Lorenzo Gaifas

> You can think of this as the bounds of the data in view :-) But that's not relaly true, since the data in view for the two layers above...

Why would you need info about the layer? You can simply return `(0, 0)` (translated as appropriate) or something like that for any dimension that's not in the data.

No, it's determined by the argument of `Layer._display_bounding_box`, which is `dims_displayed`. This does not depends on the state of the layer, it simply asks "if you were to be displayed...

I suggest we leave this open for some time so it's easier to find :)

@haesleinhuepf It would take something like the following to work with any n-d dataset: ```py displayed_extrema = np.array(viewer.dims.range)[np.array(viewer.dims.displayed)] displayed_ranges = displayed_extents[:, 1] - displayed_extents[:, 0] points_layer.size = displayed_ranges * 0.01...

Sorry for the review-request spam, but I think a change like this needs some consensus because it might be considered a bit magic/unexpected that point sizes stop changing at some...

> I like the minimum size, but dislike the maximum size. Personally I would default maximum to None. If not, then to the maximum you are comfortable with — maybe...

@alisterburt I added this to the `0.4.17` milestone. We just need to get some sort of consensus on the sensible defaults; other than that, this is a net improvement over...

I agree that `(2, 10000)` (arbitrary number higher than the highest hardware limits, since `None` is not supported) is a good default. As for settings, we can definitely do that,...

@andy-sweet I think this can be merged for 0.4.17, everything's passing, we're just missing an approval :)