TIL Googlebot is based on Chrome 41

[robzilla]

Googlebot uses a web rendering service (WRS) that is based on Chrome 41 (M41). Generally, WRS supports the same web platform features and capabilities that the Chrome version it uses.

Rendering on Google Search [developers.google.com]

I did not know this, but it's good to keep in mind that certain features that have more recently rolled out to the Chrome browser will not be available to the Google Search crawler. For reference, the subsequent version 42 was released nearly 3 years ago.

For a full list of unsupported features, see caniuse.com [caniuse.com] (bottom of page).

[aristotle]

Isn't that only for the mobile googlebot?

Agent: Mozilla/5.0 (Linux; Android 6.0.1; Nexus 5X Build/MMB29P) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/41.0.2272.96 Mobile Safari/537.36 (compatible; Googlebot/2.1; +http://www.google.com/bot.html)

[robzilla]

They don't specify, so I would assume not. Mobile presumably needs the User Agent because some sites serve a desktop or mobile view based on that. What they call the Web Rendering Service (WRS) may not be a real-time thing either, but there's very little information out here.

[aristotle]

Well I just tested "fetch and render" in GSC on an html5 page from one of my sites which has bordered divisions with rounded corners.

I did the fetch and render for both Desktop and Mobile:Smartphone googlebots. The renders for both of them showed the rounded corners. So the question is -- Would Chrome 41 show rounded corners?

[iamlost]

What is rather strange is that googlebot remains unable to handle HTTP/2 as Chrome v41 is HTTP/1.n (and FTP!) only. It must either (1) be served HTTP/1.1 as a fallback from HTTP/2 (which is what I do) or (2) it uses the outdated, superseded, and vulnerable Next Protocol Navigation aka SPDY on top of HTTP/1.n.

It rather raises the question of just how important 'search' is these days in the Google ecosystem that their main site crawler is not only 25 versions (3-years) 'out of date' but also 2-years behind the 'switch' from Google proprietary SPDY to Google involved/pushed HTTP/2.

[robzilla]

Would Chrome 41 show rounded corners?

If you're using border-radius, then yes, that's been supported since v5.

It rather raises the question of just how important 'search' is these days in the Google ecosystem that their main site crawler is not only 25 versions (3-years) 'out of date' but also 2-years behind the 'switch' from Google proprietary SPDY to Google involved/pushed HTTP/2.

Googlebot still only requests a single resource at a time, so I don't think it would really benefit from those protocol upgrades the way that users are. It may not even use Chrome while crawling (i.e. in real-time), but rather in post-crawl processing to render and analyze a page. It is likely heavily customized code, so it might be difficult keeping Chrome up to date. Most pages should display just fine on a 3-year-old browser.

Added: Actually, it looks like Chrome 41 was the first version [caniuse.com] to support HTTP/2. Googlebot is still HTTP/1.n only, though.

[aristotle]

I still think there may be a rendering difference between mobile googlebot and regular googlebot. If there's no difference, then why does GSC have "fetch and render" options for both of them?

In general, mobile browsers are more "primitive" than desktop browsers. Some of the mobile browsers don't show rounded corners.

[robzilla]

I don't disagree, there probably is. Primarily because the viewport will be different, and a mobile view will be analyzed for things like the proximity of tappable elements, whereas the desktop version will not. You can render both with the same (Chrome 41-based) rendering service, though.