The Denali Doctrine

The Rails Doctrine describes how technological advantages erode over time. Today's features are important, but Denali's vision for tomorrow is just as much so. With that in mind, here's a few philosophical points that help guide our direction, partially inspired by that same article. Expect these to evolve over time.

1. Optimize for developer productivity

   Using Denali is a user experience - and we should treat it as such. We should be continually asking ourselves: what can Denali do to help developers work faster, smarter, happier?

   A great user experience is one that feels effortless. It anticpates failure scenarios and handles them gracefully. It leaves the user with a clear picture of "what's next", of how to solve the problem, and how to accomplish their goals. With each aspect of Denali, try to keep this at the heart.

2. Find the right abstractions

   Denali is a framework for leveraging shared solutions to shared problems. The way to leverage shared solutions is to find the right abstractions. The right abstraction exposes just the right levers in just the right places; too many levers, and you haven't fully captured the <em>shared</em> aspect of the solution. Too few, and you've over-simplified the problem.

   Put differently, a good abstraction minimizes accidental complexity while maximizing control over essential complexity. It provides a simple approach to solving the common use cases, while exposing more power to those that need it.

   Finding the right abstraction is hard; inventing it from whole cloth is even harder. Instead, Denali tries to leverage it's core/userland model (see #3) to allow for experimentation to <em>discover</em> the best abstractions through real world application, without subjecting users to too much API churn.

3. Stability without stagnation

   Denali aims to foster a stable core, while encouraging progress, experimentation, and innovation through "userland" addons.

   We often talk about “core” versus “userland”, terms inspired by language used to describe an operating system kernel. “Core” (or, "kernel" in OS-talk) is a minimal set of highly trusted, rock solid code that exposes basic primitives (the right abstractions - see #2). “Userland”, meanwhile, refers to Denali apps and addons, which are less bound by the strict stability guarantees of core. This reduce stability guarantee in turn allows for greater experimentation.

   By allowing userland code to experiment and move quickly, we can uncover the right abstractions, even occasionally adopting them back into core, without excessive API churn.

4. Follow the 80/18 rule

   Denali attempts to solve 80% of use cases well. We should plan features and interfaces with this group in mind, and aim make their experience delightful.

   For most of the remaining 20% (the “18%”), we make sure to provide “escape hatches”. These use cases might not be the principal ones we design around, but they should be possible to accomplish without feeling like you are fighting the framework.

   We purposefully leave out the remaining small fraction of use cases (the “2%”). This is an admission that we’re never going to solve every use case, nor should we. There are other choices out there besides Denali that are likely better suited.

5. Marketing matters

   Most of us would like to think that technology adoption is a purely rational, well thought out process of weighing tradeoffs and picking the best technical solution. Unfortunately, that’s usually not reality.

   This can be frustrating: vanity metrics and microbenchmarks often don't measure realistic metrics, and most users would never come close to those limits in any case. It's tempting to dismiss such discussion as "unimportant", and instead focus on a "purer" implementation of Denali.

   We reject this approach. Perception and marketing matter, to some extent. To phrase it more concretely: Denali will probably never be the fastest web framework out there by any microbenchmark standard. But we at least want to show up on the same graph.

6. Respect

   The Internet is a challenging place to foster great culture, so we have to be intentional about it. A great culture can help a project like Denali blossom and thrive - and a caustic one will kill it off before it can even germinate.

   The cardinal rule for participating in Denali's community is: treat everyone with respect. Behavior to the contrary is not welcome here.

   We aim to treat everyone with respect, to always remember the human on the other side of the keyboard, to start with the assumption of good faith motivation, and the always argue the best intepretation.