Triple dot operator removes 0s

[a-maier]

Running

on names;
s <x01a1>,...,<x01a3>;
.end

shows

 Symbols
   x1a1 x1a2 x1a3

So the 0s seem to get removed. As one would expect, when changing 01 to 0 or leaving out the angle brackets the zeros are preserved.

[jodavies]

We discussed this at the workshop, and thought that it could be useful to assume leading zeros to imply width-padding hints, for e.g.:

• <x01a1>,...,<x01a3> produces x01a1,x01a2,x01a3
• <x01>,...,<x11> produces x01,x02, ... ,x10,x11
• <x01>,...,<x100> should produce x01,x02, ..., x10, ..., x100 ?

Any thoughts, @a-maier ?

In principle, this may break existing user code.

[a-maier]

I have stumbled upon this when trying to mangle variable names. For that use case, only your first bullet point should be relevant.

I am wondering how the width-padding idea would work for something like <x01>,...,<x011> and what your cases 2 and 3 should expand to when leaving out the angle brackets. It seems that x01,...,x03 currently expands to x01,x2,x3, whereas <x01>,...,<x03> expands to x1,x2,x3. I don't find that entirely intuitive.

In principle, this may break existing user code.

That's true. Of course, such code would be relying on undocumented behaviour.

[jodavies]

I would say that <x01>,...,<x011> should produce x01,x02,...,x011 since x0 is a common prefix.

<x01>,...,<x03> would then be the same as your first example: x01,x02,x03.

x01,...,x03 expanding to x01,x2,x3 feels like a bug even without changing anything.