KiCad 10 layout failure: FOOTPRINT.GetFieldByName missing during pcbnew sync

[Milind220]

pcb layout fails on KiCad 10 during footprint import, and the generated board ends up empty.

Environment

• pcb 0.3.58
• KiCad 10.0
• macOS

Repro

Run pcb layout on a simple board with a resistor and LED.

Actual behavior

Netlist generation succeeds, but footprint placement fails during the KiCad sync step:

INFO: Source: 2 footprints, 0 groups, 3 nets
INFO: Changes: +2 -0 footprints
ERROR: Failed to add footprint D1.LED: 'FOOTPRINT' object has no attribute 'GetFieldByName'
ERROR: Failed to add footprint R1.R: 'FOOTPRINT' object has no attribute 'GetFieldByName'

The resulting .kicad_pcb is effectively empty.

Expected behavior
pcb layout should place the footprints and generate a non-empty board.

Note
From the logs, this looks like a KiCad 10 compatibility issue in the Python pcbnew integration layer rather than in netlist generation.
Also, since KiCad 10+ is moving away from the legacy Python/SWIG API, is the intended path for future support still the Python integration, or is there a plan to migrate to the IPC API which would require a running KiCAD instance?

For now, KiCad 9.x seems to be the workaround.

[hexdae]

Hey thanks, we haven't migrated over to kicad 10 yet, but we will do it soon. Will ping this when ready

[Milind220]

Ah alright - thanks for the quick reply!
What's the plan for v10 though? the new IPC API is great (i've got a bindings library in Rust for it) but a little annoying that it naturally requires a running kicad instance.

[LK]

In the short term we'll stick with the current flow of using the Python pcbnew bindings + direct sexpr manipulation, since as you mentioned it works headless. Once the headless IPC API is up and running will definitely consider moving other to that!

We do use the IPC API for some other internal tools, I think KiCad 10 lands some of our upstream fixes to the API so excited to try it out. shoot me an email if you want to compare notes lenny@diode.run