Ports and cross-sections

A marker placed on the start or end of a shape drawn with a cross-section takes the cross-section. The port also gets the width of the main section and a type ("waveguide" by default), so there is nothing to write but the position:

marker {
    input  { at wg.start }   # xsection = strip, width = 500nm, type = "waveguide"
    output { at wg.end }
}

The same goes for markers placed on ports of instances: an exported port keeps the cross-section of the port it sits on.

Cross-sections become part of the connection checks. Two connected ports with different cross-sections, such as a rib waveguide butted against a strip waveguide, are an error, just like ports of different widths. Submit reports it, and so do laylight netlist and laylight lvs.