I am somewhat obsessed with clean, repeatable, maintainable dev environments and Docker has all the pieces to create one. For the past several months on several projects, I’ve been using Docker (and docker-compose) to not only run services like Postgres and Redis, but also as a virtual machine in which to run all my dev commands. My Mac is only running Docker and vim.

The problem is that Docker on Mac has very slow file system access. Running bin/rails server on a fresh Rails app can take almost a minute to come up. I recently discovered that, by using NFS, file system access is way faster. This creates a bit of a problem for webpack-dev-server, but this, too, can be fixed. I want to write out what I did in case it’s helpful.

At RubyConf last year, I announced a new book I was working on that outlines the way to write Ruby on Rails apps sustainably, so that teams, orgs, and code could continue to be valueable and easy to work with over time.

I worked remotely for Stitch Fix for six and a half years, originally as an engineer, then as a manager, director, technical project manager and architect. With all the tips and tricks going around, I want to talk about the one strategy to be successful working remotely. This strategy should be the basis for everything you do, and any productivity hacks that don’t roll up to this strategy should be set aside. The strategy is to build and cultivate trust.

Most developers, when invoking the “Law of Demeter” or when pointing out a “Demeter Violation”, do so when a line of code has more than one dot: person.address.country.code. Like the near-pointless SOLID Principles, Demeter, too, suffers from making a vague claim that drives developers to myopically unhelpful behavior.

As mentioned in the original post, I’m realizing that the SOLID principles are not as…solid as it would seem. In that post, I outlined the problems I see with the Single Responsibility Principle, and in the second, I recommended ignoring the Open/Closed Principle, since it is confusing and most reasonable interpretations give bad advice. In the third post, I talk about how the Liskov Substitution Principle is too narrowly focused on the wrong problem, and doesn’t really give usable design guidance, and the fourth is about how the Interface Segregation Principle isn’t the right way to approach problems with coupling.

Now we get to the last one, the Dependency Inversion Principle, which could also be called “The Reason 2000’s Java is Equated With Writing All Your Code in XML Principle”. The principle says that code should depend on abstractions, not concretions. Because it is a principle, the implication is that all code should depend on abstractions. No. No it should not. Depending on abstractions has a cost which the principle largely ignores. Let’s see how.

As mentioned in the original post, I’m realizing that the SOLID principles are not as…solid as it would seem. In that post, I outlined the problems I see with the Single Responsibility Principle, and in the second, I recommended ignoring the Open/Closed Principle, since it is confusing and most reasonable interpretations give bad advice. In the third post, I talk about how the Liskov Substitution Principle is too narrowly focused on the wrong problem, and doesn’t really give usable design guidance.

Now, I want to talk about the Interface Segregation Principle, which prescribes are very strange solution to the problem of coupling, and the reality is, we should just be talking directly about coupling and cohesion and be very careful about over-optimizing for one or the other.

As mentioned in the original post, I’m realizing that the SOLID principles are not as…solid as it would seem. The first post outlined the problems I see with the Single Responsibility Principle, and in the second, I recommended ignoring the Open/Closed Principle, since it is confusing and most reasonable interpretations give bad advice. Now, let’s talk about the Liskov Substitution Principle, which, as it turns out, is not design advice at all.