jQuery 1.4.4 Release Candidate 2 Released

Posted on by

We’re happy to announce that jQuery 1.4.4 Release Candidate 2 is now available! This is the second release candidate of jQuery 1.4.4 – a follow-up maintenance release to jQuery 1.4.3. The code is stable (passing all tests in all browsers we support), feature-complete (we’re no longer accepting new features for the release), and needs to be tested in live applications.

Grab the code:

How can I help?

To start, try dropping the above version of jQuery 1.4.4rc2 into a live application that you’re running. If you hit an exception or some weirdness occurs immediately login to the bug tracker and file a bug. Be sure to mention that you hit the bug in jQuery 1.4.4rc2!

We’ll be closely monitoring the bug reports that come in and will work hard to fix any inconsistencies between jQuery 1.4.3 and jQuery 1.4.4.

What’s Been Updated?

There are a few areas in jQuery that have seen changes since 1.4.3 was released:

  • (New) Added a new animation method, .fadeToggle()
  • (Enh) Calling .data() with no arguments now includes data from HTML5 data- attributes (#7222)
  • (Enh) Moved jQuery.props from support.js to attributes.js (#6897)
  • (Enh) .width() and .height() now report the width and height of hidden elements (#7225)
  • (Bug) stopImmediatePropagation was not being honoured in live/delegate event handlers (#7217)
  • (Bug) Fixed an issue where host and protocol were not compared case-insensitively when determining whether an AJAX request was local or remote (#6908)
  • (Bug) Fixed an issue where the “clone” variable was not being declared correctly (#7226)
  • (Bug) Fixed a bug where we only change the ID on nodes that don’t already have an ID for rooted qSA (#7212)
  • (Bug) Changing an HTML5 data attribute after calling .data(‘foo’) no longer causes .data(‘foo’) to also change (#7223)
  • (Bug) Fixed a bug where Opera didn’t give height/width of display: none elements with getComputedStyle but did with currentStyle – fall back to that if it exists added.
  • (Bug) Fixed a bug to ensure accessing computed CSS for elements returns ‘auto’ instead of ” consistently (#7337)

It also fixes a number of regressions in 1.4.3. One that caused:

  • (Bug) JSONP calls to fail when cleaning up after a callback (#7196)
  • (Bug) .removeData() to fail (#7209)
  • (Bug) “ready” events to fire twice when added using .bind(“ready”, foo) (#7247)
  • (Bug)  .css(‘width’) and .css(‘height’) to return 0 or negative values when trying to get the style of a hidden or disconnected element (#7225)
  • (Bug) the attribute not equals selector ([foo!=bar]) to not work in Firefox (#7243)
  • (Bug) find() to fail when selecting from forms containing inputs named “id” (#7212)
  • (Bug) .children(selector) to fail on XML documents (#7219)
  • (Bug) child (>), next sibling (+), and previous sibling (~) selectors to fail when combined with non-CSS pseudo-selectors like :last (#7220)
  • (Bug) an error “handler is null” to be raised when passing null as the event handler (#7229)
  • (Bug) it to be impossible to include a content-body with DELETE requests (#7285)
  • (Bug) it to be impossible to include data with HEAD requests (#7285)
  • (Bug) an issue where IE was firing click events on disabled elements when using live/delegate (#6911)
  • (Bug) .show() to fail if .hide() was first called on an already-hidden element (#7331)
  • (Bug) .show() to fail if an element was hidden in a stylesheet, then had .css(‘display’) manually set prior to calling .show() (#7315)
  • (Bug) Sizzle.contains to throw an error on browsers that have no support for compareDocumentPosition or documentElement.contains (#7236)

Full details concerning the release are forthcoming – for now we just need your help in catch regressions.

With your input we should be able to produce a solid release. Right now we’re looking to get the final 1.4.4 release out in about a week. Thanks for your help in reviewing jQuery 1.4.4rc2!

We would also like to remind you that the sixth maintenance release for jQuery UI 1.8 is also now out. For more information on this release, feel free to head over to the jQuery UI blog for more information.

15 thoughts on “jQuery 1.4.4 Release Candidate 2 Released

  1. I agree with Leo : jQuery should remove that “feature”, or add a second parameter, like :

    .data(‘oops’) gives the standard string (as in 41.4.2 and before)
    .data(‘oops’,’json’) makes like a JSON.parse($x.data(‘oops’))
    .data(‘oops’,’boolean’) render a boolean interpretation of the string

  2. @AzeriFire, in general .live() works on IE. If there is a specific case where it doesn’t work for you and there isn’t an open bug in the bug tracker for it, the odds are low that it will be fixed. This blog is not the bug tracker, you can find it here:


  3. For me, for future releases, and in order to improve perfomance, it would be VERY usefull to check if the browser is CSS3-transition-capable and transform effects into inner-browser effects (no every effect can be translated, but some like slide, fade, ect can be almost transformed into height or opacity transitions).
    And in modern browsers, they are hardware accelerated getting much more smooth animation.
    There is a plugin that do this but it doesn’t work perfect.

  4. Nyuszika7H on said:

    You made a typo! ;)

    “We would also like to remind you that the sixth maintenance release for *jQuery 1.8* is also now out. ”

    jQuery 1.8 -> jQuery UI 1.8

  5. After reading about the new bugs in 1.4.3 being quite serious on common methods such as width() or height(), it feels better not to upgrade to this version in a production environment…

  6. Wow! jQuery 1.4.4 rc2 now report the width and height of hidden elements using .width() and .height() ! Great work!

  7. .show() .hide() .dialog() is bugs.
    dialog close and use element hide()
    and again dialog open. but, not open to dialog.

    now. so i still use 1.4.2 version.

    my enligh very bad. sorry :(