If a path between a schematics capture program and pcb-rnd is to be implemented,
the following information needs to be passed:
<p>
<ul>
	<li> schematics -&gt; pcb (forward annotation)
		Any file format the schematics tool prefers and delivers at least this
		set of information:
		<ul>
			<li> Each component/part needs an unique identifier, a "refdes", e.g. R1, U4.
			<li> Each pin/pad needs an unique identifier within it's component/part, so at the end "U4-3" uniquely identifies pin 3 of component U4.
			<li> Each component/part needs a footprint (a free form text). Optionally arbitrary user specified key=value pairs per part can solve this and a lot more.
			<li> A list of named networks. Optionally with user specified key=value pairs per network (less important here).
			<li> A list of component-pin pairs per network.
			<li> The netlist can be flat or hierarhic.
		</ul>

	<li> pcb -&gt; schematics (back annotation)
		pcb-rnd emits a netlist patch described
		<a href="../backann"> here </a>, but it
		could emit any other format the schematics tool needs. The schematics tool
		would need to understand the following concepts while reading a back
		annotation file:
		<ul>
			<li> Unique part refdes and pin number, as above
			<li> Named networks, as above
			<li> "this is how the network looked like when we started" (net name, list of pins)
			<li> "the user removed this pin from that network" operation
			<li> "the user added this pin from that network" operation
			<li> optional: "the user changed this key=value pair of this part"
		</ul>
</ul>