Mobile placeholder proof failures should be falsified with direct desktop keypress before editing slate-react

What happened

The mobile placeholder-input proof lanes looked like a runtime bug:

• iOS had reported polluted placeholder text after typing
• Android had reported the same kind of bad readback
• the first instinct was to keep digging inside slate-react

That was the wrong first move.

A direct Chromium proof on the same placeholder?debug=1 surface showed the runtime seam was already clean:

• focus #placeholder-ime
• press a real key through Playwright
• read #placeholder-ime-debug-json

The debug output came back as:

• blockTexts: "s"
• placeholderShape: null
• slateSelection: "0.0:1|0.0:1"

So the editor was not swallowing placeholder text on the normal browser path.

At the same time, the failing slate-react unit test turned out to have a bad fixture assumption:

• the empty placeholder DOM shape does not contain a [data-slate-string="true"] node
• it contains zero-width markup inside the text host

Fixing the test to mutate the text host directly made the package test green again without inventing another editor patch.

Why this matters

When mobile automation reports placeholder pollution, there are two very different possibilities:

1. the editor DOM-commit seam is broken
2. the transport input path is lying or unstable

If you skip the direct desktop falsifier, you can waste time “fixing” slate-react when the problem is really the transport or the test fixture.

Reusable rule

For placeholder/input debugging on these proof surfaces:

• first run a direct desktop keypress on the exact same debug surface
• if desktop returns clean blockTexts, placeholderShape, and selection, stop blaming slate-react
• then inspect the transport semantics or the proof harness instead
• for empty placeholder unit tests, mutate the [data-slate-node="text"] host directly rather than assuming a [data-slate-string="true"] node exists