LA RubyConf 2012
Rack Middleware as a General Purpose Abstraction

Summarized using AI

Rack Middleware as a General Purpose Abstraction

John Bender • February 04, 2012 • Burbank, CA

In his talk at LA RubyConf 2012, John Bender addresses the common issue of large classes in Ruby on Rails applications and presents Rack middleware as a solution for object composition and code organization. He starts by identifying the challenges posed by monolithic classes, where extensive numbers of methods complicate code maintainability. Bender emphasizes that while there are various approaches to manage large classes, including mixing in functionalities, he will focus on object composition and middleware as more effective strategies.

Key Points Discussed:
- Issues with Large Classes: Bender highlights that developers often encounter unwieldy user classes in Rails applications, which can lead to poor organization and difficulties in navigation.
- Existing Solutions: He mentions that existing solutions involve mixing in functionalities or breaking down methods, but stresses the importance of context in code organization.
- Object Composition: An illustrative example he provides is refactoring the handling of bulb age from a lamp class into a separate bulb class, showcasing how separating concerns can simplify class structure.
- Understanding Middleware: Bender explains what Rack middleware is, describing it as a framework for processing requests where each middleware has the opportunity to modify requests as they pass through a stack. He clarifies the function and importance of middleware in Rails applications, especially compared to Vagrant's easier middleware management, which reduces complexity during operations such as managing virtual machines.
- Refactoring with Middleware: The speaker uses the Diaspora social network as a case study, proposing that large classes can be refactored by extracting methods into new classes, allowing for better organization and making it easier to manage dependencies.
- Conclusion: To consolidate functionalities and improve class structure, Bender suggests using middleware as a way to group behaviors logically, promoting code reusability and simplifying testing. He concludes that with careful consideration of when to apply middleware, developers can significantly enhance the maintainability and organization of their applications.

Overall, Bender advocates for middleware as a particularly powerful abstraction for improving class design and streamlining code in Ruby on Rails applications.

Takeaways:
- Middleware is beneficial for handling common functionality scattered across classes, promoting modularization and clarity in code.
- Object composition through middleware can lead to a more manageable and efficient architecture in software development.

Rack Middleware as a General Purpose Abstraction
John Bender • February 04, 2012 • Burbank, CA

We've all seen the monolithic Rails model, pages and pages of methods all dumped into one class. Inevitably someone starts moving things around just to feel better about the loc count without making any real difference. How can we reify actions on an object and simplify our classes?

In this talk we'll examine Rack middleware as a general purpose method of object composition, see examples of it at work in Vagrant, and use these ideas to simplify an existing application.

Help us caption & translate this video!

http://amara.org/v/FGkq/

LA RubyConf 2012

00:00:22.930 All right, so my name is John Bender.
00:00:26.289 You can find me in various places. I work at Adobe, where I am employed full-time with jQuery Mobile.
00:00:30.500 The reason I mention this is because I’m really grateful for the opportunity to work in a field I enjoy.
00:00:34.390 So, today I want to address a common issue in programming.
00:00:38.059 What are we really focusing on today?
00:00:39.739 This talk is about the problems with very large classes, particularly in Ruby on Rails.
00:00:49.430 I think everyone in this room is familiar with gigantic user classes in Rails applications.
00:00:52.870 Let me ask, how many of you are Rails developers? And how many have encountered a massive user class in a Rails application? Hands, please!
00:01:03.559 It’s a pretty common occurrence — these gigantic classes with pages and pages of methods. Today, I would like to address that and offer a technique for improvement.
00:01:16.100 There are existing solutions for handling large classes. One approach is to mix in functionalities.
00:01:22.820 I will present some examples but you don't need to read all the code — the key takeaway is context.
00:01:30.950 For instance, in a Rails app, you can utilize session-based modules where each class is a simple include. This way, everything is logically grouped, and you avoid scrolling endlessly through an enormous class.
00:01:50.670 Another approach is object composition, which is essential when refactoring your codebase.
00:01:56.209 An example: if I have a lamp, I need to know about maintenance methods like finding out bulb age.
00:02:02.819 This is a simple illustration, but conceptually important.
00:02:03.909 The bulb age inquiry is encapsulated within the lamp class. Instead, we can delegate the inquiry to the bulb, separating concerns.
00:02:10.180 This demonstrates object composition — moving methods to where they logically belong.
00:02:14.319 Today, we will focus on Rack middleware as an abstraction.
00:02:16.290 But first, how many of you are familiar with middleware? Raise your hands if you have worked with it before.
00:02:25.080 Not everyone, but that’s okay. I will be using the terms middleware and middlewares interchangeably.
00:02:35.720 Now let's take a quick review.
00:02:37.720 To understand middleware as a general-purpose abstraction, we need to look at it in its native environment. So, how does Rack middleware work?
00:02:50.139 For those who are not familiar, a brief overview is needed. Middleware is designed to process requests, modifying them as they pass through.
00:03:02.319 A middleware stack operates as the request flows through. At the top of the stack, the first middleware gets a chance to manipulate the request.
00:03:14.180 Let’s take a simple middleware example.
00:03:16.629 When a request comes in, it initializes the middleware, providing the next middleware in the stack as an argument.
00:03:29.799 It's critical that every middleware calls the next middleware in the chain; this distinguishes middleware from endpoints, which do not delegate.
00:03:41.919 In this scenario:
00:03:43.340 the request moves through the stack where each middleware can manipulate the request or response before passing it along.
00:03:56.030 Let’s clarify the differences between using middleware with Vagrant and traditional Rack middleware.
00:04:09.440 Vagrant's middleware processes requests in a more straightforward manner. You simply create a lambda function, and it handles calling the next middleware for you.
00:04:18.400 On the other hand, Rack assumes you’re aware of the next middleware in line, which means when coding for Rack, you need explicit calls to pass on the requests.
00:04:38.600 This is crucial to ensure every middleware in the stack can process requests appropriately.
00:04:55.000 At this point, I want to ensure everyone's clear on these foundational ideas before moving ahead. Do you have any questions?
00:05:03.370 If not, I’ll proceed, and we’ll touch on how Vagrant utilizes middleware to perform actions.
00:05:13.940 So, how many of you have used Vagrant? Great! For those who haven't, Vagrant helps manage virtual machines easily, streamlining their use in development.
00:05:28.990 Utilizing Vagrant requires various operations, including system configurations, port forwarding, and file sharing.
00:05:39.000 Let’s dive into a simple command example. When managing a virtual machine with Vagrant, suspending it is straightforward.
00:05:56.000 Here’s what the middleware stack looks like for suspending a VM. These are the necessary operations for successful suspension.
00:06:06.561 As you can see, the VM moves through each middleware in order, which facilitates the suspension process.
00:06:21.480 When we run the command, it first checks the configuration to ensure it's valid—this is essential for the functioning of the virtual machine.
00:06:31.690 If the configuration is incorrect, Vagrant raises an error and halts further processing. However, if everything is valid, it continues with the middleware processing.
00:06:46.030 Next, it verifies access to the environment to ensure no issues arise with the virtual environment.
00:07:00.090 If all checks are successful, it proceeds to actual suspension of the virtual machine using appropriate Vagrant commands.
00:07:15.470 The importance here is that Vagrant handles middleware management, streamlining the process without manual delegation.
00:07:25.000 Lastly, creating a VM from scratch involves numerous operations, but thanks to middleware, these processes can be reused efficiently.
00:07:39.800 Simplifying the operations involved prevents an overly bloated class that’s difficult to navigate.
00:07:54.880 We learned early on while developing Vagrant that we wanted to avoid large classes, as they lead to convoluted dependencies and testing complications.
00:08:11.470 So we sought an intermediate solution between large classes and middleware to streamline our development process.
00:08:23.560 To illustrate, let’s refactor a Rails application, specifically targeting the user class of the Diaspora social network.
00:08:32.070 The Diaspora user class is notably large, consisting of numerous methods that can be refactored to enhance maintainability.
00:08:48.240 As I show you this, consider grouping these methods to create a distinct class that can handle specific responsibilities.
00:09:03.319 We can extract methods that show dependencies and group logic, making it much simpler to manage.
00:09:17.080 First, we need to identify methods free from many dependencies to facilitate extraction. This will make our refactor simpler.
00:09:30.320 For instance, with the invitation method, even without diving into specifics, we notice possibilities for extraction.
00:09:43.520 Now, as I move this to a new class, the idea is to define responsibilities clearly and make proper use of middleware design.
00:09:58.890 I will create a class with the relevant methods needed for handling this aspect. It will interface smoothly with the existing environment.
00:10:12.480 The process allows for cleaner organization and a better design pattern, which ultimately aids testing and code readability.
00:10:27.570 Furthermore, it enhances flexibility, allowing method calls based on context and requirements.
00:10:41.670 The user environment facilitates injecting dependencies during these calls, making our operations seamless.
00:10:58.670 This kind of segmentation into middleware adheres to code reusability, especially when we anticipate multiple methods doing related processing.
00:11:10.410 With this implemented, adapting our architecture to fit future needs becomes less cumbersome.
00:11:25.890 To wrap up, here are some signs indicating when middleware is a suitable solution.
00:11:39.890 When you find common functionality scattered across various classes, consider consolidating actions into middleware.
00:11:52.490 Avoid relying solely on mixins as they may complicate your class structure; instead, utilize object composition to modularize your code.
00:12:05.310 Furthermore, if methods essentially wrap existing behavior, reassess their structure and possibly designate them as middleware functions.
00:12:17.590 Code organization and abstraction, such as companies where functionality is frequently utilized, are excellent candidates for this approach.
00:12:30.750 In conclusion, middleware presents a powerful method for enhancing class abstraction and organization.
00:12:43.850 If you consider these ideas, reshaping how you deal with large classes and their responsibilities will significantly benefit the overall performance of your applications.
00:13:05.600 Thank you for your time today, and I hope you found these insights helpful!
Explore all talks recorded at LA RubyConf 2012
+6