fix: pin strip-ansi to 3.x for ES5 compat

[yyx990803]

• This is a bugfix

Follow up for #1270 (comment)

strip-ansi which is required by the clients, uses ES6 syntax. Pinning it to 3.x keeps ES5 compat. (The only difference between 3.x and 4.0 is the syntax change)

[codecov]

Codecov Report

Merging #1273 into master will not change coverage.
The diff coverage is n/a.

@@           Coverage Diff           @@
##           master    #1273   +/-   ##
=======================================
  Coverage   75.72%   75.72%           
=======================================
  Files           5        5           
  Lines         482      482           
  Branches      156      156           
=======================================
  Hits          365      365           
  Misses        117      117

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 8c1ed7a...b78e249. Read the comment docs.

[graingert]

@sokra if you don't want to play whack-a-mole with deps as the npm ecosystem moves to releasing greenfield JS you should have a look at #1279

[julianxhokaxhiu]

Are you planning on merging and releasing this ASAP? I am also impacted by the same identical issue at the moment.

[graingert]

@julianxhokaxhiu FYI, i've released drop-in-replacement @procensus/webpack-dev-server with my fix in #1279

[julianxhokaxhiu]

Thanks, I'll use that temporary until this gets merged :)

[graingert]

this gets merged :)

Hopefully it won't! ;) publishing `ES${new @Date().getFullYear()}` to npm is the future. See #1279

[julianxhokaxhiu]

I tested your drop-in replacement and I still get the same error:

Syntax error on L2114:
module.exports = input => typeof input === 'string' ? input.replace(ansiRegex(), '') : input;

it seems that the ansi-module is not transpiled from arrow functions.

//EDIT: See also vercel/next.js#2747

The solution there was to downgrade strip-ansi: https://github.com/zeit/next.js/pull/2860/commits

So this PR is really needed.

[graingert]

@julianxhokaxhiu replied here: #1279 (comment)

[yyx990803]

@graingert FYI #1279 does not fix CLI --inline usage because injected entries will be processed with user config.

[TheLarkInn]

@yyx990803 I think this looks fine.

[TheLarkInn]

Would someone like to submit an issue so that we can upgrade in the future but not break IE? Pinning versions is a sort of a bandaid IMO.

[TheLarkInn]

Nevermind, I got it tracked in issues #1286

[TheLarkInn]

I've got limited time today to publish this, however I will see if I can get @SpaceK33z or @sokra to run a release for me on this.