Drop 'when _' support

[makenowjust]

Ref #4837

I think this syntax has no use-case. It can be replaced with else block always.

[bew]

Note: this is currently invalid syntax:

case foo
else
  # bar
end

But this was valid:

case foo
when _
  # bar
end

I don't know the usecase of this, but it was spec-ed (in normalize/case_spec.cr), so there's probably a reason. Maybe macro-generated case node that could have no when or sth like that.

[makenowjust]

IMHO case having only else block should be valid if it cause a problem in reality.

[ysbaddaden]

I think it was added for not generating invalid code from a macro generating case statements that would happen to only generate the else part. You use when _ instead.

Probably we could support having only an else. I don't know why we chose this syntax.

[asterite]

To be honest, I didn't know that existed. I'm pretty sure I added it at one point because it's fancy, but not very useful.

Generating code in macros might be a reason to justify this, but we could just allow a single else instead, if we ever need this.

[asterite]

Or maybe it was added when we added tuple support in case... Which is another thing I think is a bit useless, and can be replaced with an if else chain (the need for tuple case is not super common, so using if else now and then is fine). I think we should remove that feature too.

[bcardiff]

It was added together with the tuple support and it make sense since _ can be used in other places:

_, b = expr

case tuple
when {0, _}
# ...
when {_, 0}
# ...
when {x, y}
# ...
end

I think we should have some minimum amount of time to merge PR so others has time to jump in ;-) 12hs in weekend for a syntax change seems to little time.

It is a corner case, I'm ok to the merge though.

[Sija]

It's definitely useful for pattern matching.

[asterite]

There's no such thing as pattern matching in Crystal. That's why _ was a bad idea, and why tuple in when was a bad idea. It makes it look like there's pattern matching, but there's not. So we should remove both.

Case is just an if with === or is_a?, but that's not pattern matching.

[makenowjust]

A bad point of when _ I think is it can be placed anywhere. For example,

foo = 1

case foo
when _
  puts "when _"
when 1
  puts "when 1"
end

Which output is correct, when _ or when 1? Currently it shows when _. However someone expects when 1 is correct.

Such a syntax introducing ambiguity makes us confusing and it will be called as bad parts in future.

[straight-shoota]

I think we should have some minimum amount of time to merge PR so others has time to jump in ;-)

yes, please!