Consolidate cursor resets in Reboot

[mdo]

After one of the comments in #25351, I think we can consolidate our cursor resets on non-disabled anchors and buttons into Reboot.

• Removes pagination link, .navbar-toggler, .close, and .btn cursor: pointer resets.
• Adds new ones to Reboot for links and buttons.

/cc @XhmikosR @patrickhlauke

[patrickhlauke]

wondering if this doesn't get us back where we started with regards to authors using random elements, for whatever reason, with, say .btn class. like (assuming they did it at least accessibly) <span tabindex="0" role="button" class="btn">Foo</span> or something.

[mdo]

With something like that, there's still no reason to not use a <button> element, correct? Point being, we shouldn't encourage that kind of markup when semantic elements work best. Also probably not worth trying to try custom cursors to every class—just adds extra CSS.

[XhmikosR]

I too think it sort of brings us back to the previous situation but it's simpler for us this way.

Now, guys, it's your call, but personally I'd go with this and with the docs stating to prefer the proper element.

[mdo]

I'd go with this and with the docs stating to prefer the proper element.

Yeah, noted. I can get some docs updates in here.

[patrickhlauke]

this would also force cursor:pointer for any anchors (<a name="blah">, or more generally <a> elements without an href). this will likely make react people happy (as they're usually the ones complaining, if memory serves), but will end up with confusing experience on sites that do use static anchors that are not links?

[patrickhlauke]

solution if we did want to just do this for real links would be to change selector to something like &[href]:not(disabled):not(.disabled)

[ysds]

Is this code necessary? When <a> is a link, browser makes the cursor to pointer by default.

[patrickhlauke]

ah, good point. yes, it seems at that point, it's redundant (and nicely circumvents the whole "what about anchors" dilemma)

[patrickhlauke]

i'll admit i'm conflicted. on the one hand, sure, using semantically correct elements should be the preferred approach. on the other hand we know for a fact not all our authors do it, and then come here and complain that we're not adding the cursor:pointer as part of the class.

but i guess if we actually document this, advise authors to manually add their own cursor:pointer if for whatever reason they can't use the correct element, and strictly point people to that part of the docs when they come here and open issues...that'd probably be ok?

[XhmikosR]

Should we drop this in favor of #27021?