SetContainer is not derivable

[hseg]

Due to unions :: (MonoFoldable m, Element m ~ s) => m -> s, one cannot derive a SetContainer instance.
Specifically, the Element m ~ s constraint can't be coerced to the deriving type.
I suspect the same is true of the classes in Data.MonoTraversable, but haven't confirmed.

Notice, too, that as far as I can tell, the default for unions isn't overridden in any instances:
On github, the only instances for SetContainer I could find are https://github.com/AshleyYakeley/Truth/blob/master/shapes/src/Data/FiniteSet.hs, https://github.com/BlackCapCoder/dio/blob/main/src/Algebra/Divisibility.hs, and https://github.com/jnbooth/bitwise-enum/blob/master/Data/Enum/Set/Base.hs, all of which don't override the default.
And in mono-traversable itself, whether unions is overriden is inconsistent, and wherever it is the override is always of form M.unions . otoList.

All of this adds up to suggesting the appropriate definition would be unions :: Foldable f => f s -> s, with the only odd cases out being those from unordered-containers, who haven't generalized unions to arbitrary Foldables.
And if we're not averse to redefining it, we could just set unions = Foldable.foldl' union empty, as is its definition everywhere else.

[snoyberg]

I'd rather not have a breaking API change, but I'm open to a PR that keeps backwards compatibility.

[hseg]

Don't see how you could have both. As I said, the key problem is that the signature of unions makes SetContainer impossible to derive. Fixing it either requires moving unions out or changing its signature.

On July 6, 2021 4:44:04 AM GMT+03:00, Michael Snoyman ***@***.***> wrote: I'd rather not have a breaking API change, but I'm open to a PR that keeps backwards compatibility. -- You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub: #195 (comment)

-- Sent from my Android device with K-9 Mail. Please excuse my brevity.

[snoyberg]

Then I don’t think we can make the change

[hseg]

Fair enough. Will look for another way of doing what I need.

On July 6, 2021 8:10:42 AM GMT+03:00, Michael Snoyman ***@***.***> wrote: Then I don’t think we can make the change -- > You are receiving this because you authored the thread. Reply to this email directly or view it on GitHub: #195 (comment)

-- Sent from my Android device with K-9 Mail. Please excuse my brevity.

[BebeSparkelSparkel]

@hseg @gregwebs I'm not sure if this changes what snoyberg decided, but this comment is in Data.Containers

-- | Warning: This module should be considered highly experimental.

If the warning is still the case, then it seems like breaking changes are allowed. If not, then I think the comment should be removed.

[gregwebs]

I think it should be removed now.

To make a breaking change maybe we can make a new module?