From 9d29441a84858780f80ff1bc35ce9c0c5f43b0c9 Mon Sep 17 00:00:00 2001 From: JJ Liebig Date: Sun, 20 Sep 2026 00:21:32 +0200 Subject: test(windows): distinguish clipboard paste origin in input gauntlet (#4394) * test(windows): distinguish clipboard paste origin in input gauntlet refs #4314 * test(windows): correct gauntlet paste-origin evidence refs #4314 --- scripts/windows_input/README.md | 5 +++++ 1 file changed, 5 insertions(+) (limited to 'scripts/windows_input/README.md') diff --git a/scripts/windows_input/README.md b/scripts/windows_input/README.md index 5f8f463f..edebc8fc 100644 --- a/scripts/windows_input/README.md +++ b/scripts/windows_input/README.md @@ -132,6 +132,11 @@ Every run needs a **new** output directory. By default it is the runner does not dismiss unexpected dialogs or rebind Terminal shortcuts. - Image-only Ctrl+V through the remote clipboard bridge, plus a mixed image/text clipboard check that proves Windows Terminal and Herdr preserve the text paste. + The report records `paste_origin`, classified from the client's own mapper trace + (`terminal-paste`, `empty-paste`, `key-event`, or `none`). A staged image only + passes when the origin is a positively observed `empty-paste`: a paste the + terminal issued for clipboard text fails, and a missing or inconclusive trace is + `inconclusive` rather than a qualification (#4314). - A full case pass at an observed 120×30 host size; keyboard/paste sentinels at **80, 119, 120, 121, 132, 160, 240 columns**, at 24 and 50 rows; then return to 80 columns. This exercises narrow→wide→narrow resizing of the actual outer -- cgit v1.2.3