identify webviews

[stevesouders]

Hi. I'm doing a deep dive on the performance of webviews. I'd love to get RUM metrics broken out by native mobile browser vs mobile webview, but it looks like ua-parser doesn't disambiguate these. For example, here's the UA for Facebook's iOS webview:

Mozilla/5.0 (iPhone; CPU iPhone OS 7_1_1 like Mac OS X) AppleWebKit/537.51.2 (KHTML, like Gecko) Mobile/11D201 [FBAN/FBIOS;FBAV/12.1.0.24.20;FBBV/3214247;FBDV/iPhone6,1;FBMD/iPhone;FBSN/iPhone OS;FBSV/7.1.1;FBSS/2; FBCR/AT&T;FBID/phone;FBLC/en_US;FBOP/5]

Ua-parser labels it "Mobile Safari" without mentioning webview. At a minimum I'd like to see "Mobile Safari webview" and potentially "Mobile Safari webview Facebook".

What do people think of this idea? Should I just start contributing regexes?

[elsigh]

Do you know if FB (or anyone) can make a functional change to the webview? Though, I suppose that's actually what you're trying to check for, yeah?

If not I wonder if this is really a different UA, or more like OS - ie. some sort of container/annotation for a particular UA. In that case, I wonder if it might be better to either add an annotation to the OS object or device object in this case.

I agree that it would be good to be able to detect the webview case, and also to be able to mark popular webview strings somehow (twitter, fb, etc...)

[stevesouders]

Webviews are (notoriously) worse than their native browser counterpart. For example, in iOS 7 you get "mobile safari" for both a webview and mobile safari UA, but the webview does not have Nitro, IndexedDb, etc. So I'd like to separate that out in my metrics. It's also interesting to know how many people are visiting my website from a webview (Chrome for iOS, Facebook, Twitter) vs the mobile browser.

So it's not about whether FB can change the webview. Instead, it's about distinguishing webview from mobile browser traffic, and potentially slicing the webview traffic further to see the sources (FB, Chrome for iOS, Twitter, etc.).

Does that help pinpoint where to add the info?

[elsigh]

Yep - I totally agree that the string "webview" ought to be a part of the ua.family (i.e. Safari (webview)) to disambiguate it from normal Safari (assuming we can detect that difference from the UA string).

It's the latter part about the further-modified-by-distributor string (FB, Twitter) and I'm not sure where we should put that.

So I'm cool with family becoming Safari (Webview) and Chrome (Webview) - I think submitting a regex change PR is the way to go for this.

Looks like someone put up a cool site to show what ua-parser currently determines the UA to be:
http://www.whatsmyua.info/

[tobie]

Sounds good.

[broggeri]

Hello, is someone working on this ? I might try to take a stab at it.

[tobie]

Not that I know of. See related thread on the mailing list: https://groups.google.com/d/msg/ua-parser/clM4un5jgY4/b9alPXlZ6zYJ

[Ironholds]

Also note that ua-parser is deprecated as a place to throw new changes.