Regression from 1.2.0rc1: ng-include 'bugs' href/ng-href in an anchor element

[sdenier]

When I put a ng-include directive directly inside an a element, I get the following error in the console:
TypeError: Object #<Comment> has no method 'setAttribute'

For example:

      <a ng-href="#/{{href}}" ng-include="'link.html'"></a>
      <a href="#/{{href}}" ng-include="'link.html'"></a>

In the first case the element is not displayed as a link (as href is not set).

See Plunker http://plnkr.co/edit/LWa2asKqAxUYBtZL23YE?p=preview

At first I thought I was dealing with ng-include issues #3584 or #3573 . However, after testing this issue does not appear in 1.2.0rc1 and previous releases.

[sdenier]

I forget to tell. A basic workaround is to put ng-include on a child element, or directly as an element of the anchor:

<a ng-href="#/{{href}}">
  <span ng-include="'link.html'"></span>
</a>

[jankuca]

This is caused by ngInclude replacing the original element with a placeholder comment node and cloning the original element after that. If the included content is later replaced, a new copy of the original is created and placed there while the previous copy is removed. This, I believe, makes it possible to define transitions (animations) between the old and the new content.

[IgorMinar]

this should be easy to fix. since ngInclude is terminal, we should check if that could be used to prevent the attribute interpolation directive getting processed prematurely by tweaking priorities.

[IgorMinar]

@mhevery @matsko do you have another suggestion?

[matsko]

I'm starting to think about reverting ngInclude back to what is was and possibly introducing a new feature where you can style the newly created container element for animation purposes. What do you think?

Regarding this, terminal should be fine. @sdenier maybe test this out using ngView or ngRepeat first (since they both use terminal already)...

[sdenier]

Indeed, it works when I put a ng-repeat instead of the ng-include in the anchor element

http://plnkr.co/edit/nejaV3tZSqEpMkMM5YBr?p=preview

[btford]

I submitted a PR, following @IgorMinar's suggestion.