Bug 54556 – [Meta] Simplify the UA string

• Status: RESOLVED
• Resolution: FIXED
• Component: Platform

[Laszlo Gombos]

Following Gecko (http://hacks.mozilla.org/2010/09/final-user-agent-string-for-firefox-4/) and others WebKit should consider simplifying the UA string in concord with other WebKit based ports/browsers/products.

[Alexey Proskuryakov]

An obvious concern with changes like this is that sites will certainly break. Scripts that check navigator.userAgent do that in most weird ways.

For example, there used to be sites that broke if "4." was found _anywhere_ in the string, because they assumed running under Netscape 4 then (or IE 4? I don't remember all details).

[Peter Kasting]

I'm sure there will be some bustage.  Hopefully the mitigating factors are:
* Chromium has the ability to push changes to smaller populations and get real-world feedback quickly, so it would be useful for us to test this first
* If we make the same changes as Gecko and at the same time, site authors are more likely to see bustage and have more incentive to fix it correctly

We've tested UA string simplifications before so we may have some QA experience or tools from the last time we tried.  Darin Fisher might know more.

[Peter Kasting]

At this point all the UA string changes I intend to make have landed.  I'm going on vacation tomorrow, but I've sent draft blog posts to some of the Chromium folks that can post on both Chromium and WebKit blogs, in order to provide some clear, public documentation of what's changed.

There's been some bustage on Google sites, but we fixed the problems in Google's UA parsing libraries and at least some sites have pushed out updates to fix things.  Beyond that, I haven't heard much fallout so far.