0.2.1: height={nil} works, and the README stops arguing against Leaflet

[nseaSeb]

Two things surfaced by using 0.2.0 in a real application.

height={nil} was documented but illegal — and then still broken

attr :height, :string rejected the nil its own doc recommended. Phoenix emits attribute "height" ... must be a :string, got: nil — a compile warning, so an error in any project building with --warnings-as-errors. Now :any, like :class always was.

Fixing the type was not enough. style={@height && …} renders style="" for a nil value instead of omitting the attribute, and an empty inline style still beats a class in the cascade — so a map sized by class="h-96" or by a flex parent could not be sized at all, which is exactly the case the escape hatch exists for. The style is now a dynamic attribute, genuinely absent when there is no height, and placed after {@rest} so a caller's own style wins.

Three tests: height={nil} emits no style, a caller's style overrides, and the existing height/class case still holds.

The README stopped resting on someone else's library

The opening section was titled "Why not just use Leaflet?" and made a third-party project the yardstick for this one — while answering a question a Phoenix developer does not actually ask. The question in front of them is "why not just write a hook?", and that has the better answer: the first afternoon is short; the ones after it are the reconciliation, the framing, updated(), the coordinate order and the attribution.

Leaflet's name moves to Coming from a Leaflet hook — a migration table rather than a benchmark, which serves the people most likely to arrive here. It names the three things that catch them: markers need a stable :id, height beats your class, stroke opacity goes through rgba().

The two remaining mentions in the JavaScript stay; one of them explains why resolveRetina exists at all.

131 Elixir tests, 48 Node tests.

🤖 Generated with Claude Code