Reject quantifiers on zero-width assertions

[natecook1000]

A repeated assertion, like \b+ causes an infinite loop, since it checks the same position over and over without advancing in the input. We should reject regexes with this pattern and prevent RegexBuilder regexes from creating them.

[milseman]

Status?

[natecook1000]

Still to-do. Do you have ideas about how to handle this on the builder side? We don't have a way to prevent this in the type system — what kind of warning / error can we raise during compilation?

[milseman]

We would fatal error during normal compilation. This could argue for a func compile() throws future API

[hamishknight]

Ah this should probably remain open for the RegexBuilder part?

[milseman]

Does it not fail compilation? I think for the builder, absent a compile() call, it will fatal error out at run time. I wonder how much of a concern this would be when composing builders as we don't have a Regex.isQuantifiable or anything like that...

@natecook1000 was thinking about a compile() throws on Regex, and this could be useful motivation. It's trivial for us to add, and it could make sense in the Unicode pitch because it's more like options etc. miscellany

[hamishknight]

It does fail compilation at run time, yes. I wasn't sure whether it was worth leaving this issue open to track more improvements in that area, but sounds like it would be a more general thing.