You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Many lazyloading scripts try to compensate by sniffing the UA (including the included example). However it's not clear if the result of the dom manipulation (say swapping data-src to src) is indeed what the google grabs.
I understand it's chrome, not the google - but do you have any insight or knowledge of how we should interpret the results to what the google sees? Or this just a test of the events and not indicative of the googles view?
It (purposefully?) does not set the UA to a googlebot, so I'm guessing there's an interpretation we should be taking away from it.
The text was updated successfully, but these errors were encountered:
is the lazy load example script: https://github.com/GoogleChromeLabs/puppeteer-examples/blob/master/lazyimages_without_scroll_events.js supposed provide an accurate "view" of what the googlebot (image or search) sees -- visually and dom-wise?
Many lazyloading scripts try to compensate by sniffing the UA (including the included example). However it's not clear if the result of the dom manipulation (say swapping data-src to src) is indeed what the google grabs.
I understand it's chrome, not the google - but do you have any insight or knowledge of how we should interpret the results to what the google sees? Or this just a test of the events and not indicative of the googles view?
It (purposefully?) does not set the UA to a googlebot, so I'm guessing there's an interpretation we should be taking away from it.
The text was updated successfully, but these errors were encountered: