Only assign className if changed in addClass and removeClass

[jdunck]

This aids performance in some browsers (where adding/removing a
class which is already present triggers a CSS render anyway).

[jdunck]

For a bit more context: on Causes.com, we were using addClass on body, often assigning a class that already existed. On mobile, this needlessly triggers re-render and cost about 200ms. (I'm unsure which mobile browser; I'd need to ask a coworker for that detail.)

[rwaldron]

Thanks for taking the time to contribute to the jQuery project! Please make sure that your commit message cites a valid bug and that you've signed the CLA.

• CLA
• Bugs
• Style Guide
• Won't Fix

[dmethvin]

I don't see a ticket for this but it seems like a good idea. Please make a new ticket (unless you find one) and mention http://bugs.jquery.com/ticket/14164 which is the umbrella for our layout reduction effort.

[jdunck]

I just signed the CLA. I did not find any existing ticket, so created http://bugs.jquery.com/ticket/14250 and referred to 14164.

[timmywil]

👍

[mgol]

@jdunck You need to sign the CLA with the e-mail address you used for commiting; you chose a different one so we need to wait until you re-sign the CLA with the correct mail address.

[jdunck]

I just signed the CLA with the commit email as well.

[rwaldron]

Thanks @jdunck!

[mgol]

Landed at c418b94 for master and at 7dfe0ad for 1.x-master.

[richardscarrott]

@jdunck - I'd be really interested to know which browser(s) triggered a layout when assigning a className that already existed?

[dmethvin]

Chrome, for one. https://code.google.com/p/chromium/issues/detail?id=278045

[richardscarrott]

@dmethvin - Ah, it causes styles to be recalculated rather an actual layout; good to see this enhancement in!

[dmethvin]

Good point, for that particular case it causes a style recalc, which may be forced if later code queries styles.