Average Scroll Depth Metric: extracted tracker changes #4826
Conversation
RobertJoonas
changed the title
RobertJoonas
changed the title
| // also accounted for. | ||
| var count = 0 | ||
| var interval = setInterval(function () { | ||
| currentDocumentHeight = getDocumentHeight() |
There was a problem hiding this comment.
I wonder if we should also update maxScrollDepthPx here as well to account for any weird reflows.
There was a problem hiding this comment.
Makes sense, but it's tricky to optimise our script for this - a reflow might both increase and decrease maxScrollDepthPx (meaning that we would have to keep re-initialising it), but the user can already scroll during those three seconds, so that information might get lost when we blindly reset it.
I would leave it be for now. In worst case, the maxScrollDepthPx will be smaller than it really is, resulting in a lower percentage. But even the very first viewport height after DOMContentLoaded should be a sensible initial value for maxScrollDepthPx. If it turns out to be an issue, we can come back to it later.
| await page.evaluate(() => window.scrollBy(0, document.body.scrollHeight)) | ||
|
|
||
| // Wait until documentHeight gets increased by the fixture JS | ||
| await page.waitForSelector('#more-content') |
There was a problem hiding this comment.
Note: It might have been simpler to evaluate JS on the page appending a div/img rather than rely on timers which might be slightly finnicky.
There was a problem hiding this comment.
Good note! The timeout in the test fixture was intended only for debugging. I've removed it now. It now actually works by simply appending a 2000px div to the document.
Changes
This PR extracts the tracker changes from #4791, and addresses tracker related feedback as well.
Tests
Changelog
Documentation
Dark mode