Avoid preloading IE-conditional shims

[aduth]

Why: As of version 6.1.0, Rails automatically adds preload link headers to HTTP responses for every javascript_include_tag and stylesheet_include_tag. This causes modern browsers to download these script tags. However, we include a few script tags specifically for older browsers, using Internet Explorer conditional comments. These will never actually load in modern browsers. Some browsers (Chrome) even emit warnings to this effect:

"The resource https://secure.login.gov/assets/respond.min.js was preloaded using link preload but not used within a few seconds from the window's load event. Please make sure it has an appropriate as value and it is preloaded intentionally."

History:

• Introduced in Rails 6.1.0 (Automatically set Link header for each stylesheet and script rails/rails#39939)
• Improved configurability in Rails 6.1.1 (Add config.action_view.preload_links_header option rails/rails#40882)
  • The pull request comment describes the exact use-case being addressed here:

    • However not everything should be preloaded, e.g. a link to a legacy IE stylesheet has no need to be preloaded because IE doesn’t support the header and in some browsers it will trigger the preload even though it’s not used since it’s inside an IE conditional comment. This leads to increased bandwith costs and slower application performance.

Open questions:

• Should we drop support for Internet Explorer 9 and below? According to @mitchellhenke , analytics reports Internet Explorer <= 9 usage represents ~0.000575% of all traffic.

Remaining tasks:

• Unit tests, pending decision above

[mitchellhenke]

Voting to drop support given USWDS does not explicitly support these version of IE: https://designsystem.digital.gov/documentation/developers/

Currently, this means that the design system version 2.0 supports the newest versions of Chrome, Firefox, Safari, and Internet Explorer 11 and up.

[mitchellhenke]

Thanks 😀

[zachmargolis]

Can we add a request spec spec/requests that loads a page and inspects the headers?

[zachmargolis]

LGTM

[aduth]

• Should we drop support for Internet Explorer 9 and below? According to @mitchellhenke , analytics reports Internet Explorer <= 9 usage represents ~0.000575% of all traffic.

On further discussion in Slack, there's a reluctance to drop support for Internet Explorer 10 at this point. Since these shims are only actually relevant for IE8, we may still be justified in dropping them specifically. That being said, since there's a plan to follow-up with partners to better understand browser usage, it could be best to wait 'til that's completed. I'll plan to merge this as-is.