test(auto-dim): pin the minibuffer dim behavior, not just the setting

- The old assertion checked a value the module just set.
- It couldn't fail even if the fork inverted the flag.
- Three behavioral tests now drive real windows and minibuffer.
- One pins the working half, one the flag-is-t boundary.
- The third pins a known fork gap as an expected failure.

adob--rescan-windows ignores the flag, so a completion popup dims every
window. I handed that fix to the fork, which owns the code. When the fork
honors the flag, the pinned test reports an unexpected pass and can go.

Each test asserts the other window is still dimmed. Without that control,
three broken implementations look exactly like a working fix.

1 files changed, 94 insertions, 5 deletions

diff --git a/tests/test-auto-dim-config.el b/tests/test-auto-dim-config.el
index dcab7eff..8b13fbb0 100644
--- a/tests/test-auto-dim-config.el
+++ b/tests/test-auto-dim-config.el
@@ -30,16 +30,105 @@
       (progn
         (should (bound-and-true-p auto-dim-other-buffers-mode))
         (should (null auto-dim-other-buffers-dim-on-focus-out))
-        ;; Entering the minibuffer must not change what is dimmed: a dim window
-        ;; stays dim, a lit one stays lit.  The fork's `adob--update' returns
-        ;; early when this is nil and the selected window is the minibuffer, so
-        ;; nil is what keeps a minibuffer prompt from re-dimming the window the
-        ;; user was just in.
+        ;; Config intent only: this asserts the value the module just set, so it
+        ;; cannot fail even if the fork inverts what the flag MEANS.  The two
+        ;; behavioral tests below are what actually pin the behavior.
         (should (null auto-dim-other-buffers-dim-on-switch-to-minibuffer))
         (should-not (assq 'fringe auto-dim-other-buffers-affected-faces)))
     (when (fboundp 'auto-dim-other-buffers-mode)
       (auto-dim-other-buffers-mode -1))))
 
+(defmacro test-auto-dim--with-two-windows (win-a win-b &rest body)
+  "Bind WIN-A and WIN-B to two live windows with the mode on, then run BODY.
+Restores the window configuration, the buffers, and the mode's PRIOR state.
+Restoring rather than force-disabling matters: `auto-dim-other-buffers-mode' is
+global, and switching it off unconditionally left a later test in this file
+asserting the mode is on with it off.  That only stayed hidden because ERT runs
+tests alphabetically and the asserting test sorts first."
+  (declare (indent 2))
+  `(let ((config (current-window-configuration))
+         (was-on (bound-and-true-p auto-dim-other-buffers-mode)))
+     (unwind-protect
+         (let* ((,win-a (selected-window))
+                (,win-b (split-window)))
+           (set-window-buffer ,win-a (get-buffer-create " *adob-a*"))
+           (set-window-buffer ,win-b (get-buffer-create " *adob-b*"))
+           (select-window ,win-a)
+           (auto-dim-other-buffers-mode 1)
+           ,@body)
+       (when (fboundp 'auto-dim-other-buffers-mode)
+         (auto-dim-other-buffers-mode (if was-on 1 -1)))
+       (set-window-configuration config)
+       (dolist (name '(" *adob-a*" " *adob-b*"))
+         (when (get-buffer name) (kill-buffer name))))))
+
+(ert-deftest test-auto-dim-config-minibuffer-entry-leaves-previous-window-lit ()
+  "Normal: with the flag nil, entering the minibuffer leaves the previous window lit.
+This is the half that works, and the reason the flag is set to nil.
+
+Deliberately asserts the composite behavior rather than naming one function.
+Selecting the minibuffer fires the mode's own hooks, so the observable outcome is
+not attributable to the explicit `adob--update' call alone -- and the observable
+outcome is what the setting promises the user.
+
+The `win-b' assertions are positive controls.  Without them this test passes
+against an implementation where dimming is broken everywhere, which looks
+identical to the implementation being correct."
+  (skip-unless (file-directory-p test-auto-dim--fork))
+  (require 'auto-dim-config)
+  (test-auto-dim--with-two-windows win-a win-b
+    (adob--rescan-windows)
+    (should (null (window-parameter win-a 'adob--dim)))
+    (should (window-parameter win-b 'adob--dim))
+    (select-window (minibuffer-window))
+    (adob--update)
+    (should (null (window-parameter win-a 'adob--dim)))
+    (should (window-parameter win-b 'adob--dim))))
+
+(ert-deftest test-auto-dim-config-minibuffer-entry-dims-when-flag-is-t ()
+  "Boundary: with the flag t, entering the minibuffer DOES dim the previous window.
+This is what gives the nil setting meaning.  Without it the suite never shows the
+flag changing anything, so the config assertion in
+`test-auto-dim-config-applies-settings' has nothing standing behind it."
+  (skip-unless (file-directory-p test-auto-dim--fork))
+  (require 'auto-dim-config)
+  (test-auto-dim--with-two-windows win-a win-b
+    (let ((auto-dim-other-buffers-dim-on-switch-to-minibuffer t))
+      (adob--rescan-windows)
+      (should (null (window-parameter win-a 'adob--dim)))
+      (select-window (minibuffer-window))
+      (adob--update)
+      (should (window-parameter win-a 'adob--dim)))))
+
+(ert-deftest test-auto-dim-config-rescan-ignores-the-minibuffer-flag ()
+  "Error: `adob--rescan-windows' dims everything on a minibuffer selection.
+Known defect in the fork, pinned here rather than left undocumented.  The
+rescan is on `window-configuration-change-hook' and dims by window identity
+alone -- and `(window-list nil \\='n)' excludes the minibuffer, so when the
+minibuffer is selected nothing matches and every window dims.  A completion
+popup is the everyday case: `adob--update' honours the flag, this does not.
+
+Expected to fail until the fork honours the flag in the rescan too.  When it
+starts passing, ERT reports an unexpected pass -- that is the signal to drop
+this test and stop treating the gap as open.
+
+The `win-b' positive control is load-bearing here.  Three different broken
+implementations -- dimming disabled everywhere, the rescan never setting the
+parameter, `adob--update' made a no-op -- all produce an unexpected pass that
+would otherwise read as \"the fork fixed it\".  Asserting that `win-b' is still
+dimmed separates a real fix from dimming having broken."
+  :expected-result :failed
+  (skip-unless (file-directory-p test-auto-dim--fork))
+  (require 'auto-dim-config)
+  (test-auto-dim--with-two-windows win-a win-b
+    (adob--rescan-windows)
+    (should (null (window-parameter win-a 'adob--dim)))
+    (should (window-parameter win-b 'adob--dim))
+    (select-window (minibuffer-window))
+    (adob--rescan-windows)
+    (should (window-parameter win-b 'adob--dim))
+    (should (null (window-parameter win-a 'adob--dim)))))
+
 (defconst test-auto-dim--flat-dimmed-org-faces
   (append (mapcar (lambda (n) (intern (format "org-level-%d" n)))
                   (number-sequence 1 8))