> Actually it's much easier to check for a single "version number" rather then try to parse large and complex "version string" thus it'll make life of web developers easier, not harder.
Wasn't there talk of not sending the version number in UA strings at all?
> And this is why I think UA check is the only way to go. It's one thing to support few different browsers with a sane set of features, it's another to use application platform which can have important parts just randomly removed.
If that's the case, then UA strings need something along the lines of "I'm actually FooBrowser, but you can treat me like Chrome" so that new/other browsers have a *chance* of being recognized in stats. Without this or feature detection, first-party support will be nigh impossible for new/alternate browsers because "no one uses it" when it really just has to masquerade as anything but itself to even work. This is basically vendor lockout and I don't see why it should be accepted here more than anywhere else.
> This may be not all that friendly to the DIY guys, but at least this way you'll know that you can safely use local storage, for example (all major browsers support it for many years!) and now worry about fallbacks.
Good for them. However, until there's support for blacklisting random sites from dropping whatever data they like onto my machine, it gets disabled completely. Sites already have to cope with the chance that cookies don't work and if they encounter that, they put up a page saying "This site needs cookies to do random stuff, unblock us so we can do A, B, and C.", why not do the same with other features?