Avoid? No. Just use them with double backslash in RegExp constructor.

So, are you saying that we should avoid All the built-in character set such as \d, \w, \s, etc? \d is not the only char set that have a backslash (escaped). How about other escaped characters that we can't construct manually such as word boundaries? (\b & \B) ?

Agree, this does not make sense. How can this tag be removed?

Same for Javascript, but only if we use new RegExp(regString) constructor.
We must use \\d instead of \d. Or [0-9]. Or String.raw`\d`

Apparently it's advisable to not use /d if you can, since it may cause "escaping issues in some languages (like Java)."

https://stackoverflow.com/a/12643073

(However much that's worth, just putting it out there)

Why this tagged as "Regular Expressions"?

(!?1) !== (?!1).
(!?1) mean "zero or one !" then "1". So regexp can be rewritten to /0*1/g
But this solutions works despite this mistake.

Still incorrect expected output for randomised test in JS version of kata sometimes when we have 0 koef in input.

For example,

Input +1za+5z-5azb+1za+8z-12z+8zab-2az-3baz
Expected output z+0abz
But should be z