Basically, I'd need an API to fulfill the mozci contract defined at https://github.com/mozilla/mozci/blob/450c4e2c7e27b89ad3deadc295a1173894890a17/mozci/data/contract.py#L21-L46.

n-i'ing myself to remind me to take a look at this soon.

I've created a PR to add this API. It looks like:

GET /api/project/try/note/push_notes/?revision=9dd0069768515991767c2be278bf4142c8463a30

[
{
"id": 2302245,
"job": {
"task_id": "E5Cpd7V4TiGzXz59OiQ8lQ",
"job_type_name": "test-linux1804-64-qr/debug-web-platform-tests-reftest-fis-e10s-2",
"result": "testfailed",
"duration": 17
},
"failure_classification_name": "intermittent",
"created": "2020-10-08T15:19:38.879261",
"who": "cdawson@mozilla.com",
"text": "camtest 1"
},
{
"id": 2302246,
"job": {
"task_id": "NRKG0BXLSaesroAhFr2zfw",
"job_type_name": "test-macosx1014-64/debug-web-platform-tests-reftest-e10s-3",
"result": "testfailed",
"duration": 21
},
"failure_classification_name": "intermittent",
"created": "2020-10-08T15:19:55.460807",
"who": "cdawson@mozilla.com",
"text": "camtest 2"
}
]

This API appears to only return jobs that have a classification, which will cause us to miss things. We could combine this API alongside the taskcluster data source and then re-write the contracts so that we get result and duration there.

I think that would make this Treeherder API be a bit cleaner anyway (i.e, if it could drop the result and duration keys. Then it is only dealing with classifications.

Removing myself so another can take this. With the sprint cadence, I won't necessarily be the one to work on this.