527749 - Drag events out of order

Status: RESOLVED FIXED

(Core :: DOM: Copy & Paste and Drag & Drop, defect)

Description

[Daniel Bates]

User-Agent:       Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1.3) Gecko/20090824 Firefox/3.5.3
Build Identifier: Mozilla/5.0 (Macintosh; U; Intel Mac OS X 10.5; en-US; rv:1.9.1.3) Gecko/20090824 Firefox/3.5.3

The drag-and-drop processing model used in Firefox does not conform to section 7.9.4 of the HTML 5 spec. <http://dev.w3.org/html5/spec/Overview.html#drag-and-drop-processing-model>. In particular, the order of dragenter and dragleave events are reversed in Firefox. That is, Firefox fires dragleave then dragenter, but it should be dragenter then dragleave.

This issues is similar to an drag event ordering issue in WebKit bug #30754 <https://bugs.webkit.org/show_bug.cgi?id=30754>.

Reproducible: Always

Comment 1

[Daniel Bates]

Attachment: Test case

Test case derived from a similar WebKit test.

Comment 2

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Confirmed on trunk.

Comment 3

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

http://mxr.mozilla.org/mozilla-central/source/content/events/src/nsEventStateManager.cpp#3812

Should be a pretty simple fix.

Comment 4

[Kyle Huey (Exited; not receiving bugmail, old account, do not use)]

Probably a quick fix for somebody familiar with the code and an HTML5 compliance win.  Nominating for blocking.

Comment 5

[Phil Lacy]

need a dupeme maybe, But also I think trunk is correct.

Comment 6

[Phil Lacy]

oops sorry missed comment 2

Comment 7

[Benjamin Smedberg]

Enn, can you take this? (And hopefully get a test into the HTML test suite, if such a thing exists.)

Comment 8

[Neil Deakin]

Attachment: Patch

This patch moves the dragleave event to fire after the dragenter event. For developers that want the more sensible event order, including the places in browser, editor and toolbar customization code, the dragexit event still fires before the dragenter event as before.

The event order then is:
  dragexit, dragenter, dragleave

Can't be tested automatically. Works with the testcase.

Comment 9

[Olli Pettay [:smaug][[e-mail]]]

This is rather strange. Why should dragenter fire before dragleave?
Is this something that IE does always? I guess I need to test IE.

Could we just change HTML5 draft?

Comment 10

[Neil Deakin]

IE always fires dragenter before dragleave. About a year ago, Webkit was changed to this behaviour as well.

Changing our behaviour could have compatibility issues for us which is why I had to change code to use the dragexit event.

Likely though, there would be less of a problem to go from the IE behaviour to our behaviour as I can't see why the unusual event order would be useful and relied on.

Comment 11

[Hixie (not reading bugmail)]

Yeah the IE behaviour (and thus the spec) is pretty nutty here.

dropexit seems like a reasonable idea; if you'd like it formalised and specced please feel free to mail the WHATWG list mentioning it.

Comment 12

[(no longer active)]

http://hg.mozilla.org/mozilla-central/rev/d11d8ac9ffe3