Skip to main content

Laravel is Good, Facades Aren't

I've been working on some Laravel 4 based packages lately which inevitably results in me also looking at other packages.

I've noticed sometimes that people use facades at times that give me pause. The most troubling being from inside their model classes.

A quick google turned up this video which assures people "there's still an instance behind everything, we're fine".  Everything mentioned in the video is true except that there is a very glaring omission.

Scope

What usually goes out the door at the start of a long series of mishaps in software design is scope.

When the desire to obtain a solution is stronger than the desire to consider the implications of a firm approach, mistakes are sure to follow.  Sacrifices like this are made due to the assumption of a high cost to developing carefully.

What really is happening however is a false dilemma, being responded to with a convenience decision.

It's very easy to write model code like this:

     class MyModel extends Model {
          public function getUri() {

               return Request::root() . '/mymodel/' . $this->slug;

          }
     }

...and then think that you've done a great job at putting the code where it belongs, encapsulation, simplicity, elegance and other nice things.

Unfortunately, what you've also done is created a dependency by conflating an infrastructure concern with your model.  Understand that this isn't because you've used a static method, but because you've used a class resolved via the container that might not be ready yet.

So what?

You could be blithe about this and think "I'm always going to be in a request because I'm coding for the web!"  At which point, we're facing another convenient assumption and it becomes all too easy to see where things unravel:
  • Scope - Imagine you find the model class above performing in a scope that doesn't find itself within a request.  This could be in a cron, or it could be when you need your class running unauthenticated.  These are just two examples, but a change in context can and will catch you by surprise and your programs will crash trying to use a method that was made in haste.
  • Cross Cutting Concerns - Suddenly a method you've authored on a class requires a security check, but you don't have access to the necessary information from within the class to do this.  You're faced now with further convenience decisions that lead to bloated model code.  That can lead to duplication at best because you're now having to consider all angles up front.
  • Difficult usage tracking - Facades lean on some PHP magic to do their thing. This also causes a break in the dependency graph of your application.

What instead?

Make services!  There is a fine line here, but when you need to do an operation that blends disparate pieces of information and resources, make a service.  The dependencies of that service will represent the scope you wish to create for that particular operation.  Once that service is made, you can rest easy knowing it exists in a space that has all the right dependencies and state to perform the singular operation you're after.

The other benefit is that it gives you a centralized point to handle cross cutting concerns.  If one day you need to perform a secondary operation (security checks, analytics, accounting), you won't be adding more bloat to your model.  You will simply find the scope (read: service) you know the operation transpires in, and add it there.

I can assure you that by taking this approach, you'll be doing yourself a favour in the single responsibility, cohesion, repeatability and DRY departments.

Popular posts from this blog

Laravel Project Architecture: The Missing Guide

At my job, we've been doing a lot of learning and development using Taylor Otwell's Laravel 4 PHP framework.  As we've become more familiar with it, we've had to come up with better ways to structure our projects outside of what the documentation indicates and the default distribution that is provided.

If you've been working with Laravel 4 for any amount of time or come with experience from another framework and are just getting started, you've probably noticed that there are a lot of different ways to cut up your projects.


Choice is nice, but sometimes it can be paralysing or misleading. Concrete Advice This post is done in such a way that you can just skim the headings, but if you want a detailed explanation in each section, feel free to read in where necessary.

While I can't say the entirety of my advice is in practice throughout the community, I can say that we are starting to use it, and to very good effect at my job.  Especially considering that our a…

Building .NET Core Nuget Packages

My last blog post was on building and publishing npm packages, specifically for typescript projects. In my opinion, packages are an important fundamental unit in software development. They sit at the confluence of technical competence and collaboration, and represent something the software community should be proud of.
Most of the time, you're going to be creating software packages once you're comfortable with some essential pillars: CodingProject structureSoftware architectureBuildingDeliveryCommunityLicensing
After I got my npm package up and running, my next task was to do the same thing with my C# libraries. Similar to protoculture, I have made a library called praxis.  This is done by leveraging the services and tooling known in the .NET ecosystem as nuget.


In this case, praxis abstracts many of the concepts and functionality I require when producing server projects. It builds on top of ASP.NET Core, so in that sense you can almost think of it as a framework-framework. The…

Amazon in Winnipeg?

Late last week, Amazon put word out that they're starting to look for a city to locate a second headquarters.  The fuel this announcement provides has the hype train in Winnipeg going so fast, it's missing stops.

I'm curious though.  As you've been digesting this exciting news, who have you spoken or listened to?
Was it this guy and his party's historic theft of public infrastructure (pdf) and incompetence with digital strategy? Or how about this guy, who maintains that Winnipeg doesn't need a proper rapid transit option?  You know, the kind that might actually catch the eye of handsome Amazon who now has Winnipeg trying to be the belle of the ball.


Runner up articles are Winnipeg's inability to join the 21st centry on active transport and the compounding of the aforementioned digital strategy mistakes.
Stop Listening to These Guys They are not technology experts. At worst, they're unabashed self promoters. At best they're conduits for very bad pol…