devinrhode2/UA-Sniffing-FTW

## UA-Sniffing-FTW
The internet abroad doesn't believe in UA sniffing. Maybe it
complicates things too much. I believe we can have a hybrid.

I thought hard about this. What the hell is so wrong about UA sniffing?

Ok, what if we don't have a user agent? I almost took the excuse:
"well, 90% of the time we'll have a UA (if not 100% of the time)
so, why worry about the 10%?"
EVERYBODY deserves the best experience their browser is capable of.

If they don't have a userAgent header, we can delay loading the majority
of resources for the page, and in javascript get the user-agent from navigator.userAgent

But I'm skipping over myself. The current argument against UA sniffing,
as I understand it, is that you make assumptions about what code will work.

This doesn't have to be, we can have a hybrid of the approach Modernizr takes.

Feature detection. It's the most reliable way to go.

**The proposal**
Run feature detects, and find out what shims/polyfills would be needed for various things.

Then, read navigator.userAgent, and send this shim list for this UserAgent
 to a central database. Next time the server gets a request from this
exact UserAgent, it ALREADY knows what shims that browser needs, and
can also skip serving the feature detection tests.

Discuss on Hacker News: http://news.ycombinator.org/item?id=3816152 (or vote up)
Retweet and share if you support this idea! Holla @paul_irish and @modernizr if you feelz the need
	The internet abroad doesn't believe in UA sniffing. Maybe it
	complicates things too much. I believe we can have a hybrid.

	I thought hard about this. What the hell is so wrong about UA sniffing?

	Ok, what if we don't have a user agent? I almost took the excuse:
	"well, 90% of the time we'll have a UA (if not 100% of the time)
	so, why worry about the 10%?"
	EVERYBODY deserves the best experience their browser is capable of.

	If they don't have a userAgent header, we can delay loading the majority
	of resources for the page, and in javascript get the user-agent from navigator.userAgent

	But I'm skipping over myself. The current argument against UA sniffing,
	as I understand it, is that you make assumptions about what code will work.

	This doesn't have to be, we can have a hybrid of the approach Modernizr takes.

	Feature detection. It's the most reliable way to go.

	The proposal
	Run feature detects, and find out what shims/polyfills would be needed for various things.

	Then, read navigator.userAgent, and send this shim list for this UserAgent
	to a central database. Next time the server gets a request from this
	exact UserAgent, it ALREADY knows what shims that browser needs, and
	can also skip serving the feature detection tests.

	Discuss on Hacker News: http://news.ycombinator.org/item?id=3816152 (or vote up)
	Retweet and share if you support this idea! Holla @paul_irish and @modernizr if you feelz the need