Not necessarily make any claims about what's preferable (one framework to rule them all vs many competing options) - though I'm certainly curious about that as well for the front-end frameworks.
Methinks someone who left Python in 2002 is out of touch. ;-)
At first, I actually thought this post was about a PHP microframework called Flight: https://github.com/mikecao/flight
I, for one, really like Dart. It's very nice to use.
I don't really know the history of Django but I'd wager it took a lot of inspiration from Rails.
Rails had that year's head-start on Django. By the time Django came along everyone was pretty deep into the development of Rails. I think it's mostly down to first mover advantage (it really did revolutionise web development).
In terms of frontend frameworks I have no idea what's going to happen.
Backbone definitely had the initial traction and it's nice and clean, but damn AngularJS makes development quick and easy. I've implemented things in the last couple of days in Angular that would have taken a week or more previously.
The Angular approach is unorthodox and seemingly inefficient but serves the developer well at the cost of CPU cycles.
Then again, I backed Prototype during the initial js lib battles and we know how that worked out.
Not exactly. Django mainly came into existence as a formalization of a lot of helper functions that The Lawrence Journal-World newspaper found themselves building over and over again when they built web apps. Adrian Holovaty, Jacob Kaplan-Moss, and Simon Willison created Django separately from Rails, even if it wasn't released publicly for a while (and then got kinda stuck at 0.96 for a while after that).
It's like Newton and Leibniz with calculus or Darwin and Alfred Russel Wallace with natural selection; there were a lot of similar problems to be solved in the web dev space in the early '00's, particularly around accessing databases, structuring URLs, and handling authentication, it makes sense that a few similar approaches would show up around the same time.
If you're really interested in the history, this "Snakes and Rubies" video is worth watching: http://www.youtube.com/watch?v=cb9KDt9aXc8
(I should mention that I tend to personally prefer Python to Ruby, very much like Django, consider Adrian Holovaty a friend and we both once worked at msnbc.com at the same time.)
My point still stands, I think that by the time Django made it out into the world Rails was already established as the dominant framework. Django 0.96 hadn't been as thoroughly thrashed as Rails and definitely had the warts to prove it.
Django is a much more mature beast these days, though I still lean more towards Flask if there's a choice to be made.
And can you imagine, they could have patented each and every one of those solutions, and spurred their competitors to greater innovation! What a surprise that they didn't - how silly of them. /s
tl;dr - Python had no web frameworks built in or community favorite. Rails was a complete re-think of how to do web apps in the age of XML (acaffolding, convention over config, etc.) Every language since has now realized this and copied Rails. This is a good thing.
if you want a story while drinking coffee, read below...
I was a serious Python guy from 1996 til 2006. I still like and respect Python. Before Python, I was a wanna-be Smalltalk fan. (This is important towards the end)
Python always had a philosophy of 'batteries included'. This means that it will have libraries for just about everything you need as supported parts of the language. Contrast this with Perl where CPAN quality was very hit or miss. Then contrast this with Lua or Scheme. They had too many libraries and none were standard. Most of those were, again, not supported or not of production quality.
As a result of 'batteries included', Python was just a dream to work with. Everything was doable or easy and performant 'enough'. I could go on, but that still wouldn't do it justice. They had a great language design, great libraries, and portability. Python is still a great framework. They didn't have a good 'CPAN'. This was ok, since the built-ins were good. Outside of that, it was a bit of a mess.
However, over the years, it became obvious that Python had a huge flaw. They did not deal with HTTP or HTML well. People don't even talk about this anymore, but the promoted framework was 'Plone'. This framework was so complex and convoluted, that building a simple web app was just not possible. It is not spoken about anymore. New frameworks popped up as a result, and very few of those were much good. This was a huge problem. Most people had to build their own production system. I had to switch languages for a job in 2002, and I didn't deeply follow the web framework discussion in Python for quite some time. I can tell you this, though. There wasn't a single framework that popped up as the winner or as exciting.
I'm typically a conservative. When I see a new thing, my typical take away is "this has been done before, what did they not learn from prior art". For example, lets say someone creates a new OS. If they don't really understand Unix, they will get it wrong.
When the first Rails screen-cast came out, I took a look. There was a lot of buzz that was increasing. I go, 'what the heck, it is only 5 minutes and it is a video.'
What I saw was an excellent architecture for building web apps. This architecture was better than anything else out there by quite a wide margin. They supported modern HTML. Convention over configuration. ActiveRecord was truly a treasure. The default site actually looked relatively decent (vs. the typical programmer, zero whitespace, huge serif, Netscape circa 1998 site). They URL conventions looked good. This architecture was human friendly. These people understood what a modern app should look like.
I was sold. Just from that screencast.
I believe in using whatever language is needed, if the framework is excellent. I wanted to build excellent web apps, so I learned Ruby. Ruby was easy to learn, and it is as good as Python. Much to my surprise it was the first language to pull off having Smalltalk blocks! People didn't even notice this and now they really like it. Everytime I showed this to coworkers, they would go 'what is that? I don't understand'. When people eventually learn Rails, they would go.. "check out these block things. They
So, in summary, Python didn't have a web framework. Rails was an excellent design, compared to any languages framework. Rails is pretty much going to be here for quite some time.
BTW, at this point. Rails is not my first choice. I think it is too heavy and web architecture has shifted a lot. Still, it would be in my top 3 choices.
Dealing with some of the deployment and infrastructure aspects of our rails app, I also see how much more memory and resources are consumed. Anything from Unicorn processes through rake tasks, to the time it takes to launch the console or run tests seem an order of magnitude longer. It's like steering a 747 or a cruise-ship. Once it's running it's fast and quite stable, with nice and comfortable features and lots of elegance. But it doesn't feel anywhere near as nimble and responsive as python/django (and django is by no means a small framework).
I'm wondering what's your top 3 choices today.
The other choices:
I like node.
I've always liked the async/model on unix. Their community is healthy and a lot of the open source is clean. The npm system is a good system. Even if you aren't running node, people are using node as their toolchain for pure front end development (css transform,asset pipeline). If you are doing anything real-time with a browser, it is my first choice. (Erlang used to be my choice there)
As another choice, I still like sinatra with rack middleware. Very light and it works well.
Beyond that, it depends on the application. I might use C,Scala,PHP or ?? who knows what.
Scala is allowing access to some good java machine learning libraries and tools such as mahout, scalding, storm, akka, finagle, therefore i believe as a developer we should not define ourselves to a platform or language but we should always be curious and use the best tool for the job. I am not using tornado for my current project because nodejs had the async libraries i needed, that simple.
As business opportunities are arsing around big data, sensors and the internet of things our favourite platforms and languages are going to evolve as well.
I also agree with your last statement. This is not the end of the road.
Good Lord. This is like 11 years ago?
Oh yeah, this reminded me of another painful thing back then. It was so difficult to find documented, clean MySQL bindings for python. Every experience there was terrible.
What would be your #1 and #2 choice?
You might have heard of it shadow DOM etc. Basically the idea is to be able to add GUI components eg <progressbar> that are as integrated as the native ones are.
Mozilla's X-Tags implementation seems closer to the goal though: http://www.x-tags.org/ IIRC it can do this because it's using Object.observable internally to detect DOM changes.
Unfortunately, we’ll have to stick around with them for the following couple of years, probably, as for now, most of the Web Components APIs are available only in Chrome Canary and Firefox dev builds. Monkey-patching (AKA “shimming”) to the rescue…
In my opinion, the more certain nowadays libs resemble the Web Components specs, the more clear the developers’ choice towards those libs should be.
"Flight is organized around the existing DOM model with functionality mapped directly to DOM nodes"
Am I missing something?
Flight is for building UI components, not client views. I think it might be akin to jquery widget plugins, but with more structure to make it easier and potentially more extensible (eg the mixins).
So yeah, in this case I think they were right to ride on DOM events.
Enforcing components w/o return values, communicating via pub/sub is also seen in aura.js and backbone marionette. I think this thinking will lead to more stable, easy to change js apps. Exciting!
It's also just so... Yeesh. Complicated, in comparison.
A big advantage over web components (or angular, or others) is that this is much more lightweight, less polluted and has a better separation of concerns. (On the other hand, web components will be a great deal for browser-app shops developing for third parties).
This is a maintainability nightmare. I've gone down this route on hand rolled JS frameworks before and it leads to untold headaches. Sometimes a little coupling is the right solution. Many times it is better for components to have knowledge of their environment. In most applications (other than iGoogle maybe) one thing on a page depends on another. I feel it's far better to explicitly call out those dependencies by having a reference to the other components and directly invoking methods on them with known arguments of a known type.
Edit: Also, what kind of memory allocation implications are there with event subscriptions and no references to the components themselves?
I think the sandboxing from aura.js is pretty interesting here, since you can always look up which widgets are allowed to talk, it makes it easier to narrow down the problem.
Yes, you need coupling when debugging. But it need not be in the libraries and code you use.
If instead you declaratively define the state of a component via a shared model then you don't have to care how something was updated, just that it was. Any new way you come up with mutating the object will be reflected in all of the components that watch the object.
Say I want to add a component to a page. Which events do I need to trigger to make it do the kind of things I want? What do I need to pass as a payload to those events? How do I know if there's a consumer for my event or not? If there isn't a consumer that's probably an issue.
The eventing bus adds an additional layer of indirection. If your app can be truly decoupled, then this would be great. But I just don't see that use case being very common. I have never worked on an application where the broadcast model was more benefit than pain in the long run.
Look at the event source.
> When you're debugging in Firebug or someplace, what does your call stack look like?
Look at the data passed to the event. The callstack starts at the function attached as a listener to the event.
Is the event data incorrect? Then the problem is in the code that triggers the event.
Is the event data correct? Then the problem is in the code listening to the event.
> Say I want to add a component to a page. Which events do I need to trigger
> to make it do the kind of things I want? What do I need to pass as a
> payload to those events?
Documentation. That is indeed the cost of using an abstraction that decouples code, but it's also a cost that has benefits - good documentation is useful.
> How do I know if there's a consumer for my event or not?
It's not important. That's the point of an event-based system, it's a decoupling. De-coupling is an abstraction. It works on pages that are based on components or modules.
If you don't want that abstraction, then using code based on that abstraction isn't a particularly good idea.
Event-based frameworks do need documentation - but reusable code should be documented anyway. There's just more emphasis on getting the documentation done in event-based systems.
> Is the event data incorrect? Then the problem is in the code that triggers the event.
> Is the event data correct? Then the problem is in the code listening to the event.
It's difficult to reason about the event triggering code when it's call stack is not available. For a larger code base where an event is trigger in multiple locations with different logic in each location, it may be difficult to pinpoint the triggering code, let alone reason about that code. Direct function calls are much easier to debug because you can walk up the chain of calls and inspect the variables and state of each frame.
Populate the event with the event source.
> Direct function calls are much easier to debug because you can walk up the chain of calls and inspect the variables and state of each frame.
If your listener depends on knowing the state of the triggering code, it is a sign your events are too granular. If you require a complete stack trace to be able to debug a listener to the event, that sounds like you have a bigger architectural problem to deal with.
Events are coarse. The listener should not care about why the event was fired, only that the event has occurred. Events are not a replacement for direct function calls. They are a mechanism for de-coupling pieces of code that can be independent of each other. It doesn't sound like your situation is one where there's that independence.
That isn't practical debugging. If I'm debugging an event that was fired and I don't know why, or the payload doesn't match my expectations, it would be very useful to inspect the state of the triggering function and see where things went wrong.
Right, we're on the same page. I agree that in certain scenarios these decoupled event based systems make sense. However, Twitter's example app is an email client. That to me is not a use case I would typically envision as being free of coupling. Thus it's more work to maintain in the long run.
Unless you're extremely new to the game I bet you have, if only in the sense that it's a very short hop from classic MVC design to this.
This is my biggest complaint about most web apps and the number one reason that I think web apps are perceived to be less powerful. Give me a break web devs! (Or, should I say browser makers?) Let's get serious already!!
(Also if we're talking browser manufacturers, I'd really, really like to see a completely separate abstraction for apps than documents. C'mon man!! We've been inventing the same thing for 20 years, let's get it right for once :)
This just does not jive with me. My template knows that submission should be handled by `addTodo()`? Or even worse, looping. I want all of my logic to live in code, not some abstract framework specific attributes that don't add much in the way of clarity (context switching between template and view just to follow the workflow, for example).
I use (and love) Backbone because the DOM is considered a pure presentation layer, nothing else. Any behavior, logic, whatever that the code controls is defined only in the code.
Most template engines support loops. Would be very weird without that capability, really.
Also, you have to add hooks to buttons and the like somehow. There isn't much of a difference between adding a class like "js-addWhatever" or using Angular's in-your-face approach.
Your template will also use names of properties, arrays, and objects directly. There also isn't a way around that.
At some point you have to connect your pieces. They can't float around in mid-air forever.
The presentational things you mentioned are different. We can separate those things more, we can use abstractions, and we can make it extremely reusable.
It doesn't matter if that identifier is a class or something else.
Well, if it's a class people may feel inclined to reuse that identifier on the CSS side, which is a bad thing, because now you've tied 3 parts together instead of just 2. If I use classes for this, I prefix them with "js-" to indicate that they are only meant for scripting purposes.
You don't actually have to do it that way in angular if you don't want to. You can put it all in code instead. The framework specific attributes you're looking at aren't exactly framework specific, they're kinda application specific (just happens they ship with the framework).
You can write your own ones to move any of that stuff into code instead of markup. I personally think that loops should be in markup - I happen to prefer templating systems with less restrictions.
I love Angular, but if you are rendering server side HTML (either as partials, or bootstrapping a page) it is currently lacking.
Okay, in that case I can see a need here.
> It's agnostic to how requests are routed, which templating language you use or even if you render your HTML on the client or the server.
While it is possible that they are collaborating with the current-Twitter developers working on this thing, it doesn't seem like much of a sure thing.
Ryan is one of the core contributors to knockout, by the way, so you can be assured of its quality.
Really, I was using KO and it was really nice at the time but since I've tried angularJS I don't see a reason to go back to KO.
Two way binding without ko.observable/ko.observableArray/ko.computed, just plain js objects is really a big win.