And when you have gone and reimplemented every single operation from some "bloated" library.
Then you will be enlightened.
I was already leaning towards 'no' regarding the project, but this has sealed the deal.
And, sure, you can use something lighter-weight than the full ca. 100Kb jQuery reference implementation, but "1Kb MVP framework", while it sounds awfully nifty, still lacks something in basic honesty by comparison with "1Kb MVP framework, oh, and you have to write a ton of jQuery glue to make anything work with it."
[Edit: Don't believe me? Take a look at the Riot.js TodoMVC example. This isn't a framework; this is jQuery with extra hair.]
That's the point I'm making: I prefer, inasmuch as possible, not to do manual DOM manipulation at all, which renders your question somewhat moot.
Perhaps it's different for someone who hasn't written that kind of code enough times to grow thoroughly tired of it, but I have written it that many times, and I am therefore very much open to the suggestion that there's value in a framework which handles the great bulk of such cases automatically, and in so doing gives me more time and effort to spend on whatever project I'm actually working on. Angular, in particular, I've found to be quite good for this; Riot.js, on the other hand, not only doesn't bother with it at all, but takes what I consider to be an unduly snobbish attitude toward the whole concept.
For a tiny chat app that's designed to be embedded in third-party pages, I can see some sense in taking the lightweight microframework-style approach. But when its developer and its devotees promulgate it as the prescription for every problem, while denigrating much more capable and general-purpose solutions as "bloated" and casting veiled aspersions upon the competence of the developers who employ them? That's a real problem, and I strongly misdoubt I am the only one to think so.
No doubt Riot.js is quite useful in a specific subset of cases, albeit only those closely similar to the one in which it found its origin. But this sort of behavior, on the part of those who already have a stake in it, can only make it less likely that anyone who doesn't have a stake will decide they want one.
That's precisely the decision I made upon reviewing the original Riot.js thread on HN, and this latest article from the Riot.js developer, in which he arguably outright lies both about the drawbacks of his competition and the benefits of his own entry in the great framework shootout, only reinforces that decision -- and, again, I much misdoubt I'm the only one.
I can see from your comment that you are biased towards MVVP and Angular in particular. You prefer to express the logic in the HTML template and use two-way data binding and avoid using jQuery (Official Angular suggestion). It's a very different approach and you cannot really compare Riot and Angular directly. It's a much broader comparison between MVP and MVVM patterns. A massive topic.
I strongly think that MVP scales much better than MVVP and is better for bigger applications/teams. It's easier structure the codebase in modular fashion. Riot documentation focuses solely on this topic. Moot is not a "tiny chat app", it has pretty vast feature set and the code is split into 50 different files. MVP definitely scales.
The demo application has several scripts after /body. `navi` should be `nav`.
> Pure HTML is cleaner and passes W3C validator.
The demo does not pass the W3C validator.