I'm seeing enough different interpretations to think things aren't quite specified -- but I'm not sure if it matters. (1) Is any of this something that should affect computation, or is it really just a question of how to interpret possibly ambiguous documentation? (2) Are any of these troubling cases something that a person should actually write for a normal situation? Or are they just arguments about which abbreviations are acceptable? Or about how automatically-generated (inferred) type descriptions should be written? (3) Are the slice-expansion questions all assumed to be indexing an n-dimensional array, as opposed to [start, stop, step]? Is that explicit in the PEP, and just not in the extracts here? (4) Expanding multiple * shouldn't be ambiguous; the problem is figuring out what to condense into which if two are adjacent. So s1, s2 =[a,b], (1,2,3) [*s1, *s2] should turn into [a, b, 1, 2, 3] The problem is that [*s3, *s4] = (a, b, 1, 2, 3) is ambiguous ... and I didn't really get that distinction from Petr's question or the answers. I can't tell whether I've missed something crucial, or others are arguing over angels on a pinhead ... so whatever the PEP ends up deciding, it should be explicit. (I *think* the earlier parts of this thread are consistent with this, and discussing whether to say explicitly that certain formats are forbidden (but maybe not enforced by the grammar), meaningless, or valid but currently meaningless outside of typing.) -jJ