Using alternates within escaped parentheses doesn't work about cucumber-expressions

[mpkorstanje]

The idea that the \( and \) go together only exists in your head. The alternation is bounded only by spaces.

Compare:

\(b c/d\)
some rat/cats

Just because there is an s or ( at the start, it doesn't mean that the s or ) at the end is paired with it .

[mpkorstanje]

Maybe a better question, since the locally/globally isn't captured, why even mention it? Or use "globally or locally" to make it clear that it doesn't matter.

[mattwynne]

In fact the locally/globally is captured (or at least was), so this isn't the right expression anyway for the particular problem at hand, but that's not really the point.

I was just surprised by the behaviour and thought it might be useful to make it less surprising.

The alternation is bounded only by spaces.

That's the rule I was questioning. Maybe it's more sensible to have it bound by any non-word character? I can just about imagine people wanting to write I have eaten {int} cucumbers/carrots, {int} apples/pears and having the same problem with the comma. The intent there would clearly be (?:cucumbers|carrots) not (?:cucumbers|carrots,)

(Personally I can't remember when I last used a comma in a Gherkin step, but hopefully it at least illustrates the point. 😀)

[mpkorstanje]

Maybe it's more sensible to have it bound by any non-word character?

So any non-word character is [^a-zA-Z0-9_] which probably isn't what you mean.

There is the P category in unicode for punctuation, and several subcategories to choose from but the Po category that contains the full stop "." also contains the ampersand "&" so that doesn't work either.

There are a few more caveats. But truth be told, I don't think we can sensibly define the bounds of a alternation beyond space in a way that doesn't involve writing our own multi-language table of characters. And even if we could, we'd have to explain it to people. And then we'd lose the advantage of having a simpler form of regular expressions. So if this is what you want, use a regex. :)

https://en.wikipedia.org/wiki/Unicode_character_property#General_Category

https://www.fileformat.info/info/unicode/category/Po/list.htm

[mattwynne]

Maybe it's more sensible to have it bound by any non-word character?

So any non-word character is [^a-zA-Z0-9_] which probably isn't what you mean.

That is exactly what I meant, yes! You seem to be saying that obviously wouldn't work but I can't see why. What did I miss?

[mpkorstanje]

Не все пишут на латыни. Ή αγγλικά για αυτό το θέμα. そして、ガーキンは多くの言語で動作します。Y realmente, incluso el español tiene algunos caracteres que no son palabras en medio de sus palabras.

[mattwynne]

🤦