This test has started to consistently return values between 90-100 in CI, so lets lower it.
I investigated a bit, and something that sticks out is that wastedMs (aka download time, according to the comment) being used here did not match the ?delay param at all. This is after simulating the fcp graph, but I'd expect the download times to be longer not shorter. The headers start timing on the trace event contains the delay as expected, so perhaps we are building the lantern graph incorrectly.
debug the artifacts yourself - the http://localhost:10200/dobetterweb/dbw_tester.css?delay=2200 is one request that has a wastedMs of much less than its delay:
This test has started to consistently return values between 90-100 in CI, so lets lower it.
I investigated a bit, and something that sticks out is that
wastedMs(aka download time, according to the comment) being used here did not match the?delayparam at all. This is after simulating the fcp graph, but I'd expect the download times to be longer not shorter. The headers start timing on the trace event contains the delay as expected, so perhaps we are building the lantern graph incorrectly.debug the artifacts yourself - the
http://localhost:10200/dobetterweb/dbw_tester.css?delay=2200is one request that has a wastedMs of much less than its delay:https://github.com/GoogleChrome/lighthouse/actions/runs/9169845726/artifacts/1522909511
node cli "-A=/Users/cjamcl/Downloads/Smokehouse (ubuntu; chrome ToT)/2/dbw" http://localhost:10200/dobetterweb/dbw_tester.html