Components inside a disabled Container are not disabled

[Artur Signell]

When a for example a panel is disabled, components inside it used to be (not sure?) disabled. Currently components inside a disable panel are rendered as disabled but are not really disabled.

[Artur Signell]

Same problem with Form.

[Jani Laakso]

I've noticed this too, but is this a bug? Isn't this correct functionality?

[Jani Laakso]

Ignore last comment, mixed it up with setVisible which seems to work.

So it seems that setVisible(false) is cascaded to all components within the container but setEnable(false) is not.

[Jani Laakso]

On Millstone this has worked (if Artur's memory serves right)

[Jani Laakso]

See also #767

[Matti Tahvonen]

Currently it is up to developer to do child component disabling. Maybe it will be implemented later.

[Matti Tahvonen]

will not fit into 5.2

[Artur Signell]

Fixed for AbstractComponentContainers? and Form in [5252] and [5253]

[Matti Tahvonen]

Fix is not valid. Makes situation just weird with multiple levels. Improved test case in 5450.

Also I wouldn't like to see public setDisabledByContainer method in AbstractComponent?. It is leaked to all components and is more easily hit by developer than setEnabled method.

Correct way would be to recursively check the whole tree in isEnabled, but this'd sure add some overhead for render phase.

If a well working solution is not found, I'd suggest that we keep in the old implementation (enabled flag is only for component it self, not childs), emphasis this in documentation and maybe add helper to disable/enable whole child tree.

[Marc Englund]

Sort of with Matti here; 'climbing' upwards would be a great solution, except that the default case (no disabled parents) is actually the worst case performance-wise.

As a compromise, perhaps the naming of the setDisabledByContainer() should be changed, so that it's not used accidentally?

Some alternatives:

• An additional setEnabled(boolean enabled, boolean parentEnabled) would perhaps be sufficiently weird so that it's not used accidentally?
• An additional setEnabled(boolean enabled, ComponentContainer? setBy) would also make the developer read the javadoc, and could throw if setBy!=parent.

Also note that in any case, the ComponentContainer? should reset this state when a component is attached to it (addComponent())

[Matti Tahvonen]

Note, naming issue is only a side story. Issue with current implementation: two disabled parent containers -> child enabled (although theme works).

[Marc Englund]

Mostly fixed in [5513], Table still needs fixing.

[Marc Englund]

Table fixed in [5514]

[Marc Englund]

Oops, [5515] too...