It seems that we have the option to just support OnStart/StopRequest off the main thread, or also add support for AsyncOpen (which would likely require supporting redirects) OMT.

We should try to collect telemetry for how long we spend waiting on the main thread to be available before running each of these steps, to get an idea of the relative value of supporting them OMT.

I think we can limit this just to preloading image requests from the html5 parser, just as a potential consumer we could target with OMT work.

Comment on attachment 9224160 [details]
request.md

DATA COLLECTION REVIEW RESPONSE:

Is there or will there be documentation that describes the schema for the ultimate data set available publicly, complete and accurate?

Yes.

Is there a control mechanism that allows the user to turn the data collection on and off?

Yes. This collection is Telemetry so can be controlled through Firefox's Preferences.

If the request is for permanent data collection, is there someone who will monitor the data over time?

No. This collection will expire in Firefox 96.

Using the category system of data types on the Mozilla wiki, what collection type of data do the requested measurements fall under?

Category 1, Technical.

Is the data collection request for default-on or default-off?

Default on for Nightly only.

Does the instrumentation include the addition of any new identifiers?

No.

Is the data collection covered by the existing Firefox privacy notice?

Yes.

Does the data collection use a third-party collection tool?

No.

Result: datareview+

Data from testing a few sites locally looks like this:

DOCUMENT_PRELOAD_IMAGE_ASYNCOPEN_DELAY
89 samples, average = 7.8, sum = 694

 0 |#########################  14  16%
 1 |#######################  13  15%
 2 |##############  8  9%
 3 |################  9  10%
 4 |#####  3  3%
 5 |#####  3  3%
 6 |#######  4  4%
 7 |#########  5  6%
 8 |##################  10  11%
 9 |####################  11  12%
11 |####  2  2%
47 |#############  7  8%
55 |  0  0%

HTTP_PRELOAD_IMAGE_STARTREQUEST_DELAY
88 samples, average = 55.7, sum = 4,898

  0 |#########################  36  41%
  1 |###  4  5%
  2 |#######  10  11%
  3 |#  2  2%
  4 |#  1  1%
  5 |#  2  2%
  7 |###  5  6%
  8 |#  2  2%
  9 |##  3  3%
 11 |#  1  1%
 13 |#  2  2%
 29 |#  1  1%
 55 |#  1  1%
103 |#  1  1%
120 |#  2  2%
140 |###  5  6%
164 |#  1  1%
192 |#  1  1%
357 |######  8  9%
417 |  0  0%

That suggests that there's significantly more delays waiting to process OnStartRequest than waiting to open the channel initially. Will update with results from the Nightly population when they're available.

https://hg.mozilla.org/mozilla-central/rev/4d9de2cde07d