jQuery doesn't filter (and probably everything else) child selector properly

[vidmich]

jQuery('<div><div class="cls1"><div class="cls2"><div><div class="cls2"><div class="cls3"></div></div></div></div></div></div>').find('.cls1 > .cls2 .cls3').length === 1

should be true

having elements with classes .cls1 .cls2 .cls2 .cls3 selector '.cls1 > .cls2 .cls3' returns 0 elements

propably cause jQuery stop searching correct pattern of parents when found that nearset .cls2 parent have no .cls1 parent...

[scott_h]

I've encountered this bug too. Here are my findings:

It looks like >, + and ~ combinators are too pessimistic -- if they fail to match on the first element they're tested against, jQuery will refuse to test any ancestors.

I've created a test case but there are many more subtleties my test case doesn't capture. Consider testing the selector '.a > .b > .c *' against the DOM tree represented by '.a > .b > .b.c > .c > x' -- when '.a' fails to match on .b, it needs to backtrack one element and resume testing from '.c' to find the match. Chains of > and + need to be treated like units.

[dmethvin]

​http://jsfiddle.net/dmethvin/uEL2c/

Must have been fixed at some point.