Semigroup on refinements?

[ocharles]

I wanted to set outline-style:none for both a:active and a:visited, and thought I could do (":active" <> ":hover") & outlineStyle none, but Refinement is not a Monoid or Semigroup instance. In the end I did:

forM_ [":active", ":focus"] (& outlineStyle none)

Which does the job, but it is a tad verbose.

It also doesn't produce a:active, a:focus as I wanted, but instead gives me two separate blocks.

[jameshfisher]

I wanted to set outline-style:none for both a:active and a:visited

thought I could do (":active" <> ":hover")

forM_ [":active", ":focus"]

Just to clarify, should the :visited, :hover and :focus things be the same?

[jameshfisher]

Does this do what you wanted?

((a # active) <> (a # hover)) ? outlineStyle none

Refinement isn't a Monoid, but Selector is.

[ocharles]

It does, but if I am already "in" a then I have to make another rule, which wasn't quite what I was expecting to do.

And yes, all of those pseudo selectors should have the same rules.

[jameshfisher]

icwym, I experienced this the other day.

[sebastiaanvisser]

Makes sense. You do want to be able to say a # (active <> hover), but this requires some additional machinery.

[jhenahan]

I'm a bit curious about how you'd like this to work in certain cases.

With the trivial Monoid instance I'm playing with, this renders as a:active:hover, which is not what you want (and, strictly speaking, is ineffective since :hover must come before :active), but I think it fits the spirit of the Monoid better. I'm mainly thinking about cases likea:hover:last-child.

Conceptually, I think this is the right way to make Refinement a Monoid, but I'm open to being convinced that the split case is the better default.

[ocharles]

I agree that what your suggesting. I can't say I gave the semigroup much thought, but seeing as pseudo selectors "stack" I'd be ok with your proposal.

[Profpatsch]

I stumbled over Refinements not being a semigroup, too. But I think what the OP wants is not what the instance should be: a:focus:hover selects all elements that are focused and hovered over (as far as I understand it).

So it would more sense to make it a commutative semigroup that combines its effects in that manner.

[turion]

I guess it would make sense to make Refinement a Semigroup, but I'm not sure I understand this well enough. PR welcome.

[github-actions]

Stale issue message