Specify network body data collection for data: scheme responses

[juliandescottes]

Similar to #985

We should also specify what to do for data URLs, which would currently not trigger the data collection. This response doesn't even have a stream to clone.

[css-meeting-bot]

The Browser Testing and Tools Working Group just discussed Data collection for data scheme.

The full IRC log of that discussion

<AutomatedTester> topic: Data collection for data scheme
<AutomatedTester> github: https://github.com//issues/986
<AutomatedTester> jdescottes: this is the same topic as the previous... we don't have a stream we can clone
<blaze_apple> q+
<AutomatedTester> ... we've updated Firefox to be similar to chrome. Are people happy for me to return a response body for data urls?
<AutomatedTester> ack next
<sadym> q+
<AutomatedTester> blaze_apple: In general this is fine. What do we do if the page is navigated away... e.g. if the process dies or ...
<AutomatedTester> ack next
<AutomatedTester> jdescottes: we don't clean collected data. We currently do this with a queue
<AutomatedTester> ... and we can handle the situation when collectors are cleaned up
<AutomatedTester> ... but we don't do anything for navigation
<AutomatedTester> sadym: I wonder what selenium thinks about this?
<jimevans> q+
<AutomatedTester> ack next
<AutomatedTester> jimevans: how does selenium feel about cached responses or when the navigation has happened?
<AutomatedTester> sadym: about data schema
<AutomatedTester> jimevans: I am of the opinion that we are doing network interception or of data
<AutomatedTester> ... whether it comes over any scheme we should be able to retrieve that data
<AutomatedTester> ... and I expect selenium users to think the same
<AutomatedTester> q?

[juliandescottes]

For info, I have wdspec tests in progress/review at https://phabricator.services.mozilla.com/D267405 for this.

What is left is to check how we need to update the spec to ensure we can handle data scheme.

[juliandescottes]

This response doesn't even have a stream to clone.

I think I was misreading the spec here. The body of a data: scheme request is set as

body is dataURLStruct’s body as a body.

as a body is the result of safely extracting bytes, which does create a stream made of the byte sequence found in the URL (https://fetch.spec.whatwg.org/#concept-bodyinit-extract)

Meaning the spec as is normally fine (minus the fact that we are missing the integration with fetch atm).

Going to close this one since we should have tests landed via https://phabricator.services.mozilla.com/D267405