`prefer-lookaround`: Report cases that replacement repeating pattern

[fisker]

Should we prefer

'a'.replace(/(?<=a)/, '-')

over

'a'.replace(/a/, 'a-')

?

I'm not sure about this. Since it may make code unreadable.

'Hi'.replace(/Hi/, 'Hello') // Good
'Hi'.replace(/(?<=H)i/, 'ello') // Really BAD

[ota-meshi]

Thank you for posting the issue. Hmm🤔. I'm not sure the prefer-lookaround rule should report that far. As you wrote Hi->Hello is easier to read the first one.

[RunDevelopment]

My thoughts:

• In the first example, I prefer 'a'.replace(/a/, 'a-') because it's simpler.
• The only advantage of this transformation seems to be that it reduced redundancy.
• This makes regexes more complex. Using a lookbehind to reduce some redundancy is a clever trick, but it's also not trivial to understand.
• The prefix needs to be at least 6 characters long to produce shorter source code.
• The resulting regexes might be harder to change. Whether you can do this trick depends on the pattern, so you might need to undo the trick to change the regex.
• In the first example, I prefer 'a'.replace(/a/, 'a-') because it's simpler.

The downsides outweigh the upsides, IMO.