http://jsfiddle.net/qD5Se/show/

At the above jsFiddle example, we have extremely simple markup:

<input tabindex="0" name="Username" />
<input tabindex="1" name="Password" />
<input type="hidden" name="Hidden" value="value" />
<input tabindex="2" type="submit" value="Submit" />
<a href="#" id="ForgotPassword">Forgot Password?</a>​

When I do the following jQuery selectors in the JavaScript console of Chrome Dev Tools or Firebug...

$('[tabindex="0"]')
$('[tabindex="0"]:visible')

...I get confusing results. The first selector does as I intended (i.e., select the single html element on the page that has an attribute of

tabindex

with a literal value of

0

, which is the

Username

input text field). I would then expect the second selector to produce the same result, as it has the same selector but with one extra restriction that the element be visible. However, I actually get more results than the first, despite the more restrictive selector (i.e., it matches

Username

text field but also matches the

ForgotPassword

anchor link as well).

It appears what happens is that when the

:visible

selector is added, the attribute selector is not doing a literal match anymore. When I do the same attribute selector, but without quotes...

$('[tabindex=0]')
$('[tabindex=0]:visible')

...I now get all elements which effectively have a tab index of zero by default, not just the one that literally has the attribute in its markup (i.e., now we match the

Username

text field and the

ForgotPassword

link, and additionally the hidden input field in the case of the selector without the visibility check).

What I'm gathering is that when the

:visible

selector is appended to the attribute selector of

[tabindex="0"]

, it switches to behaving like the

[tabindex=0]

attribute selector instead.