In the "FIXME Roulette: Creating Path2D objects from SVG path strings" video by @awesomekling, at 32:00 I noticed that for the CanvasRenderingContext2DPrototype::fill generated function (and more generally for functions with effective overload sets, that there are two cases that the code handles based on the argument count:

1. There are more than one effective overload. -- In this case, the code needs to build the overload set in order to work out which function to use so the generated code to calculate the effective overload set is necessary.

2. There is only one effective overload, (E.g. the 0-argument and 2-argument variants of fill) -- In this case, it should be possible to avoid generating the effective overloads and call the correct version of the function directly (after any error checking and type conversion of the arguments).

Therefore, the overload resolution logic and overhead should then only apply when the IDL generates 2 or more possible overloads for the function name and argument count.