Open yoavweiss opened 1 year ago
After looking at the examples again, I'm less confident that it's a false positive:
So it seems like it may be reasonable to consider this a soft navigation to a similar (yet different) product page.
^^ @anniesullie
I agree. To me there's a user interaction, a state change, a UI change, and a URL change. That to me seems like it should be a soft nav.
The only concern is if there is not a navigation history. On the Walmart example, for example, clicking back does not take me to the previous option selection.
So, rather than a URL update, should we insist on a URL pushState
and exclude replaceState
? Perhaps that should be a separate issue though...
So it seems like it may be reasonable to consider this a soft navigation to a similar (yet different) product page
Would love to collect a list of issues like this where it's a bit subjective whether it's a navigation and hear working group feedback!
The reliance on contentful paints should potentially help here. If we see a bunch of cases where the change is "too small" to be considered a navigation, we could add a limit based on the %age of viewport changed. However, experiments with that approach did not prove particularly useful.
crbug/1371927 outlines a few examples where the heuristic's result is arguably a false positive: product options where the user clicks to e.g. change the item's color result in a URL change and a DOM modification, and hence are considered a soft navigation.
More research is needed to see if this can be avoided.