Can anyone who has tried Ember and Angular on moderately sized apps (500-1k+ LOC) compare their experiences?
I'm using Angular, I love how little code I have to write to make testable apps with their two-way binding, and dependency injection. I went from 0 to Todo App in about an hour after I started to learn Angular. Should I give Ember a try?
Ember is a lot harder to go from 0 to Todo App in an hour. There's a fairly steep learning curve, but once you've grokked the "Ember Way", building more complex apps is significantly easier.
With Angular, you can build a Todo App in no time, but building anything more complex requires you to understand e.g. Services, Directives, and Transclusion.
With Ember, you are just learning about Models, Views, & Controllers -- it doesn't get significantly more complex than that initial learning curve.
Learning directives, services, and the scary sounding "transclusion" has not been an issue at all for my team and I who are using angular on a very large project and loving it. Directives are quite simple once you understand the basic principles. Transclusion when actually explained (http://www.egghead.io/video/cjrBTjMvruI) is a trivial and obvious concept. I understand the confusion one may have around services/factories/providers because they are all very similar and in some ways little more than syntactic sugar for a single API, but once their differences are sorted out as in this blog post (http://iffycan.blogspot.com/2013/05/angular-service-or-facto...) they too become very natural.
For me it wasn't learning the terms, I didn't have an issue there. The problem was how limited Angular was in comparison to Ember, particularly the idea of directives. I actually spent a large amount of time on freenode discussing in the Angular chatroom how I might construct a directive to replicate a common jQuery plugin like Zurb's Joyride. In short it couldn't be done in anyway that seemed remotely non-hackish.
Ember has jQuery included, so using existing plugins is generally easy or at least possible. Beyond that, it has a far more powerful router (try doing deep routing with Angular. Angular UI's router may be changing this, but you would still have to include outside code to get the power of the standard Ember router) and in the future much more powerful REST support with Ember-Data (currently volatile, but still very powerful).
I started building a large application with Angular using FuelPHP for the REST server. Angular simply didn't have the power to do the things I needed it to at that time. Hopefully that's changing, since it seems to be the dominant framework these days.
I'd say that once you've mastered Angular concepts like services and factories, dependency injection, etc., it's far easier to build complex applications in it than in Ember. That's because the abstractions Angular introduces are focused around writing decoupled and reusable code.
If you want to build an app in an hour, I'd suggest knockout instead of either ember or angular.
This is the same experience I had. I built a simple Angular app quickly but I soon realized I didn't understand how to go from the sample code on their website to doing more complicated things as the structure of the framework is very different from what I am accustomed to.
I haven't built anything in Angular yet, but I can vouch for the Ember side of this. Getting started was pretty painful, but I now feel remarkably productive even in more complicated scenarios.
This was my experience exactly. I started a project with Angular and got the basics set up without difficulty. When it came to integrating Google Maps, it was a nightmare and I switched to Ember. It took me a little while to figure out "The Ember Way" but the integration is very clean and understandable now.
Very high learning curve but it levels out. It doesn't get more complex unless you make it. It's hard to transition to a fully async environment if you're coming from the server side.
from my limited experience with Ember, it appears that it hosts entire application from a single base url, and overrides forward/back buttons so that the page doesn't reload in order to keep state. instead it uses query string view paths and tokens to navigate around right? Does Angular support anything like this?
I'm throwing my 2 cents in, although I haven't built a large scale app with either, I did do a full evaluation of both before my latest large app and I decided on neither one. There were many reasons but every time I wanted to do anything that was remotely non-trivial both fell apart. Although most of that had to do with the MVC pattern more than the actual framework (but the two are tied intricately) In the process of evaluating these two frameworks I found KnockoutJS and I chose it for my large project that is one year into development. After one year with it I love it more now then when I first started. It's not perfect but it is a really well thought out framework, it does lack some documentation, ko.utils I'm looking at you. But it's documentation is vastly superior to the other two frameworks, at least it was a year ago. It's very stable and I prefer the team's approach. The two binding and subscription model is great for solving some intricate problems. It's easily extensible with custom bindings and unlike angular doesn't favour bad coding habits that lock you in to the framework, I forget the name now but their custom HTML tags (I think they call them directives) is nothing but bad news and vendor lock in.
That's pretty much the same conclusion I came to. Wake me up once ember-data is stable and can support things like range queries and pagination, and just generally dealing with a small subset of data.
The Ember app that I'm working on has about 20 controllers plus all the models, templates, etc. It's a few thousand LOC right now.
We use it because we don't actually serve (much) HTML. We server up basically an empty DOM and the Ember app uses Handlebars templates to generate the entire webpage as it pulls data from our REST API.
Akamai hosts all our Ember code + handlebars templates + css + images, so the client only hits our servers for the REST API.
I've done some Angular with a simple internal tool, but it didn't see possible (or at least very easy) to take a similar approach so we stuck with Ember.
I'm obviously misunderstanding something in your comment because that sounds exactly like how you'd build an app with Angular. Could you explain the issue in a little more detail.
Angular is vastly superior to Ember technology wise; Their approach is also more future proof (No stupid getter/setter, string templates, the architecture is based on something very close to web components, etc)
Thank you for your well-reasoned comment. However, it sounds like you're basing your opinion on some misinformation. That's okay, it happens to everybody. :)
Ember.js is extremely future-proof. Yehuda Katz, who I work with on Ember.js, is on both the W3C and TC39, the committee designing the next version of JavaScript. Needless to say, we have a pretty good idea of what's planned for the future of the web platform; we're also very involved with the standards bodies to make sure it is as powerful as possible for building web applications.
Setters and getters are a trade-off. While we'd prefer not to require them, the alternative is that you have to diff objects every time a change happened. And not just a change, but a possible change—which means in response to almost any event. By using getters and setters, we avoid this particular performance footgun.
Actually, Ember views have much more in semantically common with Web Components than Angular directives, despite the fact that many people like to repeat this misinformation. For more, see this thread on Google Groups: https://groups.google.com/forum/#!msg/polymer-dev/4RSYaKmbtE...
Yes, forgive my tone. I should just have said that when it comes to picking a third party framework (I don't always do), I personally prefer the compromises made by Angular than the ones made by Ember. Of course, the biggest compromise of Angular (dirty checking all over the place just to be sure) also make it unsuitable in a number of situations :)
About future-proof, I mean the Angular programming model is closer to what we'll be doing in a few years (There won't be any get/set or computed properties anymore); I never implied the Ember team is not aware of what's coming.
To me none of the frameworks are ideal, although I find Ember or Angular well architectured, as far as implementing the vision of the authors go. With the current capacities of Js and no compilation steps, compromises can not be avoided :-(
I'm happy to see RC4. Ember has really come a long way since I started using it. Ember itself has really stabilized. I'm really looking forward to 1.0.
There is one more thing I'd really like to see before hitting 1. https://github.com/tildeio/router.js/pull/19 is probably the most important thing happening right now. 1.0 will not be there until these problems are addressed. I hope to see the PR make it to 1.0.
Scary breaking change though. Something that was called automatically before isn't anymore. I personally would have opted for a safer alternative to implement this breaking change. I guess it's a little detail in the grand scheme of ember thing, but for the poor developers who will have RC4 on prod, there could be some very sneaky bugs.
That being said, happy to see ember getting closer to 1.0 Final, it's a beautifully designed framework.
I'm using Angular, I love how little code I have to write to make testable apps with their two-way binding, and dependency injection. I went from 0 to Todo App in about an hour after I started to learn Angular. Should I give Ember a try?