EscapeRegExpPattern: references to Pattern omit setting for N

[jmdyck]

In the abstract operation EscapeRegExpPattern, step 1 says

Let _S_ be a String in the form of a |Pattern[~U]| (|Pattern[+U]| if _F_ contains `"u"`)  ...

However, since "RegExp named capture groups" was merged (95ec0c6), Pattern now takes an additional grammatical parameter, N. Thus, the above references to Pattern must be updated, and possibly the logic extended to decide between ~N and +N.

(attn @mathiasbynens)

[jmdyck]

I think it might suffice to always use +N. A Pattern can't always be parsed with +N, but there's always a +N form that's equivalent (right?), which is all that's required of _S_.

On the other hand, what do implementations do?

[bakkot]

On the other hand, what do implementations do?

$ eshost -e '/\k<>/.source'
#### Chakra
\k<>

#### JavaScriptCore
\k<>

#### node
\k<>

#### SpiderMonkey
\k<>

#### V8
\k<>

#### V8 --harmony
\k<>

#### XS
\k<>

They seem to preserve useless escapes, or at least this particular useless escape, which means that they can produce strings which do not match Pattern_[~U, +N].

[jmdyck]

Hm, that's going to make it tougher to match web reality.