Consider this JSFiddle. It works fine in Firefox (14.0.1), but fails in Chrome (21.0.1180.75), Safari (?) and Opera(12.01?) on both Windows (7) and OS X (10.8). As far as I can tell the issue is with either the setData()
or getData()
methods on the dataTransfer
object. Here's the relevant code from the JSFiddle.
var dragStartHandler = function (e) {
e.originalEvent.dataTransfer.effectAllowed = "move";
e.originalEvent.dataTransfer.setData("text/plain", this.id);
};
var dragEnterHandler = function (e) {
// dataTransferValue is a global variable declared higher up.
// No, I don't want to hear about why global variables are evil,
// that's not my issue.
dataTransferValue = e.originalEvent.dataTransfer.getData("text/plain");
console.log(dataTransferValue);
};
As far as I can tell this should work perfectly fine and if you look at the console while dragging an item you will see the id written out, which means that it's finding the element just fine and grabbing it's id attribute. The question is, is it just not setting the data or not getting the data?
I'd appreciate suggestions because after a week of working on this with three attempts and some 200+ versions, I'm starting to loose my mind. All I know is it used to work back in version 60 or so and that specific code hasn't changed at all...
Actually, one of the major differences between 6X and 124 is that I changed the event binding from This has been debunked. The event binding method has no effect on the issue.live()
to on()
. I don't think that's the issue, but I've come to see a couple failures from Chrome when it comes to DnD while working on this.
UPDATE
I've created a new JSFiddle that strips out absolutely everything and just leaves the event binding and handlers. I tested it with jQuery 1.7.2 and 1.8 with both on()
and live()
. The issue persisted so I dropped down a level and removed all frameworks and used pure JavaScript. The issue still persisted, so based on my testing it's not my code that's failing. Instead it appears that Chrome, Safari and Opera are all implementing either setData()
or getData()
off spec or just failing for some reason or another. Please correct me if I'm wrong.
Anyway, if you take a look at the new JSFiddle you should be able to replicate the issue, just look at the console when you're dragging over an element designated to accept a drop. I've gone ahead and opened a ticket with Chromium. In the end I may still be doing something wrong, but I simply don't know how else to do DnD at this point. The new JSFiddle is as stripped down as it can get...
Ok, so after a bit more digging around, I found that the problem actually isn't with Chrome, Safari, and Opera. What gave it away was that Firefox was supporting it and I just couldn't say the other browsers are failing, since that's something I'd normally accept for IE.
The real cause of the issue is the DnD specification itself. According to the spec for the drag
, dragenter
, dragleave
, dragover
and dragend
events the drag data store mode is protected mode. What is protected mode you ask? It is:
For all other events. The formats and kinds in the drag data store list of items representing dragged data can be enumerated, but the data itself is unavailable and no new data can be added.
That translates to, "you have no access to the data that you set, not even in read only mode! Go f@&# yourself.". Really? Who'se the genius that came up with this?
Now, to get around that limitation you have few choices, and I'm only going to outline two that I've come up with. Your first one is to use an evil global variable and pollute the global namespace. Your second choice is to use the HTML5 localStorage API to perform the EXACT same functionality that the DnD API should have provided to begin with!
If you go down this route, which I have, you're now implementing two HTML5 APIs not because you want to, but because you have to. Now I'm starting to appreciate PPK's rant about the disaster that the HTML5 DnD API is.
The bottom line is this, the spec needs to be changed to allow for access to the stored data even if it's only in read only mode. In my case, with this JSFiddle, I'm actually using the dragenter
as a way to look ahead at the drop zone and verify that I should allow a drop to occur or not.
In this case Mozilla apparently opted out of full compliance with the spec which is why my JSFiddle was working just fine in it. It just so happens that this is the one time I fully support not supporting the full specification.