Revert "Ensure Maybe propagates error information (#411)"

[Tankanow]

This reverts commit da3cc96.

See description in #439. I don't think this was intended to be a breaking change, but it will be for many consumers. The extra error information is not worth breaking consumers.

[coveralls]

Coverage remained the same at 95.707% when pulling c42ecf8 on Tankanow:ISSUE-439 into 5419cff on alecthomas:master.

[Tankanow]

@alecthomas, who has write access to this repo to consider this PR?

[spacegaier]

Alec as the owner has, but he is not active here anymore, so it's basically only me currently.

I am not convinced about your PR, hence why I did not comment yet. My current leaning is towards: That is a nonsensical validator in the Maybe context, since the goal of a Maybe is to specifically allow None.

Based on that, you could argue it is not breaking any "valid" use cases. Also given that yours is the only feedback after 0.12.1 has been out for a month, you seem to be the only one relying on that nonsensical validator.

[Tankanow]

@spacegaier, this is definitely a valid use case. The example I used in the Issue was just a demonstration. If I'm writing my own validators, I expect them to be composable with Maybe. In other words, I shouldn't have to do a null check in my own validator if I use it with Maybe, that's what Maybe is for.

Also, regardless of my example, this is a behavioral change in the library that is also breaking for many consumers.

[Tankanow]

@spacegaier, can we please revisit this issue. This is still a breaking change. Honestly, it makes the Maybe validator useless.

[alecthomas]

Agreed, merged.