Attribute normalization broken

[caitp]

In AngularJS, attributes are normalized in such a way that the x- and data- prefixes are ignored, allowing authors to write valid HTML documents, without breaking the applications.

I was just testing a quick demo app (based on the angular.dart.tutorial), and to my surprise, the ng-model attribute works, while the data-ng-model attribute does not. (This is using the essentially the same pubspec and dart code from chapter 1 of the tutorial).

Curiously, data-ng-app works as expected, so maybe ngBootstrap() is clever about this.

...

TIL angular.dart is not normalizing attributes at all, and doing so would mean changing the way the NodeAttrs class works, hmm.

Down the line I think this would be a worthwhile thing to do, but changing this now would mean changing a lot of stuff and I'm not sure I'm acquainted with dart enough to do that.

[vicb]

Are you sure this is working for for ng-app ? What I mean is that w/o the ng-app attribute you'll fallback to the first dom node inside your document.

Beside that don't know if there are plan to support x- or data- prefixes.

[bgourlie]

What was the original intent of supporting x- and data- prefixes in angularjs? Was it a browser support thing? If so, Dart targets IE9+ so some browser compatibility issues that were relevant in angularjs may be non-issues in angular.dart.

[caitp]

@bgourlie browsers will generally support invalid attributes, however they do result in markup which is considered invalid, which makes relying on tools like the W3 validators much more of a pain. I wasn't involved with the project when attribute normalization was brought in, but I have to figure it was related to that, as well as support for XHTML documents.

[mhevery]

The reason why we have x- and data- prefix is that some people like to have their HTML to pass validators. I personally Don't like typing more. So normalization was the compromise.

Not sure how I feel about it in AngularDart. What are your thoughts? Should it be supported here too, or should we drop that support.

[zoechi]

I personally don't care much. I find those long names make the code harder to read. (I'm rather new to web development and haven't encountered any disadvantages yet)

[bgourlie]

It's hard to say. On one hand, I feel like people should be allowed to write valid html5 markup. On the other hand, what is valid per the spec (which itself is still evolving) and what browsers will accept are two different things. Personally, I could do without normalization.

[caitp]

Apparently the x-* attribute stuff is not valid HTML5, if it ever was, so that's one thing not to worry about adding. I'm not sure the XHTML namespace normalization really matters (and it historically has caused problems with SVG/MathML stuff). So maybe the solution is so simple as to just make the data- prefix invisible.

Anyways, my opinion is that authors should be able to write valid markup, but it's just my opinion. If people think that's terrible, alright

[mhevery]

Ok let's add data- Anyone up to do a PR? I think we would need to change the selector class to remove data-.