Virtual DOM in Elm(elm-lang.org) |
Virtual DOM in Elm(elm-lang.org) |
They play to the strengths of the virtual dom approach by manipulating the benchmark via dom instead of what ever interface was implemented within each TodoMVC implementation.
So I forked the benchmark and changed both the Backbone and Mercury implementations to work through their respective apis.
Here it is: https://github.com/smelnikov/todomvc-perf-comparison
as you can see the giant gap between Backbone and mercury is now gone while both tests perform the exact same tasks. (feel free to step through it in the debugger to see for yourself)
Here's my commit log: https://github.com/smelnikov/todomvc-perf-comparison/commit/...
Note: I've added a new method to the Backbone implementation for toggling completed state as the old one was horribly in-efficient. This is not something inherit in Backbone but rather is specific to this TodoMVC implementation. See my comments in the commit log.
Note 2: Exoskeleton, which is basically backbone without the jQuery dependency is roughly 2-3x faster than vanilla backbone, I'm going to predict that it will actually be significantly faster than mercury.
Note 3: I think the virtual dom is great and seemingly has many benefits but I feel as though the speed benefit has been greatly exaggerated.
React originally was designed for developer efficiency and not performance. It is a port of XHP (in PHP) that we use at Facebook to build the entire front-end and we're really happy with. It turns out that the virtual dom and diff algorithms have good properties in term of performance at scale. If you have ideas in how we can communicate it better, please let me know :)
This benchmark was taken from the webkit source code then forked into http://vuejs.org/perf/ then forked to include mercury then forked again to include elm.
Neither elm nor mercury came up with this benchmark and just added themself to it.
What this benchmarks shows is that async rendering is really fast. Mercury, vue & elm all use async rendering where DOM effects are batched and only applied once.
A better, close to real user experience benchmark would be one using mousemove since that's an event that can happen multiple times per frame.
The way that the Backbone TodoView is designed does not take into account the possibility of a user adding 100 items using the dom within a tight loop. Probably because such a use case is impossible outside of this type of benchmark. By doing so the Backbone implementation ends up performing a lot of unnecessary renders. Therefore as far as Backbone performance is concerned this benchmark is not indicative of any real world scenario.
Just to re-iterate; when you're loading a set of todos from your local storage to display when the user first opens the page, you would not populate the "new todo" input box and fake an enter event for each item that you want to add. Instead you would reset the Backbone.Collection with a list of your new todos (go through the interface). That's basically the change I made to the benchmark. Sorry if it wasn't clear.
See my proof of concept video: https://vimeo.com/100010922
And actually runnable example that you can edit without refresh: https://github.com/gaearon/react-hot-loader
I plan to write a blog post explaining how to integrate it into React project soon.
I've just had a look at the code and React is using a localstorage backend, while Backbone is using its models + collections with a localstorage extension... so I'd expect there to at least be some overhead there, but apparently not.
Does anyone have any quick thoughts on what might be happening here? I can't shake the feeling that these benchmarks might not be terribly useful.
The same in PureScript would be
profile user = mkUI spec do div [ className "profile" ] [ img [ src user.picture ] , span [ text user.name ] ]
See, types everywhere. https://github.com/purescript-contrib/purescript-react/blob/...
Disclaimer: I'm the author of Vue.js. The benchmark is a fork of a fork of the Vue.js perf benchmark (http://vuejs.org/perf/).
React looks interesting, but only if it gives significant advantages (time-to-market, maintainability, etc) vis-a-vis AngularJS in managing a large code-base.
Any first hand reviews?
(I'm not affiliated, just a very happy user.)
I hear you about the API surface. Currentlu AngularJS has too many weird/new concepts.
node : String -> [Attribute] -> [CssProperty] -> [Html] -> Html
div : [Attribute] - > [CssProperty] -> [Html] -> Html
div = node "div"
bareDiv : [Html] -> Html
bareDiv = div [] []
Update: the benchmark uses a correct implementation available here: https://github.com/evancz/todomvc-perf-comparison/tree/maste... so it was a false alarm on my part. Tried it out at https://rawgit.com/evancz/todomvc-perf-comparison/master/tod...
I hope the new framework gets animation right, I'd love to see it as flexible as in D3, in my opinion this is something currently React currently lacks (the transitions are there but they are too simplistic to cover complex web app cases).
I'm sure others are experimenting as well, so we should see more implementations soon.
Or is this the limit of an overly mutable language like js?
The Om example does some kind of Event delegation using channels which is much faster.
I am a bit concerned about the lack of typeclasses and what that could mean if I try and build something bigger using it. Maybe I could use Purescript and Elm together.
The problem is that it's very easy to force a full recalculate of the whole page layout. Whenever you call .offsetHeight or .offsetWidth or .getComputedStyle, you're doing it. The full list of properties is about 2 dozen strong:
http://gent.ilcore.com/2011/03/how-not-to-trigger-layout-in-...
Most web developers don't know this, and so they're actively making their pages slow. Worse, many popular frameworks build this into the library, and so if you use them, there is no way to keep your pages responsive. JQuery, for example, can easily cause 4-5 layouts with a single call to .css; on a mobile phone and a moderately complex page, that's about a second of CPU time.
Also, the fact is that the web is stuck in a Web Components-driven approach to building apps which is pretty orthogonal to how this works.
I haven't dug deep enough to form an educated opinion on the language, but so far its refreshing to see such a drastically different process than popular languages today.
Not by default because it's missing laziness and immutability: because just about everything in javascript is mutable, React can't prune out the changeset as the developers could be modifying the application state behind its back in ways unknown (or worse, could be using state which is not under React's control, neither props nor state, but globals and stuff).
That is, for safety/correctness reasons the default `shouldComponentUpdate` is `function () { return true; }`. The result is React has to re-render the whole virtual DOM tree (by default) and the only possible gain is when diffing the real and virtual trees.
Because Clojurescript and Elm are based on immutable structures they can take the opposite default (and if you're going behind their back and mutating stuff it's your problem).
Also, I'm not sure React defers rendering until animationFrame.
An optimized React application (or at least one which uses PureRenderMixin[0]) ought have closer performances to the others's (Om is a bunch of abstractions over React after all, so you should be able to have at least as good performances in pure React as you get in Om).
[0] http://facebook.github.io/react/docs/pure-render-mixin.html
* Development React is slower than production React. There are a bunch of extra checks all over the place along with a profiler [1].
[1] http://facebook.github.io/react/docs/perf.html
* Speed isn't the top priority of the framework, predictability is. There's a virtual event infrastructure and other browser normalization work going on. Om is using React under the hood and more or less represents the best case scenario.
* React isn't magically fast. The diff process still has to visit all the nodes in the virtual DOM, generate the edit list, and apply it, which is a substantial amount of overhead. I'm used to seeing React even or behind when benchmarked with a small number of nodes. The explanation I've seen is that these benchmarks aren't considered important since the goal isn't to be as fast as possible but rather to never be slower than 16ms.
The trick behind most of the "React is fast" articles is that React is O(N_dom) instead of O(N_model) so if you can shrink the size of the output DOM, React goes faster. The Om line demonstrates this and doing screen-sized render over a sliding window of data in a huge data set (most grid/scrolling list demos) is another common example. There are perf knobs that probably aren't being turned here but if the app renders fast enough why would you waste your time turning them?
I consistently got the result of Angular utterly and completely destroying React. Initially I blamed the virtual DOM approach, but after seeing other frameworks utilizing it and outperforming Angular by a huge margin, it seems to me that React is not written for performing well on small DOM documents. (There might be a turning point, considering how bloated Facebook pages it was designed for.)
My limited understanding of React is that it fails in (a), (b) and (c), and only limited measures can be applied to improve them. Re-creating the entire DOM on each update probably does not help. I have no information if (d) is possible with it.
I am using Angular.dart for a while now, and it can be used to get all of them in an optimal way.
Disclaimer: I'm working at Google.
[1] http://swannodette.github.io/2013/12/17/the-future-of-javasc...
These is certainly something wrong with the benchmark. Since Om is a layer on top of React, it is obvious that React itself cannot be necessarily slower than Om. (Perhaps idiomatic React usage is slower than idiomatic Om usage for this case, though?)
edit: Rather, see masklinn's comment that describes what actually happens. Point being, vanilla React does extra work to account for anything a developer might do, but allows Elm and Om, which have more rigorous standards for their users, to override that behavior.
edit: I was running the original test instead of your fork.
1. Two-way bindings complicate things because there is no single source of truth. (See http://vimeo.com/92687646 at 30:00)
2. Template/directive separation is superficial, in fact these are single concern and should be together.
3. Separation between `props` and `state`, as well as documenting `props` via `propTypes` encourages very natural modularity.
Overall, it's been a really enjoyable experience - a stark contrast to the horribly buggy mess that is iOS 8.
Personally I have high tolerance for this, but I know some folks would get frustrated quickly.
They're literally announcing the untyped base library layer and the post specifically calls out the desire to build higher level abstractions. Elm is moving incredibly fast, so your question is totally unreasonable.
Since Om or Elm assume immutable inputs and pure components they can skip rendering components altogether when the inputs have not changed (mentioned in TFA's "Making Virtual DOM Fast").
React can do that, but it has to be opted in component by component either by using PureRenderMixin or by implementing shouldComponentUpdate.
When you use something like mori, but you need to define data transforms, such as float precision, derived attributes, and so on, how do you make that work well in your apps without too much complexity in your components? One thing I really enjoy about ampersand-state and ampersand-collection (forks of Backbone) is that I can define very simple, standard functions per model so that my views don't have to have any idea what was passed to them. How would that be possible in Mori? Do you use something like a `__type` attribute so an external utility can suss out what the object is and transform it correctly?
I don't know what you mean by "derived attributes", or why it would be difficult to do data transformations on the models that deal with mori data structures. It'd be great to discuss this somewhere, maybe on #reactjs IRC? I'm jlongster there.
That said, I'm probably being overly optimistic and I'm just starting to research it. I don't quite like how addons.update feels like a bandaid, but maybe it is good enough. Haven't done enough research yet. I definitely don't like writing updates the way addons.update forces you to, but sweet.js macros could solve that (and I was going to write macros for mori anyway).
https://github.com/dustingetz/react-cursor/blob/master/examp... https://github.com/dustingetz/react-cursor/blob/master/js/Cu...
(It is backed by react.addons.update, it provides a mechanism like mori.assoc_in for immutable subtree updates, and it preserves reference equality for equivalent cursors (value/onChange tuples))
(Cursors are also not vulnerable to issue#122 https://github.com/facebook/react/issues/122)
React.addons.update uses normal JavaScript arrays. So it won't scale as well, but at least you get immutability.
(b) You can implement shouldComponentUpdate in order to have a quick way not to re-render a sub-tree if nothing changed.
(c) See (b) but we're also working on changing the internal representation of the virtual DOM to plain js objects that can be reused[1]. We were super worried about GC but it turns out that it hasn't been the bottleneck yet for our use cases.
(d) If you are writing pure React, all the actions are batched and actually, it's write-only, React almost never reads from the DOM. If you really need to read, you can do it in componentWillUpdate and write in componentDidUpdate. This will coordinate all the reads and write properly.
A really important part of React is that by default this is reasonably fast, but most importantly, when you have bottlenecks, you can improve performance without having to do drastic architecture changes.
(1) You can implement shouldComponentUpdate at specific points and get huge speedup. We've released a perf tool that tells you where are the most impactful places to[2]. If you are bold, you can go the route of using immutable data structures all over the place like Om/the elm example and you're not going to have to worry about it.
(2) At any point in time, you can skip React and go back to raw DOM operations for performance critical components. This is what Atom is doing and the rest of their UI is pure React.
[1] https://github.com/reactjs/react-future/blob/master/01%20-%2... [2] http://facebook.github.io/react/docs/perf.html#perf.printwas...
You seem to be an implementor, so two questions that maybe spare me looking at the source code :-)
1. How do you batch updates? 2. I am currently using an algorithm for longest increasing subsequences for avoiding superfluous dom insertions of children when diffing lists. I also make sure that the node containing the active element will not be removed from the tree during the diffing (if possible at all). Are you doing the same?
It's possible to change the batching boundaries via "Batching Strategies" but we haven't exposed/documented it properly yet. If you are interested, you can look at requestAnimationFrame batching strategy. https://github.com/petehunt/react-raf-batching
2. We cannot really use normal diff algorithms for list because elements are stateful and there is no good way for React to properly guess identity between old and new. We're pushing this to the developer via the `key` attribute.
See this article I wrote for a high level overview of how the diff algorithm is working: http://calendar.perfplanet.com/2013/diff/
2. I don't think we spend a lot of time trying to make this super optimal, but git grep ReactMultiChild to see what we do.
DOM reuse is not the same thing moving a DOM subtree to a different place. DOM reuse is e.g. getting an already-rendered table row, binding a new value to it, and modifying only the DOM properties in the complex DOM structure that actually did change. E.g. you modify only an Element.text deep in the first column, and a few other values in the other columns. Or maybe you need to do more, but all you do is delta. You don't just annotate a DOM structure with a key at row level, as it is closer to a hash of the DOM, not speaking of the data-dependent event handlers.
Calculating the DOM (virtual or not) is expensive, compared to not calculating at all. Creating a virtual DOM structure and not using it afterward creates GC pressure, compared to not creating at all. We are talking about optimizations in the millisecond range. A large table with complex components inside will reveal the impacts of these small things.
DOM coordination is not just making the DOM writes in one go. Complex components like to interact with each other, depending on their position and size on their page, and the changes in the underlying structure. They read calculated style values, and act upon those values, sometimes causing reflows. And if such things happen at scale, forced reflows may cripple the performance, and coordinating such changes may be more crucial than the framework you are choosing.
I am sure that people who are familiar with React may have their way get these stuff. I have looked at it, and I haven't seen it to happen automatically, while with Angular.dart, I get it without effort.
The virtual DOM determines which mutations are necessary and performs only those, batched. It also reuses DOM nodes as appropriate. DOM nodes can even be keyed against data in the case that, between transitions, it isn't entirely clear which nodes correspond to which data.
(a) Using the key property, will give React a manner to determine likeness of elements
(b) Don't calculate the expensive things at render time, do them when loading or modifying state.
(c) Is related to a, but I haven't run into large problems with this personally.
(d) React does batch changes to an extent I believe.
Perhaps you meant virtual DOM here? (in any case, the actual DOM is not recreated on every update)
DOM node reuse is perhaps the central theme of React so it's odd that you bring this up as a criticism (see https://www.youtube.com/watch?v=1OeXsL5mr4g)
Calculating the virtual DOM does come with some processing and GC overhead, yes. But any system that tracks changes for you (data binding) comes with overhead and React makes the right set of tradeoffs for real apps (since it is a function of how large your render output is, not your underlying data model). React has about a 70% edge in CPU time on Angular in the "long list" class of benchmarks (which drops to a mere 40% with the Object.observe() performance unicorn). And steady state memory usage is almost always better with a virtual DOM approach since again it only tracks what you actually render which is usually smaller than your data model (https://www.youtube.com/watch?v=h3KksH8gfcQ).
DOM coordination boils down to non-interleaving of reads and writes to the DOM. React manages the writes for you which happen in one go. Components have a well-defined lifecycle which is also batched and are only allowed to read from the DOM during specific points in the lifecycle, which are coordinated system-wide. So out of the box if you follow the guidelines you will thrash the DOM much less (see http://blog.atom.io/2014/07/02/moving-atom-to-react.html)
<div class="row">
<div class="cell">
<div class="align-left">
Value.
</div>
</div>
</div>
<div class="row">
<div class="cell">
<div class="align-center">
Value B.
</div>
</div>
</div>
I agree that you can take this really far. At this point I need to sit back and thing about it. :) I think mori could provide better performance for certain types of apps, but it does come at a cost if interop. Time to hit the hammock.
Email me if you want to talk about it.
Why do all of this instead of just using Om? I'm currently waffling between Om and straight react.js for a project of mine.
Able to use cljs and don't care about designer friendly -> what are you waiting for!!!
Thank you!
On the HTML markup, there are many valid reasons you may want to use non-TABLE based tables:
- it allows better rendering control for infinite scrolling (DOM reuse, re-positioning, detached view for sticky header and column)
- it allows you to have real (CSS style-able) row groups, or if your structure is hierarchical, it allows you a better control to create a treetable (reduced rendering time if you expand a node and insert a bunch of rows in the middle).
- it allows you to have multiple grid systems inside the table (e.g. a detail row may use up the entire row, and it may have its own table inside, which you'd like to synchronize across multiple detail rows). I guess this later benefit is just redressing the fact that you do need to implement an independent grid system anyway :)
I tried to make the example as minimal as possible, so I don't show off a lot of the features (i.e. state, event handling), but I did use JSX, an optional syntax extension for function calls.
[1]: http://facebook.github.io/react/docs/pure-render-mixin.html