@gjsify/react-native 0.45.0 → 0.47.0

package/PROPS.md

@@ -0,0 +1,525 @@
+<!-- GENERATED by scripts/generate-exports.mjs — do not edit. -->
+
+# What `@gjsify/react-native` does with each prop
+
+One section per primitive, generated from `src/primitives/table.ts` — the same data the
+renderer executes and `@gjsify/react-native/prop-table` publishes, so this document cannot
+disagree with what a render does.
+
+The **answer** column is `acceptsProp()`'s: `property`, `event`, `style`, `file`, `gesture` and
+`announcement` reach GTK; `ignored` is a DECLARED no-op, which is an answer and not a refusal;
+`refused` throws a `PrimitiveError` at render time. A prop that appears in no row throws too,
+naming the ones that are here.
+
+Ask this from a test rather than reading it: `acceptsProp("Text", "onPress")` is `false` and
+`explainProp("Text", "onPress")` is the sentence a render would have thrown (ADR 0039).
+
+## `<View>`
+
+widget `GtkBox` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `pointerEvents` | property | `can-target` | — |
+| `testID` | property | `name` | — |
+
+## `<Text>`
+
+widget `GtkLabel` · a text child writes `label`.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | announcement | `notify::label`, `Gtk.Accessible.announce()` | — |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `adjustsFontSizeToFit` | refused | — | GTK has no shrink-to-fit text; a label ellipsizes or wraps. Use `numberOfLines` with `ellipsizeMode` |
+| `allowFontScaling` | ignored | — | text scaling is a desktop-wide GNOME setting, not a per-label opt-in |
+| `ellipsizeMode` | property | `ellipsize` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `maxFontSizeMultiplier` | ignored | — | see `allowFontScaling` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `numberOfLines` | property | `lines` | — |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onPress` | refused | — | a `Gtk.Label` emits no `clicked` (measured: `activate-current-link`, `activate-link`, `copy-clipboard`, `move-cursor` and nothing else). Wrap it in a `<Pressable>`, which is a real `Gtk.Button` |
+| `selectable` | property | `selectable` | — |
+| `testID` | property | `name` | — |
+
+## `<Pressable>`
+
+widget `GtkButton` · a text child writes `label`.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `android_ripple` | ignored | — | an Android-only ripple; Adwaita has its own press feedback |
+| `delayLongPress` | refused | — | see `onLongPress` |
+| `disabled` | property | `sensitive` | — |
+| `hitSlop` | refused | — | GTK hit-tests the allocation and has no way to grow it past the widget. Pad the button (`p-*`), which enlarges the real target |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onLongPress` | refused | — | a long press is a `Gtk.GestureLongPress` controller added to the widget, not a property or a signal on it, so it is not something this layer can route as data |
+| `onPress` | event | `clicked` | — |
+| `onPressIn` | refused | — | press state is the GTK CSS `:active` pseudo-class, not React state (ADR 0032 § 7) — write `active:opacity-70` and GTK animates it with no re-render at all |
+| `onPressOut` | refused | — | see `onPressIn` — `active:` is the mechanism |
+| `pointerEvents` | property | `can-target` | — |
+| `testID` | property | `name` | — |
+
+## `<ScrollView>`
+
+widget `GtkScrolledWindow` · children go into `GtkBox` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `contentContainerClassName` | ignored | — | read directly as the content node’s class list |
+| `contentContainerStyle` | ignored | — | read directly as the content node’s style set |
+| `horizontal` | property | `orientation`, `hscrollbar-policy`, `vscrollbar-policy` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `keyboardDismissMode` | ignored | — | see `keyboardShouldPersistTaps` |
+| `keyboardShouldPersistTaps` | ignored | — | there is no on-screen keyboard to dismiss |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onScroll` | refused | — | GTK reports scroll position through the `Gtk.Adjustment` objects behind `hadjustment`/`vadjustment` — `notify::value` on an adjustment, not a signal on the scrolled window — so there is no prop to route. Reach the adjustment through a ref |
+| `refreshControl` | refused | — | GTK has no pull-to-refresh idiom and `RefreshControl` is `refused` in the support table for that reason. Give the desktop build a refresh action |
+| `scrollEnabled` | refused | — | a `Gtk.ScrolledWindow` has no switch that stops scrolling — setting both policies to `never` hides the bars AND clamps the child, which is a different thing. Render the content without the ScrollView when it must not scroll |
+| `scrollEventThrottle` | refused | — | see `onScroll` |
+| `showsHorizontalScrollIndicator` | property | `hscrollbar-policy` | — |
+| `showsVerticalScrollIndicator` | property | `vscrollbar-policy` | — |
+| `testID` | property | `name` | — |
+
+## `<ActivityIndicator>`
+
+widget `AdwSpinner` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `animating` | property | `visible` | — |
+| `color` | style | `style: color` | — |
+| `hidesWhenStopped` | ignored | — | a stopped `Adw.Spinner` is a hidden one here — there is no property that stops it while it is on screen (measured), so this is already the behaviour |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `size` | property | `width-request`, `height-request` | — |
+| `testID` | property | `name` | — |
+
+## `<TextInput>`
+
+widget `GtkEntry` · a text child writes `text` · a `ref` receives the `text-input` handle, not the widget.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `autoCapitalize` | ignored | — | an on-screen keyboard behaviour; a hardware keyboard has no shift state to preset |
+| `autoComplete` | ignored | — | is Android’s autofill hint, and there is no autofill service on a GTK desktop to hint to — a password manager fills a field through the accessibility bus or a paste, neither of which the widget declares anything for. `keyboardType` is the one hint that does land: it writes `Gtk.Entry:input-purpose`, which an input method reads |
+| `autoCorrect` | ignored | — | an on-screen keyboard behaviour |
+| `autoFocus` | refused | — | focus is `Gtk.Widget.grab_focus()`, which only works once the widget is mapped — an imperative call at a moment this layer does not own. Call it from a ref in an effect |
+| `defaultValue` | property | `text` | — |
+| `editable` | property | `editable` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `keyboardAppearance` | ignored | — | there is no on-screen keyboard to theme |
+| `keyboardType` | property | `input-purpose` | — |
+| `maxLength` | property | `max-length` | — |
+| `multiline` | ignored | — | it chose the widget; it is not also a property |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onBlur` | refused | — | see `onFocus` — the same controller |
+| `onChangeText` | event | `notify::text` | — |
+| `onFocus` | refused | — | focus arrives through a `Gtk.EventControllerFocus`, a controller rather than a signal on the widget, so it is not a prop this layer can route |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onSubmitEditing` | event | `activate` | — |
+| `placeholder` | property | `placeholder-text` | — |
+| `returnKeyType` | ignored | — | a desktop keyboard has one Return key and no label to change |
+| `secureTextEntry` | property | `visibility` | — |
+| `submitBehavior` | ignored | — | chooses what the Return key does, and each GTK widget has exactly one answer with nothing to select between: a `Gtk.Entry` always emits `activate` (which is `onSubmitEditing`) and can never insert a newline, and a `Gtk.TextView` always inserts a newline and never submits — which is why `onSubmitEditing` is refused on a multiline input. `multiline` is the prop that picks the behaviour, because it picks the widget |
+| `testID` | property | `name` | — |
+| `textContentType` | ignored | — | is `autoComplete`’s iOS spelling and has the same absent addressee. Note `Gtk.Accessible` does carry an `autocomplete` PROPERTY, and it is a different thing: it describes a widget that completes its own text (a combo box), not a system that fills it in |
+| `value` | property | `text` | — |
+
+## `<TextInput multiline>`
+
+widget `GtkTextView` · takes no text child · a `ref` receives the `text-input` handle, not the widget.
+
+One React Native prop, two GTK widgets: `multiline` selects this row instead of the one above.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `autoCapitalize` | ignored | — | an on-screen keyboard behaviour; a hardware keyboard has no shift state to preset |
+| `autoComplete` | ignored | — | is Android’s autofill hint, and there is no autofill service on a GTK desktop to hint to — a password manager fills a field through the accessibility bus or a paste, neither of which the widget declares anything for. `keyboardType` is the one hint that does land: it writes `Gtk.Entry:input-purpose`, which an input method reads |
+| `autoCorrect` | ignored | — | an on-screen keyboard behaviour |
+| `autoFocus` | refused | — | focus is `Gtk.Widget.grab_focus()`, which only works once the widget is mapped — an imperative call at a moment this layer does not own. Call it from a ref in an effect |
+| `defaultValue` | refused | — | `Gtk.TextView` keeps its content in a `Gtk.TextBuffer` rather than in a property (measured: 61 properties, no `text`), so there is nothing for this layer to set as data. Drop `multiline` for a `Gtk.Entry`, or reach the buffer through a ref |
+| `editable` | property | `editable` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `keyboardAppearance` | ignored | — | there is no on-screen keyboard to theme |
+| `keyboardType` | property | `input-purpose` | — |
+| `maxLength` | refused | — | `Gtk.TextView` installs no `max-length` (measured) — the limit belongs to the buffer |
+| `multiline` | ignored | — | it chose the widget; it is not also a property |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onBlur` | refused | — | see `onFocus` — the same controller |
+| `onChangeText` | refused | — | `Gtk.TextView` keeps its content in a `Gtk.TextBuffer` rather than in a property (measured: 61 properties, no `text`), so there is nothing for this layer to set as data. Drop `multiline` for a `Gtk.Entry`, or reach the buffer through a ref |
+| `onFocus` | refused | — | focus arrives through a `Gtk.EventControllerFocus`, a controller rather than a signal on the widget, so it is not a prop this layer can route |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onSubmitEditing` | refused | — | Return inserts a newline in a multiline field; there is nothing to submit |
+| `placeholder` | refused | — | `Gtk.TextView` installs no `placeholder-text` (measured); the Adwaita idiom is a label above the field |
+| `returnKeyType` | ignored | — | a desktop keyboard has one Return key and no label to change |
+| `secureTextEntry` | refused | — | `Gtk.TextView` installs no `visibility` (measured), and a multiline password field is not a thing a desktop offers |
+| `submitBehavior` | ignored | — | chooses what the Return key does, and each GTK widget has exactly one answer with nothing to select between: a `Gtk.Entry` always emits `activate` (which is `onSubmitEditing`) and can never insert a newline, and a `Gtk.TextView` always inserts a newline and never submits — which is why `onSubmitEditing` is refused on a multiline input. `multiline` is the prop that picks the behaviour, because it picks the widget |
+| `testID` | property | `name` | — |
+| `textContentType` | ignored | — | is `autoComplete`’s iOS spelling and has the same absent addressee. Note `Gtk.Accessible` does carry an `autocomplete` PROPERTY, and it is a different thing: it describes a widget that completes its own text (a combo box), not a system that fills it in |
+| `value` | refused | — | `Gtk.TextView` keeps its content in a `Gtk.TextBuffer` rather than in a property (measured: 61 properties, no `text`), so there is nothing for this layer to set as data. Drop `multiline` for a `Gtk.Entry`, or reach the buffer through a ref |
+
+## `<Switch>`
+
+widget `GtkSwitch` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `disabled` | property | `sensitive` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `ios_backgroundColor` | ignored | — | an iOS-only fallback colour |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onValueChange` | event | `notify::active` | — |
+| `testID` | property | `name` | — |
+| `thumbColor` | refused | — | see `trackColor` |
+| `trackColor` | refused | — | Adwaita paints the switch from the theme’s accent colour, and its track is a CSS `slider`/`trough` subnode rather than a property. Style it from the application stylesheet |
+| `value` | property | `active` | — |
+
+## `<Image>`
+
+widget `GtkPicture` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `alt` | property | `alternative-text` | — |
+| `blurRadius` | refused | — | is a per-pixel effect. GTK composites a `Gdk.Paintable` and has no filter property on the widget that draws it — the counterpart is a `Gsk` render node or a `Gtk.Snapshot` subclass, which is a widget of its own rather than a prop on this one |
+| `capInsets` | refused | — | is iOS nine-part stretching. GTK’s counterpart is a CSS `border-image` on a widget’s background, which is a different widget and a different property set |
+| `defaultSource` | refused | — | is the placeholder shown while the real image loads, and `Gtk.Picture` has one file at a time with no load event to swap on (measured: it emits no signals). Render a placeholder widget beside it and remove it yourself |
+| `fadeDuration` | ignored | — | an Android cross-fade on load; there is no load event here to fade from (measured) |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `loadingIndicatorSource` | refused | — | see `defaultSource` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onError` | refused | — | is a load event, and `Gtk.Picture` emits NO signals at all (measured: its GType lists none) — it is handed a file or a paintable and draws it, reporting neither progress nor failure. A missing file leaves `paintable` null with no diagnostic, so a callback that never fires would hide exactly the case it exists to report. Load the bytes yourself and set `paintable` through a ref |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onLoad` | refused | — | is a load event, and `Gtk.Picture` emits NO signals at all (measured: its GType lists none) — it is handed a file or a paintable and draws it, reporting neither progress nor failure. A missing file leaves `paintable` null with no diagnostic, so a callback that never fires would hide exactly the case it exists to report. Load the bytes yourself and set `paintable` through a ref |
+| `onLoadEnd` | refused | — | is a load event, and `Gtk.Picture` emits NO signals at all (measured: its GType lists none) — it is handed a file or a paintable and draws it, reporting neither progress nor failure. A missing file leaves `paintable` null with no diagnostic, so a callback that never fires would hide exactly the case it exists to report. Load the bytes yourself and set `paintable` through a ref |
+| `onLoadStart` | refused | — | is a load event, and `Gtk.Picture` emits NO signals at all (measured: its GType lists none) — it is handed a file or a paintable and draws it, reporting neither progress nor failure. A missing file leaves `paintable` null with no diagnostic, so a callback that never fires would hide exactly the case it exists to report. Load the bytes yourself and set `paintable` through a ref |
+| `onProgress` | refused | — | is a load event, and `Gtk.Picture` emits NO signals at all (measured: its GType lists none) — it is handed a file or a paintable and draws it, reporting neither progress nor failure. A missing file leaves `paintable` null with no diagnostic, so a callback that never fires would hide exactly the case it exists to report. Load the bytes yourself and set `paintable` through a ref |
+| `progressiveRenderingEnabled` | ignored | — | GTK decodes an image in one pass |
+| `resizeMethod` | ignored | — | an Android decoder hint |
+| `resizeMode` | property | `content-fit` | — |
+| `resizeMultiplier` | ignored | — | an Android decoder hint |
+| `source` | file | `file` | — |
+| `testID` | property | `name` | — |
+| `tintColor` | refused | — | is a per-pixel effect. GTK composites a `Gdk.Paintable` and has no filter property on the widget that draws it — the counterpart is a `Gsk` render node or a `Gtk.Snapshot` subclass, which is a widget of its own rather than a prop on this one |
+
+## `<ImageBackground>`
+
+widget `GtkOverlay` · children go into `GtkBox` · backdrop `GtkPicture` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `alt` | property | `alternative-text` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `imageRef` | refused | — | hands back the inner `Image`’s instance. This element’s `ref` is the `Gtk.Overlay`; the picture is its main child and `overlay.get_child()` reaches it, so a second ref would be a second way to say the same thing |
+| `imageStyle` | ignored | — | read directly as the picture node’s style set |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `pointerEvents` | property | `can-target` | — |
+| `resizeMode` | property | `content-fit` | — |
+| `source` | file | `file` | — |
+| `testID` | property | `name` | — |
+
+## `<TouchableOpacity>`
+
+widget `GtkButton` · a text child writes `label`.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `activeOpacity` | refused | — | sets the pressed appearance, which on GTK is the CSS `:active` pseudo-class rather than a prop (ADR 0032 § 7). Write `active:opacity-70` — it resolves through the same variant mechanism as every other utility, reads the project's own token scale, and GTK animates it with no re-render at all |
+| `delayLongPress` | refused | — | see `onLongPress` |
+| `delayPressIn` | refused | — | see `onPressIn` — there is no press-in event to delay |
+| `delayPressOut` | refused | — | see `onPressIn` |
+| `disabled` | property | `sensitive` | — |
+| `hitSlop` | refused | — | GTK hit-tests the allocation and has no way to grow it past the widget. Pad the element (`p-*`), which enlarges the real target |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onLongPress` | refused | — | a long press is a `Gtk.GestureLongPress` controller added to the widget, not a property or a signal on it, so it is not something this layer can route as data |
+| `onPress` | event | `clicked` | — |
+| `onPressIn` | refused | — | press state is the GTK CSS `:active` pseudo-class, not React state (ADR 0032 § 7) — write `active:opacity-70` and GTK animates it with no re-render at all |
+| `onPressOut` | refused | — | see `onPressIn` — `active:` is the mechanism |
+| `pointerEvents` | property | `can-target` | — |
+| `pressRetentionOffset` | refused | — | see `hitSlop` — GTK hit-tests the allocation and nothing around it |
+| `testID` | property | `name` | — |
+| `touchSoundDisabled` | ignored | — | an Android touch sound; a desktop button makes none |
+
+## `<TouchableHighlight>`
+
+widget `GtkButton` · a text child writes `label`.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `activeOpacity` | refused | — | sets the pressed appearance, which on GTK is the CSS `:active` pseudo-class rather than a prop (ADR 0032 § 7). Write `active:opacity-70` — it resolves through the same variant mechanism as every other utility, reads the project's own token scale, and GTK animates it with no re-render at all |
+| `delayLongPress` | refused | — | see `onLongPress` |
+| `delayPressIn` | refused | — | see `onPressIn` — there is no press-in event to delay |
+| `delayPressOut` | refused | — | see `onPressIn` |
+| `disabled` | property | `sensitive` | — |
+| `hitSlop` | refused | — | GTK hit-tests the allocation and has no way to grow it past the widget. Pad the element (`p-*`), which enlarges the real target |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onHideUnderlay` | refused | — | see `underlayColor` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onLongPress` | refused | — | a long press is a `Gtk.GestureLongPress` controller added to the widget, not a property or a signal on it, so it is not something this layer can route as data |
+| `onPress` | event | `clicked` | — |
+| `onPressIn` | refused | — | press state is the GTK CSS `:active` pseudo-class, not React state (ADR 0032 § 7) — write `active:opacity-70` and GTK animates it with no re-render at all |
+| `onPressOut` | refused | — | see `onPressIn` — `active:` is the mechanism |
+| `onShowUnderlay` | refused | — | see `underlayColor` — there is no underlay to show |
+| `pointerEvents` | property | `can-target` | — |
+| `pressRetentionOffset` | refused | — | see `hitSlop` — GTK hit-tests the allocation and nothing around it |
+| `testID` | property | `name` | — |
+| `touchSoundDisabled` | ignored | — | an Android touch sound; a desktop button makes none |
+| `underlayColor` | refused | — | sets the pressed appearance, which on GTK is the CSS `:active` pseudo-class rather than a prop (ADR 0032 § 7). Write `active:bg-<token>` — it resolves through the same variant mechanism as every other utility, reads the project's own token scale, and GTK animates it with no re-render at all |
+
+## `<TouchableWithoutFeedback>`
+
+widget `GtkBox` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `delayLongPress` | refused | — | see `onLongPress` |
+| `delayPressIn` | refused | — | see `onPressIn` — there is no press-in event to delay |
+| `delayPressOut` | refused | — | see `onPressIn` |
+| `disabled` | property | `can-target` | — |
+| `hitSlop` | refused | — | GTK hit-tests the allocation and has no way to grow it past the widget. Pad the element (`p-*`), which enlarges the real target |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onLongPress` | refused | — | a long press is a `Gtk.GestureLongPress` controller added to the widget, not a property or a signal on it, so it is not something this layer can route as data |
+| `onPress` | gesture | `Gtk.GestureClick::released` | — |
+| `onPressIn` | refused | — | press state is the GTK CSS `:active` pseudo-class, not React state (ADR 0032 § 7) — write `active:opacity-70` and GTK animates it with no re-render at all |
+| `onPressOut` | refused | — | see `onPressIn` — `active:` is the mechanism |
+| `pointerEvents` | property | `can-target` | — |
+| `pressRetentionOffset` | refused | — | see `hitSlop` — GTK hit-tests the allocation and nothing around it |
+| `testID` | property | `name` | — |
+| `touchSoundDisabled` | ignored | — | an Android touch sound; a desktop button makes none |
+
+## `<Button>`
+
+widget `GtkButton` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `color` | refused | — | is the button’s background on Android and its text colour on iOS — one prop, two meanings, and GTK has a third answer: `suggested-action` and `destructive-action` are Adwaita’s own classes for the two cases a colour is usually asking for. Set them from the application stylesheet, or use `<Pressable>` |
+| `disabled` | property | `sensitive` | — |
+| `hasTVPreferredFocus` | ignored | — | an Android TV focus hint |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nextFocusDown` | ignored | — | an Android TV focus hint |
+| `nextFocusForward` | ignored | — | an Android TV focus hint |
+| `nextFocusLeft` | ignored | — | an Android TV focus hint |
+| `nextFocusRight` | ignored | — | an Android TV focus hint |
+| `nextFocusUp` | ignored | — | an Android TV focus hint |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onPress` | event | `clicked` | — |
+| `testID` | property | `name` | — |
+| `title` | property | `label` | — |
+| `touchSoundDisabled` | ignored | — | an Android touch sound; a desktop button makes none |
+
+## `<SafeAreaView>`
+
+widget `GtkBox` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `pointerEvents` | property | `can-target` | — |
+| `testID` | property | `name` | — |
+
+## `<Icon>`
+
+widget `GtkImage` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `color` | style | `style: color` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `name` | property | `icon-name` | — |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `size` | property | `pixel-size` | — |
+| `testID` | property | `name` | — |
+
+## `<KeyboardAvoidingView>`
+
+widget `GtkBox` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `behavior` | ignored | — | describes how to move out of an on-screen keyboard’s way, and none appears |
+| `contentContainerStyle` | refused | — | styles the inner view React Native adds to do the avoiding, and there is no avoiding here — so there is no second node to style. Put it on `style`, which lands on the one box this element is |
+| `enabled` | ignored | — | see `behavior` — the avoidance is already off |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `keyboardVerticalOffset` | ignored | — | see `behavior` |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `pointerEvents` | property | `can-target` | — |
+| `testID` | property | `name` | — |
+
+## `<FlatList>`
+
+widget `GtkBox` · children go into `GtkScrolledWindow` · takes no text child.
+
+| prop | answer | GTK | why |
+|---|---|---|---|
+| `ItemSeparatorComponent` | refused | — | puts a widget BETWEEN rows, and a `Gtk.ListView` has no between — it has rows. `Gtk.ListView:show-separators` draws the Adwaita separator itself; set it through a ref, or give the row its own bottom border with `border-b` |
+| `ListEmptyComponent` | ignored | — | rendered by the component instead of the scroller |
+| `ListFooterComponent` | ignored | — | rendered by the component, after the scroller |
+| `ListHeaderComponent` | ignored | — | rendered by the component, before the scroller |
+| `accessibilityHint` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLabel` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityLiveRegion` | refused | — | declares that a screen reader should speak this element when its CONTENT changes. GTK has the announcement — `Gtk.Accessible.announce(message, priority)` since 4.14 — but not the watch: it is an imperative call, and this element’s content is a SUBTREE, which GTK emits no signal for. `<Text>` answers this prop, because its content is one property (`Gtk.Label:label`) and so is both the moment and the message. Elsewhere, call `announce()` on the widget through a ref when you change the content |
+| `accessibilityRole` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessibilityState` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `accessible` | refused | — | GTK carries accessibility through `Gtk.Accessible.update_property()`, an imperative call, not through a widget property — so there is nothing for this layer to set as data. `AccessibilityInfo` (tier P3) is the entry that owns this, and GTK’s model maps onto the props well once it exists |
+| `columnWrapperStyle` | refused | — | see `numColumns` |
+| `contentContainerStyle` | refused | — | styles the box React Native puts the rows in, and there is no such box: `Gtk.ListView` builds each row from the factory and owns the space between them. Style the row inside `renderItem`, or the list itself through `style` |
+| `data` | ignored | — | read by the component, which drives the Gio.ListStore from it |
+| `extraData` | ignored | — | read by the component; a change re-renders every bound row |
+| `getItem` | ignored | — | read by the component — VirtualizedList’s accessor form of `data` |
+| `getItemCount` | ignored | — | see `getItem` |
+| `getItemLayout` | refused | — | tunes React Native’s own virtualisation, and `Gtk.ListView` does that job itself — it creates and recycles rows through the factory as the viewport moves (measured: 205 of 500 rows set up in a 400×300 window) and installs no property that changes the batching. There is nothing for this to set |
+| `horizontal` | property | `orientation`, `hscrollbar-policy`, `vscrollbar-policy` | — |
+| `id` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `initialNumToRender` | refused | — | tunes React Native’s own virtualisation, and `Gtk.ListView` does that job itself — it creates and recycles rows through the factory as the viewport moves (measured: 205 of 500 rows set up in a 400×300 window) and installs no property that changes the batching. There is nothing for this to set |
+| `initialScrollIndex` | refused | — | scrolls to a row before the first paint. `Gtk.ListView.scroll_to(position, flags, scroll_info)` is the counterpart and it is an imperative call on a realised view, not a property — reach the view through a ref and call it |
+| `inverted` | refused | — | draws the list bottom-up. GTK has no reversed list view; the honest equivalent is to reverse `data` yourself, which is one line and visible in the code |
+| `keyExtractor` | ignored | — | read by the component, to key the model rows |
+| `keyboardDismissMode` | ignored | — | see `keyboardShouldPersistTaps` |
+| `keyboardShouldPersistTaps` | ignored | — | there is no on-screen keyboard to dismiss |
+| `maxToRenderPerBatch` | refused | — | tunes React Native’s own virtualisation, and `Gtk.ListView` does that job itself — it creates and recycles rows through the factory as the viewport moves (measured: 205 of 500 rows set up in a 400×300 window) and installs no property that changes the batching. There is nothing for this to set |
+| `nativeID` | refused | — | would write `Gtk.Widget:name`, and so does `testID`. Two props for one property is the silent-drop shape the host refuses by name (`signalTaken`), so this layer picks one: use `testID` |
+| `numColumns` | refused | — | lays the rows out in a grid, and a `Gtk.ListView` is one column of rows by construction. `Gtk.GridView` is the widget that does this and it is a different tag with a different model contract, so it is a primitive of its own rather than a prop on this one |
+| `onEndReached` | ignored | — | bound by the component to the scroller’s own Gtk.Adjustment |
+| `onEndReachedThreshold` | ignored | — | see `onEndReached` |
+| `onLayout` | refused | — | GTK reports its allocation through `Gtk.Widget.vfunc_size_allocate`, a SUBCLASS override rather than a signal, so there is no handler to bind. `useWindowDimensions` (tier P2) is the window-level answer |
+| `onRefresh` | refused | — | is pull-to-refresh, which GTK has no idiom for and should not grow one — `RefreshControl` is `refused` in the support table for the same reason. Give the desktop build a refresh action in the header bar |
+| `onRowError` | ignored | — | read by the component, and handed to each row’s own React root |
+| `onScroll` | refused | — | GTK reports scroll position through the `Gtk.Adjustment` objects behind the scroller’s `hadjustment`/`vadjustment` — `notify::value` on an adjustment, not a signal on the widget. `onEndReached` is the one question this layer answers from it; reach the adjustment through a ref for the rest |
+| `onViewableItemsChanged` | refused | — | reports which rows are on screen. `Gtk.ListView` binds and unbinds rows itself and exposes no viewport range — the closest fact is `Gtk.ListItem:position` inside the factory, which is per row rather than a set, so a faithful answer would be a guess assembled from bind order |
+| `refreshControl` | refused | — | is pull-to-refresh, which GTK has no idiom for and should not grow one — `RefreshControl` is `refused` in the support table for the same reason. Give the desktop build a refresh action in the header bar |
+| `refreshing` | refused | — | is pull-to-refresh, which GTK has no idiom for and should not grow one — `RefreshControl` is `refused` in the support table for the same reason. Give the desktop build a refresh action in the header bar |
+| `removeClippedSubviews` | refused | — | tunes React Native’s own virtualisation, and `Gtk.ListView` does that job itself — it creates and recycles rows through the factory as the viewport moves (measured: 205 of 500 rows set up in a 400×300 window) and installs no property that changes the batching. There is nothing for this to set |
+| `renderItem` | ignored | — | read by the item factory |
+| `renderSectionHeader` | ignored | — | read by the item factory |
+| `scrollEventThrottle` | refused | — | see `onScroll` |
+| `sections` | ignored | — | read by the component — SectionList’s form of `data` |
+| `showsHorizontalScrollIndicator` | property | `hscrollbar-policy` | — |
+| `showsVerticalScrollIndicator` | property | `vscrollbar-policy` | — |
+| `stickySectionHeadersEnabled` | refused | — | GTK’s own sticky headers are a `Gtk.ListView:header-factory` over a model that implements `Gtk.SectionModel` (MEASURED: `Gio.ListStore` does not, every selection model and `Gtk.FlattenListModel` do). This layer flattens the sections into one model with header ROWS instead, so a header scrolls with its section — which is what `stickySectionHeadersEnabled={false}` asks for and the only thing it can deliver |
+| `testID` | property | `name` | — |
+| `updateCellsBatchingPeriod` | refused | — | tunes React Native’s own virtualisation, and `Gtk.ListView` does that job itself — it creates and recycles rows through the factory as the viewport moves (measured: 205 of 500 rows set up in a 400×300 window) and installs no property that changes the batching. There is nothing for this to set |
+| `viewabilityConfig` | refused | — | see `onViewableItemsChanged` |
+| `viewabilityConfigCallbackPairs` | refused | — | see `onViewableItemsChanged` |
+| `windowSize` | refused | — | tunes React Native’s own virtualisation, and `Gtk.ListView` does that job itself — it creates and recycles rows through the factory as the viewport moves (measured: 205 of 500 rows set up in a 400×300 window) and installs no property that changes the batching. There is nothing for this to set |