Bug Tracker

Opened 12 years ago

Closed 12 years ago

Last modified 11 years ago

#7912 closed bug (fixed)

$.fx.prototype.cur() is incompatible with cssHooks

Reported by: anonymous Owned by:
Priority: high Milestone: 1.5.1
Component: effects Version: 1.5
Keywords: Cc:
Blocked by: Blocking:


To calculate the start value of a property being animated, $.fn.animate() uses $.fx.prototype.cur(). This function then tries to access the DOM property first and, should it fail, falls back to $.css(), which is cssHooks aware. Unfortunately, the value returned by $.css() is invariably parsed as a Float, which breaks values for properties such as color, transform, gradient, etc.

var r = parseFloat( jQuery.css( this.elem, this.prop ) );
return r || 0;

Currently, cssHooks workaround this problem by re-calculating the start value on the very first step of the animation, see color.js and transform.js examples. Obviously, this solution is neither clean nor efficient.

Before returning the value, cur() could simply do the following:

var r = jQuery.css( this.elem, this.prop ),
  parsed = parseFloat(r);
return isNaN(parsed)? r : parsed;

But there's probably a good reason for it to currently return 0 when parseFloat fails. I need help finding this reason so that I can fix this bug.



Change History (9)

comment:1 Changed 12 years ago by Rick Waldron

Component: unfiledcss
Keywords: needsreview added
Priority: undecidedhigh

Thanks for taking the time to contribute to the jQuery project! Please provide a reduced jsFiddle test case to help us assess your ticket!

Additionally, test against the latest jQuery release and the jQuery 0 GIT version to ensure the issue still exists. Be Excellent to eachother!

comment:2 Changed 12 years ago by anonymous

I've opened a pull request with the proposed code and a new unit test: https://github.com/jquery/jquery/pull/165

comment:3 Changed 12 years ago by @…

i think with this little tweak that would be backward-compatible anyway :

var r = jQuery.css( this.elem, this.prop ),
  parsed = parseFloat(r) || 0;
return isNaN(parsed)? r : parsed;

comment:4 Changed 12 years ago by lrbabe

Sorry, the code you propose and the current code would have the exact same effect, the returned value will be 0 for a complex property.

I'm wondering if we were defaulting to 0 because some browsers return 'undefined' or other weird values for properties that have not been set previously...

comment:5 Changed 12 years ago by lrbabe

Of course, the 0 is here as a fallback for "" and "auto" values. Browsers never seem to return undefined...

I've updated my pull request to take this fact into account and I've also added unit tests.

Last edited 12 years ago by lrbabe (previous) (diff)

comment:6 Changed 12 years ago by dmethvin

Component: csseffects
Keywords: needsreview removed
Milestone: 1.next1.5.1

comment:7 Changed 12 years ago by snover

Status: newopen

comment:8 Changed 12 years ago by louisremi

Resolution: fixed
Status: openclosed

Fixes #7912. Make sure .cur() only returns 0 as fallback value when it needs to ("", auto, undefined, null).

This change makes .cur() more .cssHooks friendly. .cur() now returns the unmodified value by .css() if it isn't a number, number-alike or a value that needs a fallback to 0. This way fx.start doesn't need to be recalculated for complex values.

Changeset: 85d9343271da85fc945bf37a604873eaf247a3a7

comment:9 Changed 12 years ago by jitter

Version: git1.5
Note: See TracTickets for help on using tickets.