It really does. Isomorphic is the new standard. Doing it in a very manual way is not good enough. Every new JS framework is expected to do this out of the box and the older frameworks are scurrying to gain support.
React literally does this out of the box. Om does it (based on react). Reagent does it (based on react). Using js/React (raw js interop) does it. There is actually nothing to do. You exec your js in a nashhorn thread. That gives you some html. You ship it your browser on initial GET, then initialize react on it.
This is identical to what you'd be doing on react+node. With a macro or two, it's easier than anything node provides.
As an aside, isomorphic isn't a new anything. We were rendering pages on the server, rendering updates on the server with the same templates, sticking them in xml representing jquery dom manipulation, and pipelining those in a single request since at least 2007 (when I first encountered the technique and shipped with it, I believe we used a javascript library called Taconite, maybe since 2005 from looking at the sourceforge account). Progressive enhancements were rendered off the same templates and inserted asynchronously, automatically, based on whether the controller on the server detected it was an ajax request or a bare GET. This incidentally made things very cacheable.
> React literally does this out of the box. Om does it (based on react). Reagent does it (based on react). Using js/React (raw js interop) does it. There is actually nothing to do. You exec your js in a nashhorn thread. That gives you some html. You ship it your browser on initial GET, then initialize react on it.
You're being disingenuous, I read the ClojureScript mailing list, this subject comes up regularly. People are still figuring it out, Nashorn is not well documented and the use of it for this purpose is not well documented. Please don't pretend like something is a common thing when it's not.
EDIT: The example in the link provided builds on the now deprecated CLJX, so it's not entirely up to speed. (CLJX was the way to write clj/cljs in the same codebase until "reader conditionals" came along).
