Transclude + ngClass = clone unexpected class

[maxisam]

This is a rare case and I am not 100% positive about my theory.

So let's say if there is a directive with transclude feature, and there is a ngClass in the transclude part, it will generate unexpected class during clone process. Because during clone process, transcludeFn(scope, function(clone) { }); , I think the clone part is compiled with the scope you feed. But if the data is not ready at that moment, the unexpected class will be generated. And unfortunately the unexpected class will be cloned as well, and there is no way it will be cleared by ngClass later.

Here is the example plunker
Another example this one is simpler.
You can inspect the icon in field4, and you will see there are 2 classes for it. It supposes to be one. The first one was generated during clone process.

I can walk around this issue but I wonder if we need to feed scope to transclude function, since it is likely the clone part will be compiled later with the scope you feed it. Why do we need to feed transclude scope then ?

[maxisam]

edit note: It turns out there is a bug in transclude with IE. I put it in different issue here

There is another issue related to this one. In IE (All versions), none of the row is showing correctly. The data binding is not working correctly this way.

It seems like in the second example, data binding works fine in IE. So it might be just table

[petebacondarwin]

Related to #7388

[maxisam]

Actually there are 2 issues here. One is related to #7388 the other one is not. The 2nd post is the issue related to #7388. The first one is not. FYI, this is the reason why I used transclude.