errors when adding polygons with 4 points

[joamatab]

When adding polygons with 4 points we get an error

poly1 = kdb.Polygon(
    [
        [-8, -6],
        [6, 8],
        [7, 17],
        [9, 5],
    ]
)

TypeError: Ambiguous overload variants - multiple method declarations match arguments in Polygon.__init__

@amccaugh @sebastian-goeldi

[sebastian-goeldi]

This works, but it seems the .pyi didn't pick up the new function overload.

>>> import klayout.db as kdb
>>> kdb.Polygon([[0,0],[10,0],[10.5,10]])
(0,0;10,10;10,0)
>>> kdb.DPolygon([[0,0],[10,0],[10.,10]])
(0,0;10,10;10,0)

[sebastian-goeldi]

The issue seems to be if the list has exactly 4 points, then the creation fails

>>> poly1=kf.kdb.DPolygon([(0,0),[-10,0],[10.,10],[10,0]])
Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
TypeError: Ambiguous overload variants - multiple method declarations match arguments in DPolygon.__init__

@klayoutmatthias could it be that klayout tries to cast to a Box first here?

[klayoutmatthias]

Grml ... I knew too much smartness would finally create some trouble ... :(

Basically, the whole scheme why this is works is the inference logic KLayout uses: if an argument expects an object, but is passed an array, KLayout tries to call the object constructor from the arguments of the array. This also works for array elements.

This is way (0,0) can create a DPoint in the DPolygon([DPoint, DPoint, ... ]) constructor. In the same fashion a four-element array can be matched to the arguments of DBox in the DPolygon(DBox) constructor. This is what happens here and creates that ambiguity.

I can fix this by giving precedence of array matching array (I think).

The best workaround is to use "assign_hull" which does not have a Box overload:

poly = kf.kdb.DPolygon()
poly.assign_hull([(0,0),[-10,0],[10.,10],[10,0]])

Matthias

[joamatab]

awesome! that worked, thank you Matthias!

feel free to close this

[klayoutmatthias]

Very good, thanks for the feedback.

I'll keep that ticket open to remind me of this stupid ambiguity. I shall be less permissive on the inference paths.

Matthias