Add options to draw indicator as polygon

[Zabot]

Description

I needed a hexagon unlock indicator for my config, thought it turned out pretty neat and was worth sharing. This adds three new config options.

• --polygon-sides=n enables drawing a regular polygon instead of a circle for the indicator. It defaults to 0, which draws a circle, and can be any integer greater then 2.
• --polygon-offset=d specifies a rotation in degrees to apply to the polygon. At zero offset the first vertex is at (ind_x+radius,ind_y).
• --polygon-highlight={0,1,2} defines the highlight mode while typing. 0 highlights a random edge on each keystroke. 1 highlights sequential edges walking clockwise around the indicator. 2 walks counter clockwise. The edge to highlight is based on the buffer position, so backspacing characters reverses the direction of rotation.

The value of --radius is the center to vertex distance of the polygon. --line-*, --inside-*, and --ring-* are all respected. Separators are not drawn.

Screenshots/screencaps

Release notes

Notes: Added config options to draw indicator as a polygon.

[Raymo111]

This is SO COOL. I'll review and merge soon!

[Raymo111]

Great work! This was a static review, just a couple of code points to fix. Once they're updated I'll do a full test and then approve+merge!

[Raymo111]

Good changes, couple more. Please add more commits instead of amending so it's easier for me to review, thanks!

[Raymo111]

Missed this

[Raymo111]

Should document the required arg (number of sides, in addition to current info here)

[Raymo111]

Can't be 0 either

[Raymo111]

Use EXIT_FAILURE

[Raymo111]

Use EXIT_FAILURE

[Raymo111]

Review — Request changes

Clean, self-contained feature and the screencaps look great. A few real defects plus the merge conflict to clear first. Most severe first:

Merge blocker — rebase needed. GitHub reports this as CONFLICTING / DIRTY; the branch is stale against master and must be rebased before it can merge.

1. [Medium] --polygon-sides=0 is fatally rejected despite being the documented "draw a circle" value. The guard is if (polygon_sides < 3) errx(...) with message "must be greater then 2 or 0", but 0 < 3 triggers the fatal errx, so explicitly passing --polygon-sides=0 kills the process. Should be if (polygon_sides != 0 && polygon_sides < 3).

2. [Medium] bash completion advertises a flag that doesn't exist. i3lock-bash adds --polygon-offset, but the actual getopt long option (and the i3lock-zsh / i3lock.1 docs) is --polygon-rotation. The completion suggests a non-existent option and omits the real one. Rename to --polygon-rotation. (The PR description uses the old name too.)

3. [Low] draw_polygon() truncates vertex coordinates to int — int x = radius * cos(theta) — introducing up to ~1px asymmetry per vertex, worst at small --radius. Cairo takes doubles everywhere else; use double.

4. [Low, visual] Polygon ring isn't path-closed. It returns to the first vertex with an explicit cairo_line_to rather than cairo_close_path; fill auto-closes fine, but the stroked ring renders start/end as two line caps instead of a mitered join, leaving a small notch at vertex 0. Use cairo_close_path.

Nits: inconsistent error handling — case 404 uses errx(EXIT_FAILURE, ...), 405/406 use errx(1, ...), and out-of-range --polygon-highlight warns-and-continues while out-of-range --polygon-sides is fatal; pick one convention. Typo "greater then 2" (also in the man page) → "than". The counterclockwise -input_position % polygon_sides yields a negative index but is harmless (fed through periodic cos/sin) — a comment would help.

Recommendation: Request changes — rebase to clear the conflict, then fix the --polygon-sides=0 rejection and the --polygon-offset→--polygon-rotation completion mismatch (both real defects); the double-precision and close_path items are quick polish. Worth landing after that.