Jason Cheatham

This is related to (or at least touches a lot of the same code as) #1064.

The frequency of timeouts when connected to BrowserStack has gotten a bit silly at this point.

A potential fix has been pushed in 7b7b110. This commit adds retry functionality for all requests made from Node. Only two classes of failures lead to retries: ETIMEDOUT (a TCP...

This is in Intern master, but hasn't yet been backported to Intern 4.x (`@theintern/common`, specifically).

A fix for this should also be back ported to https://github.com/theintern/common.

`@theintern/[email protected]` has been published with a possible fix. To test it, `npm install @theintern/[email protected]` and manually remove `node_modules/intern/node_modules/@theintern/common`, then try running tests.

This is likely related to theintern/digdug#65

Using Intern 4.8, common 0.2.3, and mitmproxy running on http://localhost:8080, test runs fail with ``` Error: Download server returned status code 400 for https://chromedriver.storage.googleapis.com/79.0.3945.36/chromedriver_mac64.zip @ node_modules/src/Tunnel.ts:400:21 @ node_modules/@theintern/common/index.js:670:29 at processTicksAndRejections...

Or that could just be a [problem with mitm proxy](https://github.com/mitmproxy/mitmproxy/issues/848). Proxies are kind of a pain. 😄

Well, after making a small modification to mitmproxy (it's a bit picky about URL schemas) by commenting out [this block](https://github.com/mitmproxy/mitmproxy/blob/fbbbbb678e77e3cff5ab35b1fd55e85e875ed3f5/mitmproxy/proxy/protocol/http.py#L138), connections to https addresses go through when using `@theintern/[email protected]`.