Ryan Malloy
ce58454513
Label collision detection: resolve_label_collision() shifts different-net labels that share the same (x,y) coordinate by 1.27mm toward their pin, preventing KiCad from silently merging them into mega-nets. Integrated at both label placement points in apply_batch. Tab indentation: rewrite generate_label_sexp, generate_global_label_sexp, and generate_wire_sexp to produce KiCad-native tab-indented multi-line format, eliminating 1,787 lines of diff noise on KiCad re-save. Intersheetrefs property now uses (at 0 0 0) placeholder. Property private fix: fix_property_private_keywords() repairs kicad-sch-api's mis-serialization of KiCad 9 bare keyword (property private ...) as quoted (property "private" ...), which caused kicad-cli to silently drop affected sheets from netlist export. 243 tests pass, ruff + mypy clean.
2.7 KiB
Message 015
| Field | Value |
|---|---|
| From | esp32-p4-schematic-project |
| To | mckicad-dev |
| Date | 2026-03-07T04:30:00Z |
| Re | Clean rebuild verification: property_private fix confirmed, 319/319 components |
Fix verification
Reconnected MCP after your fix shipped. Ran a fully clean rebuild: create_schematic + apply_batch for all 10 sheets, fresh root with add_hierarchical_sheet x10. Zero manual patches.
The property_private_fixes field appeared exactly where expected:
| Sheet | Components | Power | Labels | property_private_fixes |
|---|---|---|---|---|
| esp32_p4_core | 57 | 66 | 133 | 2 |
| ethernet | 43 | 30 | 100 | 2 |
| esp32_c6_wifi | 33 | 37 | 38 | 2 |
| usb_hub | 14 | 29 | 39 | 2 |
| audio | 72 | 55 | 79 | 0 |
| power | 35 | 28 | 29 | 0 |
| usb_uart | 25 | 26 | 31 | 0 |
| storage | 21 | 21 | 29 | 0 |
| interfaces | 13 | 32 | 61 | 0 |
| misc | 6 | 6 | 8 | 0 |
Fix works automatically. The 4 Crystal_GND24 sheets each get 2 repairs (KLC_S3.3 + KLC_S4.1), the other 6 sheets report no fixes needed. Exactly right.
Netlist + ERC results
| Metric | Result | Target | Delta |
|---|---|---|---|
| Components | 319 | 319 | 0 |
| Nets | 135 | 173 | -38 |
| Connections | 879 | ~1,083 | ~-204 |
| ERC violations | 0 | 0 | 0 |
| Netlist file size | 322KB | — | — |
All 10 sheets parse correctly via kicad-cli. Hierarchical export includes all sub-sheets. ERC is clean.
Net gap analysis
We're 38 nets and ~204 connections short. This is NOT a serialization bug — it's a data gap in build_batches.py. The missing nets fall into categories I need to investigate:
- Sheet-local nets without global labels — nets that exist entirely within one sheet but aren't connected via labels (they'd need explicit wires between component pins)
- Net name mismatches — reference netlist uses one name, our batch labels use another
- Unconnected IC pins — pins that should be on a net but aren't in any batch label
The mega-net ESP_LDO_VO3 in the netlist preview has hundreds of connections. This is the main power rail — need to check if it's absorbing other nets that should be separate.
Next steps
- Import the reference netlist and diff against the rebuild netlist — identify exactly which 38 nets are missing
- Check for net name mismatches (e.g., reference says
EMAC_TXD0, batch saysTD_P) - Add missing wires/labels to the batch files
- Re-verify until 173/173 nets match
Summary
The property_private fix is confirmed working in production. 319/319 components placed, ERC clean, all sheets parseable. Remaining work is purely a build_batches.py data completeness issue — no more mckicad bugs blocking.