8.2.3.10.8, specifies a single referenced location for each intentional short; another contact between the same nets requires another declaration. This replaces the earlier, incorrect whole-graphic NetTie exemption.

Partial overlaps, edge/point-only contacts, uncertain contact topology, multi-layer declarations, and declarations with more than two NetRefs remain unsupported and fail closed. In particular, the isolated real-geometry antenna regression completes, but a routed feed extending outside the bridge can still require unsupported partial-contact handling. This PR does not establish that the complete routed board passes DFM.

Companion exporter work: https://gitlab.com/diode-inc/kicad/-/merge_requests/35. The corrected exporter changes are now published to that fork MR under explicit user approval. They use native net_tie_pad_groups, explicit Set/net, and contact-specific NetShort locations. No bespoke properties or board changes are used.

Validation

• cargo nextest run -p pcb-ipc2581-tools -p pcb-ir -p ipc2581: 627 passed; no snapshots changed.
• Scoped Clippy with --all-targets -- -D warnings passed.
• Scoped doctests: 2 passed, 1 existing ignored test.
• Real antenna geometry retained in fixtures/antenna.xml; regressions exercise standard and jlcpcb-1oz profiles, separate declarations for separate contacts, nearby same-pair gaps, third-net shorts, unrelated drills, missing ownership, metadata-only Sets, repeated layout occurrences, and partial/uncertain geometry including a 10 nm sliver.
• Real findings remain, including the ground pad's 0.1 mm annular ring.

---

Note

Medium Risk
Changes copper clearance exemptions and conductor ownership in DFM—a correctness-sensitive fab path—but behavior is heavily regression-tested and invalid declarations fail closed.

Overview
Adds initial DFM support for IPC-2581C NetShort (intentional net ties): the parser and import path now read Set/NetShort as electrical intent (not painted copper), keep declarations on step definitions even when a layer has no geometry, and scope import diagnostics to the checked step/layer.

Copper clearance is updated so a valid declaration exempts only the specific contact at the declared location—graphic↔pad or pad↔pad on the same component—without merging nets or waiving unrelated shorts, third-net overlaps, or clearance elsewhere. Net-tie copper gets distinct conductor identity (ConductorId with optional tied object); unattributed functional copper still fails closed, with clearer Set/feature/component messaging. Unsupported or ambiguous NetShort metadata marks ownership-dependent rules Incomplete rather than silently passing.

Includes a real antenna IPC fixture and broad regression tests (panel/array occurrences, negative clear copper, location tolerance, KiCad-like pad entry).

^{Reviewed by Cursor Bugbot for commit 310e6b2. Bugbot is set up for automated code reviews on this repo. Configure here.}

---

[cursor]

Stale comment

Left a non-blocking comment; no reviewers were assigned. Bugbot and the Security Agent completed with no findings, but this is a functional DFM change rather than a small fixup, so it still needs human review.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}

[cursor]

Stale comment

Left a non-blocking comment; no reviewers were assigned. Bugbot completed with no findings and the Security Agent was not running, but this is a functional DFM change rather than a small fixup, so it still needs human review.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}

[cursor]

Stale comment

Left a non-blocking comment; no reviewers were assigned. Bugbot completed with no findings and the Security Agent was not running, but this is a functional DFM change rather than a small fixup, so it still needs human review.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}

[devin-ai-integration]

Note

Newer findings are available below. Devin Review posted a newer report on this PR, in addition to the findings presented here.

Devin Review found 1 new potential issue.

[cursor]

Stale comment

Left a non-blocking comment; no reviewers were assigned. Bugbot completed with no findings and the Security Agent was not running, but this is a functional DFM change rather than a small fixup, so it still needs human review.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}

[cursor]

Stale comment

Left a non-blocking comment; no reviewers were assigned. Bugbot completed with no findings and the Security Agent was not running, but this is a functional DFM change rather than a small fixup, so it still needs human review.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}

[cursor]

Left a non-blocking comment; no reviewers were assigned. Bugbot completed with no findings and the Security Agent was not running, but this is a functional DFM change rather than a small fixup, so it still needs human review.

  

_{Sent by Cursor Approval Agent: Pull Request Router and Approver}