test(e2e): More e2e test improvements

[knagaitsev]

• This is a bugfix
• This is a feature
• This is a code refactor
• This is a test update
• This is a docs update
• This is a metadata update

For Bugs and Features; did you add new tests?

Yes

Motivation / Use-Case

• There was some problematic logic in ClientOptions.test.js
• Client.test.js and Progress.test.js could potentially write to the same main.css, so I split their tests to use two different configs

Breaking Changes

None

Additional Info

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (master@bbe410e). Click here to learn what that means.
The diff coverage is 50%.

@@            Coverage Diff            @@
##             master    #2163   +/-   ##
=========================================
  Coverage          ?   93.85%           
=========================================
  Files             ?       33           
  Lines             ?     1270           
  Branches          ?      366           
=========================================
  Hits              ?     1192           
  Misses            ?       71           
  Partials          ?        7

Impacted Files	Coverage Δ
client-src/clients/WebsocketClient.js	57.89% <50%> (ø)
client-src/clients/SockJSClient.js	61.53% <50%> (ø)

---

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 bbe410e...0de0d86. Read the comment docs.

[knagaitsev]

/cc @evilebottnawi I don't think just adding an empty client.onerror handler is a breaking change, right? Or is it preventing something from being thrown that would normally get thrown?

Edit: I think it should not effect anything to have an empty handler https://developer.mozilla.org/en-US/docs/Web/API/WebSocket/onerror

I got the idea that this might solve the problem from: websockets/ws#246, websockets/ws#656, but not entirely sure why it would change anything

[alexander-akait]

Code looks good, it is interesting questions, in theory will be better logging error

[knagaitsev]

Page crashes still happening. I've thought about making some mechanism to retry the test if a page crashes, but that might just be sweeping an issue under the rug.

Maybe we should just wait until there is further progress on this puppeteer/puppeteer#1454

[alexander-akait]

Need use logger

[alexander-akait]

We should use logger, instead console

[knagaitsev]

@evilebottnawi This is another case where the path from SockJSClient to utils/log is not the same in the client-src directory compared to the client directory. I think we should wait and add this after reorganizing client directories on next, rather than do more of this redirection:

webpack-dev-server/client-src/default/webpack.config.js

Lines 21 to 31 in c2da532

	new webpack.NormalModuleReplacementPlugin(
	/^\.\/clients\/SockJSClient$/,
	(resource) => {
	if (resource.context.startsWith(process.cwd())) {
	resource.request = resource.request.replace(
	/^\.\/clients\/SockJSClient$/,
	'../clients/SockJSClient'
	);
	}
	}
	),

[knagaitsev]

Let me remove the console.error for now, since it did not exist before and we won't introduce anything new in console output.

[alexander-akait]

/cc @hiroppy