tests: update many test traces, support .json.gz#16007
Conversation
| "optimistic": 2538, | ||
| "pessimistic": 2820, | ||
| "timing": 2679, |
There was a problem hiding this comment.
Would it be better to use a more controlled trace to replace this? I don't envy the next person making changes to the simulator having to manually determine if any changes to these numbers are reasonable from a massive cnn trace.
Alternatively, if it needs to be more complex than what's in our smoke tests, in #14525 we had discussed using the test sites in https://github.com/ChromeDevTools/performance-stories as the source of generating test artifacts. Controlled, intentional, but also a decent variety of scenarios.
There was a problem hiding this comment.
I don't see any story there that has much JS work going on.
I can switch to something else. intercom-widget has a value here. But I don't know why for observed I'm getting this:
1) Metrics: TotalBlockingTime
should compute an observed value:
LighthouseError: NO_TTI_CPU_IDLE_PERIOD
at Interactive.findOverlappingQuietPeriods (file:///Users/cjamcl/src/lighthouse/core/computed/metrics/interactive.js:134:11)
at Interactive.computeObservedMetric (file:///Users/cjamcl/src/lighthouse/core/computed/metrics/interactive.js:164:41)
at Interactive.compute_ (file:///Users/cjamcl/src/lighthouse/core/computed/metrics/metric.js:91:21)
I wonder if user flows are ending too early?
There was a problem hiding this comment.
@paulirish and I investigated the NO_TTI error and found an unrelated issue
Updates many old test traces. Also adds
.json.gzsupport toasset-saver.js. I wanted to add fixtures for cnn.com, but they were far too big to commit without compression.ref #15841