pointer-events styles breaks title prop and cursor styles on disabled buttons

[vdh]

Checklist

• [x] I'm using Bulma version [0.1.0]
• [x] My browser is: Chrome 52.0.2743.82 (64-bit)

  Description

The pointer-events: none styles being used on .is-disabled or .button[disabled] selectors breaks any title/alt properties applied to button tags, and also prevents the cursor: not-allowed style from being applied to the button.

• .is-disabled on =control: sass/utilities/controls.sass:30
• .is-disabled on %control-with-element: sass/elements/form.sass:53, sass/elements/form.sass:55
• The global .is-disabled: sass/base/helpers.sass:98
• In a related style, .is-loading uses pointer-events: none as a hack to disable buttons: sass/elements/button.sass:106

I think it's a dangerous precedent to rely on pointer-events to control browser behaviour for disabling form controls. The disabled property is the more accurate and reliable way to disable a control. I can see how using this CSS style could make it easier to disable multiple elements at once by applying the is-disabled class and cascade it down to child elements, but that behaviour could be replicated with a disabled fieldset.

As noted on the MDN article about pointer-events:

Note that preventing an element from being the target of mouse events by using pointer-events does not necessarily mean that mouse event listeners on that element cannot or will not be triggered. […]

So without a dedicated disabled property, buttons and controls could still potentially be interacted with by non-mouse events such as keyboard input, or browser plugins (e.g. autofill or spellcheck plugins)

Steps to Reproduce

<button class="button is-disabled" title="Cannot remove the last item" disabled type="button">
  Remove
</button>

<!-- Or wrapped in a control -->
<div class="control is-disabled" title="Cannot remove the last set">
  <button class="button" disabled type="button">
    Remove
  </button>
</div>

Expected behavior: When hovered over, the browser toggles the not-allowed cursor and shows the title text in a tooltip.

Actual behavior: Both the tooltip and cursor style are not activated.

[jgthms]

Fixed by https://github.com/jgthms/bulma/pull/615

[homeworkprod]

@jgthms: It seems to me that the disabled attribute is not applicable to anchors as per the HTML5 standard.

However, you have changed examples to use the attribute on anchors.

This seems to be invalid (anyone, please correct me if I'm wrong).

I am under the impression that in order to style anchors as "disabled", the re-introduction of a corresponding CSS class is necessary.

[vdh]

The issue I raised technically only applied to form controls (buttons and inputs), I'm not quite sure why that change was also done for anchor tags.

@homeworkprod But even if a CSS class (or something like a[disabled] { pointer-events: none; }) were (re)added, it wouldn't completely disable anchors. Keyboard events will still be able to interact with them. Unfortunately, AFAIK the only way to completely disable anchors in the current spec is to bind a Javascript click event handler that calls event.preventDefault();.

[markedro]

Is this getting reverted back specifically on the tags? disabled doesn't appear to be considered as a valid attribute for anchors. It does appear that if disabled gets hard coded to an anchor directly when defining that, yes the element seems to show in a "disabled" state but subsequently it isn't able to be programmatically enabled / disabled beyond its original definition which means you are stuck with having to specifically define 2 separate anchors (specifically one enabled and one disabled) and then perform some type of show/hide toggle between them which is very sub-optimal. Or is there something I'm missing?

Yes this is still an issue in regard to anchors.

[hovancik]

Still issue (with anchors) in 2018. If anyone comes here, check the status at https://github.com/jgthms/bulma/issues/885

[vans163]

This is a two edged sword. By not relying on a class, you cannot click things inside the disabled element. For example we have an input box that is disabled, but when clicking the little lock which is INSIDE the input box, we want to enable it.

Cannot do that if disabled is a property of the input box, as the browser blocks onclick events.

[itised]

I think using an invalid attribute (disabled on an a tag) instead of a regular class is enough of a reason to bring back the is-disabled class, but there are valid reasons to bring back that is-disabled class even for elements where disabled is a valid attribute.

For example, I'd like to style a button as disabled until a certain condition is met. However, when the user clicks on the disabled button, I would like to let them know exactly why it has been disabled. Right now I can't do that, because the only way to style the button as disabled it to use the disabled attribute, which means I can't respond to click events.

I think keeping the is-disabled class but not disabling pointer events would be the best solution. Let the programmer handle the functionality, and let Bulma handle the styling.

For elements that actually have disabled attributes, I think it would be redundant to also require the is-disabled class. The same disabled styles can be applied to elements with the is-disabled class or the disabled attribute, but the programmer can decide which to use based on the functionality they want.

[tpartee-zspace]

I agree with the other commentors here that Bulma shouldn't be trying to intefere with the programming/logic and interaction side of things here. Keep it purely style-based. As it is, this attribute nonsense completely breaks modern frameworks like Angular and React. Just bring back is-disabled as a CSS class and be done with it. No pointer hacks, no attributes, just plain CSS styling class.

[anthonyboutinov]

I was unpleasantly surprised to see the ".is-disabled" being deprecated in Bulma. [disabled] introduces lots of accessibility issues when used in most scenarios.

Here's a great article about this: https://www.smashingmagazine.com/2021/08/frustrating-design-patterns-disabled-buttons/

It breaks keyboard navigation, making it a nightmare for accessibility.

There are literally just a few instances where it's okay to use [disabled]. In the majority of cases, you should not use it.