[feature request] anchors for line's coordinates

[tapyu]

In

line(
    (0,1),
    (0,2),
    (0,3),
    mark: (end: ">"),
    name: "line"
  )

it would be interesting to have anchors to the coordinates that makes the line up. AFAIK CeTZ doesn't provide such anchor.

I propose the syntax "line.i" to access the ith coordinate. For instance, in "line", "line.1" would access (0,1), "line.2" would access (0,2), and so on.

PS: Path anchor (e.g., "line.30%") is not convenient when the line is very complex, so it is not a reasonable alternative.

[johannes-wolf]

Since <element>.<int> is already used for absolute length along a path, we could add anchors with a prefix like pt-: line.pt-1. What do you think?

[tapyu]

I see...

we could add anchors with a prefix like pt-: line.pt-1.

Are you proposing to add such prefix for absolute length anchors and leave the syntax "line.i" for my proposed anchor, right? If so, I think it is a good idea. However, I would suggest a small change: instead of "line.pt-1", I would adopt "line.1pt" for two reasons:

1. 1pt matches with the syntax used for length types variables.
2. No extra fancy - character is required.

Moreover, I think this "new" (AFAIU) anchor type deserves a subsection in sec 2.2 of CeTZ. It is distinct enough from named, border, and path anchor type. What do you think? I think so because the way we access it not based on a path, compass or angle, or a mere name. Rather, it is the ith coordinate used to construct the object (in this case, a line). I think the terms "coordinate anchor" or "order anchor" are good candidates to refer to it.

[fenjalien]

Are you proposing to add such prefix for absolute length anchors and leave the syntax "line.i" for my proposed anchor, right?

No, when using shorthand anchors, anchors that start with a digit are reserved for path anchors if the element supports them. What @johannes-wolf has proposed is that for each coordinate given to the line function a pt-i named anchor will be created, where i is the index of the coordinate. This is similar to how a bezier's control points are given anchors.

[tapyu]

Ah :( Well, it is up to you. I think pt-i too wordy, to be honest, but okay.

[johannes-wolf]

Something like pt-0, pt-1 is too wordy?

[tapyu]

Yes. Four characters when only one is actually useful.

This is obviously a matter of opinion, though. Fell free to disregard my point of view. Although it is verbose, pt-i is good, don't get me wrong. More importantly, this is the syntax used in CeTZ, like catmull, so I think it is the way-to-go for the sake of consistency.