core: add hidden audits for each insight #16312
Conversation
connorjclark
changed the title
| @@ -44,7 +44,7 @@ function buildStandaloneReport() { | |||
| outfile: 'dist/report/standalone.js', | |||
| format: 'iife', | |||
| bundle: true, | |||
| minify: true, | |||
There was a problem hiding this comment.
drive by. this is very useful. surprised we didn't already do this.
| const navigationId = traceEngineResult.data.Meta.mainFrameNavigations[0].args.data?.navigationId; | ||
| if (!navigationId) { | ||
| throw new Error(`Lantern metrics could not be calculated due to missing navigation id`); | ||
| } |
There was a problem hiding this comment.
shifts in types with trace engine...
| * @param {(obj: Record<string, string>, key: string) => void} cb | ||
| * @param {Set<object>} seen | ||
| */ | ||
| function recursiveObjectEnumerate(obj, cb, seen) { |
There was a problem hiding this comment.
I think there is some risk that we will resolve node details for a bunch of nodes we don't necessarily need. An insight might hold on to a trace event with a nodeId that never gets used in one of the insights.
I guess the alternative is to manually curate the list of nodeIds but then we would potentially miss some. False positives > false negatives so this seems good.
| // Can only resolve elements for the latest insight set, which should correspond | ||
| // to the current frame id. Can't resolve elements for pages that are gone. | ||
| const insightSet = [...traceEngineResult.insights.values()] | ||
| .reverse().find(insightSet => insightSet.frameId === mainFrameId); |
There was a problem hiding this comment.
Pretty sure this will always be the last insight set in the list. The main frame ID does not change across navigations.
There was a problem hiding this comment.
This should actually just grab the last one, and then check if it is valid by looking at navigation id. updated.
| */ | ||
| static async audit(artifacts, context) { | ||
| return adaptInsightToAuditProduct(artifacts, context, 'DOMSize', (insight) => { | ||
| if (!insight.maxDOMStats?.args.data.maxChildren || !insight.maxDOMStats?.args.data.maxDepth) { |
There was a problem hiding this comment.
Technically we will still have a valid totalElements value even if these are missing. In practice though, this will only happen if the document element is removed so it's an edge case that really doesn't matter.
adaptInsightToAuditProducthelper, which is given the insight model and the caller returns aLH.Audit.Productyarn generate-insight-auditsscript, to quickly make and configure new audits (just run when new insights exists, no flags). Calling in CI so we don't forget to add a new insight when upgrading trace enginedom-size-insightandviewport-insighthave been implemented. The other 12 are blank implementationstrace-engine, which grabs all the node ids referenced in insights and collects their node information