blocking="render" attribute on <link>, <script> and <style>

[xiaochengh]

Request for Mozilla Position on an Emerging Web Specification

• Specification Title: blocking="render" attribute on <link>, <script> and <style>
• Specification or proposal URL:
  • Full proposal: https://gist.github.com/xiaochengh/9404abbecdc3b45c0e9f3d5e99fbc65d
  • WHATWG discussion: Feature proposal: blocking="render" attribute on <link>, <script> and <style> whatwg/html#7131
• Caniuse.com URL (optional):
• Bugzilla URL (optional):
• Mozillians who can provide input (optional):

Other information

This proposal also includes the standardization of a general render-blocking mechanism, which will standardize the existing stylesheet loading behaviors and support the proposed attribute. The proposed behavior slightly differs from Firefox's current behavior: the proposal does not allow requestAnimationFrame() callbacks to fire until all render-blocking stylesheets are loaded, while Firefox currently allows it.

[smaug----]

Well, UA can always stop the blocking. Otherwise there would be security/privacy issues.
But yes, we could stop calling rAF. Though, need to test how resize (and other) events behave in other implementation while rendering is blocked.

I think in general the idea is quite reasonable.
I filed https://bugzilla.mozilla.org/show_bug.cgi?id=1751383 just in case we want to implement this.

@emilio

[khushalsagar]

But yes, we could stop calling rAF. Though, need to test how resize (and other) events behave in other implementation while rendering is blocked.

@smaug---- can you clarify the question. Did you mean the impact of resizing the window? As per the spec text here there's no rendering opportunities while we're blocked. So no events which happen during the rendering lifecycle run.

[noamr]

Intending to amend the current spec for render-blocking scripts to also include inline scripts, see whatwg/html#10035

[zcorpan]

I think this is reasonable and useful. My main concern is that it might be a bit too easy to block rendering on a defer (or module which is defer by default) script, and thus block until the timeout hits or until the full document is parsed. I'll file a spec issue.

[zcorpan]

Filed whatwg/html#10068

I think that issue can be resolved without changes to how the feature works.

Suggest positive.