Clearing an experiment restores an input element's default

The maintainer's ruling on the ambiguity the Android session surfaced:
after a clear the buffer gets the DEFAULT, not the position the control
was left in. Clearing is how a user deliberately returns an experiment to
its starting state, and it is what makes clearGroup coherent - a setting
that should survive the trash is exactly what that attribute exists for,
so without this ruling it would be answering a question clearing did not
ask.

Recorded on the default attribute of edit, toggle and dropdown: a value
the user has set is not overwritten while it is in the buffer, and
clearing it restores the default.

Android's toggle keeps the widget position and is the thing to change;
its edit already does the right thing and is the shape to copy, and the
dropdown has not been established either way. iOS already complies -
seedInputDefaults fills the emptied buffer and the switch view reads it
without writing back - so its item is a test to pin it, since it holds by
construction rather than by intent.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>

Author: Staacks

spec/views.yml

@@ -906,11 +906,15 @@ elements:
         remark: >-
           Applied whenever the buffer is EMPTY, not only once at load: when the
           experiment is opened, after the data is cleared, and at the start of
-          each analysis cycle. A value the user has entered is never
-          overwritten. Without the last two an experiment loses the setting the
-          moment anything empties the buffer - a clear, or an analysis input
-          without keep - and only gets it back if that element's own view
-          happens to be on screen.
+          each analysis cycle. Without the last two an experiment loses the
+          setting the moment anything empties the buffer - a clear, or an
+          analysis input without keep - and only gets it back if that element's
+          own view happens to be on screen. A value the user has set is not
+          overwritten while it is in the buffer; CLEARING IT RESTORES THE
+          DEFAULT, even though the control was showing something else, because
+          clearing is how a user deliberately returns an experiment to its
+          starting state. An author who wants a setting to survive that says so
+          with clearGroup on the data container, which is what it is for.
         default: 0.0
         summary: Value the field starts with.
         agreement: agreed
@@ -1015,11 +1019,15 @@ elements:
         remark: >-
           Applied whenever the buffer is EMPTY, not only once at load: when the
           experiment is opened, after the data is cleared, and at the start of
-          each analysis cycle. A value the user has entered is never
-          overwritten. Without the last two an experiment loses the setting the
-          moment anything empties the buffer - a clear, or an analysis input
-          without keep - and only gets it back if that element's own view
-          happens to be on screen.
+          each analysis cycle. Without the last two an experiment loses the
+          setting the moment anything empties the buffer - a clear, or an
+          analysis input without keep - and only gets it back if that element's
+          own view happens to be on screen. A value the user has set is not
+          overwritten while it is in the buffer; CLEARING IT RESTORES THE
+          DEFAULT, even though the control was showing something else, because
+          clearing is how a user deliberately returns an experiment to its
+          starting state. An author who wants a setting to survive that says so
+          with clearGroup on the data container, which is what it is for.
         default: 0.0
         summary: State the switch starts in.
         agreement: agreed
@@ -1145,11 +1153,15 @@ elements:
         remark: >-
           Applied whenever the buffer is EMPTY, not only once at load: when the
           experiment is opened, after the data is cleared, and at the start of
-          each analysis cycle. A value the user has entered is never
-          overwritten. Without the last two an experiment loses the setting the
-          moment anything empties the buffer - a clear, or an analysis input
-          without keep - and only gets it back if that element's own view
-          happens to be on screen.
+          each analysis cycle. Without the last two an experiment loses the
+          setting the moment anything empties the buffer - a clear, or an
+          analysis input without keep - and only gets it back if that element's
+          own view happens to be on screen. A value the user has set is not
+          overwritten while it is in the buffer; CLEARING IT RESTORES THE
+          DEFAULT, even though the control was showing something else, because
+          clearing is how a user deliberately returns an experiment to its
+          starting state. An author who wants a setting to survive that says so
+          with clearGroup on the data container, which is what it is for.
         default: 0.0
         summary: Value selected initially.
         agreement: agreed