On Dec 8, 2013, at 12:19 PM, Daniel Freedman <dfre...@google.com> wrote: > Developers want data-binding, and the auto cloning <template> does not give > them a favorable timing model. > They want to set those up before the ShadowDOM is stamped, on a per-instance > level. > If they were to use the automatic template, it would be far too late, and > there could be unnecessary network requests or FOUC.
Indeed. This is why we've included a data binding feature in our original proposal: http://lists.w3.org/Archives/Public/public-webapps/2013OctDec/0418.html We can refine it further to make it more library/framework friendly. In fact, I'm already in contact with Rafael. W. about this. > To remove a bit of vaguness from this scenario, Polymer elements use > data-binding in almost all cases. > Event handlers, computed properties, MVC, everywhere. > As such, no Polymer element would use the automatic <template> registration > argument. > I doubt that elements created with other libraries like Ember or Angular > would make much use of it either. Sure, but that doesn't mean non-framework authors wouldn't use this feature for simple use cases to stamp out the shadow DOM with a static template. > However, if some low level data-binding primitives were introduced to the > platform, there would be some real merit in an automatic template argument. > There would have to be some modifications to the proposal, such as adding > hooks for data-binding information to be given to the template instance, but > I think those details can be discussed when such a data-binding spec arrives. That seems entirely reasonable, and we're indeed interesting in spec'ing such a feature. > Until data-binding primitives arise, I think this automatic template is a > premature discussion. Let us spec that automatic template mechanism then. - R. Niwa