[css-easing-2] Intended possibility of discontinuity of `linear()`?

[yarusome]

2.1.2 Parsing ensures that points in <linear-stop-list> have (not necessarily strictly) increasing inputs. And in 2.1.4 Output of a linear easing function (irrelevant steps omitted):

2. Let pointAIndex be index of the last item in points with an input less than or equal to inputProgress, or 0 if there is no match.

[…]

4. Let pointA be points[pointAIndex].

5. Let pointB be points[pointAIndex + 1].

[…]

Following these steps, linear(0, 0.6 50%, 0.4 50%, 1) produces a discontinuous piecewise linear function:

Is this really intended by the spec? If so, I'll add an example in mdn/content#27167.

[tabatkins]

We can't stop people from writing linear functions with multiple outputs for the same input, so yes, it's intended by the spec. I'm not certain the old evaluation text actually handled discontinuities properly, but the new text I've written definitely does, and matches your diagram.

(Additionally, if your function starts with a discontinuity, or ends with one, then inputs before or after the first/last point give a constant output equal to the first/last point.)