check_for_equals_in_render_object_setters

v1.0.0WarningConfigurableWidget Best Practices

Flags a RenderObject setter that calls a mark-dirty method without comparing the incoming value first.

updateRenderObject runs on every rebuild and assigns every property, whether or not it differs. An unguarded setter therefore turns each rebuild into a full relayout or repaint of that subtree, even when nothing about it changed. At best that is wasted frames on a hot path; when the layout pass itself causes another rebuild, the two feed each other and the app stops rendering.

Flutter’s own render objects all use the same shape: compare, then assign and mark dirty.

This rule is in the opinionated preset, so it is on with preset: opinionated and preset: pedantic.

Don’t

class RenderBadge extends RenderBox {
  RenderBadge(this._color);

  Color _color;

  set color(Color value) {
    _color = value;
    markNeedsPaint();   // repaints on every rebuild, unchanged or not
  }
}

Do

The early return is the idiomatic form:

class RenderBadge extends RenderBox {
  RenderBadge(this._color);

  Color _color;

  set color(Color value) {
    if (_color == value) return;
    _color = value;
    markNeedsPaint();
  }
}

The wrapping form works just as well:

set color(Color value) {
  if (_color != value) {
    _color = value;
    markNeedsPaint();
  }
}

Layout and paint are separate decisions

A property that changes the object’s size needs markNeedsLayout; one that only changes how it draws needs markNeedsPaint. Both need the same guard:

class RenderBadge extends RenderBox {
  RenderBadge(this._padding, this._color);

  double _padding;
  Color _color;

  set padding(double value) {
    if (_padding == value) return;
    _padding = value;
    markNeedsLayout();      // size changed
  }

  set color(Color value) {
    if (_color == value) return;
    _color = value;
    markNeedsPaint();       // only the painting changed
  }
}

The recognised methods are markNeedsLayout, markNeedsPaint, markNeedsCompositingBitsUpdate, markNeedsSemanticsUpdate and markNeedsLayoutForSizedByParentChange.

Teaching it about a project wrapper

If your render objects mark themselves dirty through a helper of your own, name it so the rule keeps working:

analysis_options.yaml
many_lints.yaml

many_lints:
  rules:
    check_for_equals_in_render_object_setters:
      additional_methods: [markNeedsCustomPass]

rules:
  check_for_equals_in_render_object_setters:
    additional_methods: [markNeedsCustomPass]

// Reported under the config above
set weight(double value) {
  _weight = value;
  markNeedsCustomPass();
}

Known limitations

Guard detection is deliberately loose. Any ==, != or identical call anywhere in the setter body counts as a guard, even one comparing something else entirely. A false positive on a setter that is in fact guarded would be far more annoying than missing an exotic shape.

Only setters that mark dirty are considered. A setter that merely assigns has nothing to guard against, and is never reported.

Only a block body is examined. An expression-bodied setter is skipped.

No quick fix. Which field to compare — and whether an unchanged value should really skip the work — is a decision about the render object.

Options

Option	Type	Default	Description
`additional_methods`	list of strings	`[]`	Extra methods treated as marking the render object dirty

Turning this rule off

rules:
  check_for_equals_in_render_object_setters: false

To keep the rule on but skip certain paths, use per-rule exclude.

always_pass_global_key — Don’t create a GlobalKey inside build.
avoid_conditional_hooks — Never call hooks inside conditionals, loops, or ternaries.
avoid_deep_widget_nesting — Keep a widget tree within a nesting budget.
avoid_flexible_outside_flex — Only use Flexible and Expanded as direct children of Row, Column, or Flex.