Type System

Why and how we built Flow for JavaScript

JavaScript: The Good and the Bad

Developers like coding in JavaScript because it helps them move fast. The language facilitates fast prototyping of ideas via dynamic typing. The runtime provides the means for fast iteration on those ideas via dynamic compilation. This fuels a fast edit-refresh cycle, which promises an immersive coding experience that is quite appealing to creative developers.

However, evolving and growing a JavaScript codebase is notoriously challenging. Developers cannot move fast when they break stuff. They hit frequent interruptions, spending a lot of time debugging silly mistakes, unraveling assumptions and guarantees made by libraries written by others, etc.

In principle, this overhead can be mitigated by adding a layer of types to the codebase, and building tools that use type information to solve the above problems. For example, types can be used to identify bugs, to document interfaces of libraries, and so on.

The idea of using types to manage code evolution and growth in JavaScript (and related languages) is not new. In fact, several useful type systems have been built for JavaScript in recent years. However, type systems differ in their goals. On one end of the spectrum are permissive type systems that provide some degree of linting against likely errors, without regard for correctness. On the other end of the spectrum are restrictive type systems that can guarantee the correctness of static code optimizations, at the cost of interoperability. Another area that has not seen a lot of focus is the performance of type checking itself.

Flow: Goals

Flow is a static type checker for JavaScript that we built at Facebook. The overall mission of Flow is to deliver an immersive coding experience for JavaScript developers—a fast edit-refresh cycle—even as the codebase evolves and grows. In engineering terms, we identify two concrete goals that are important to this mission: precision and speed. These goals pervasively drive the design and implementation.

Precision

JavaScript bugs can have significant impact at Facebook. Developers want to find and fix as many bugs as they can by the time their code rolls out into production. So we must care about soundness. At the same time, we must also care about not reporting too many spurious errors, because a low signal/noise ratio implies a low fix rate.

In other words, we want Flow’s analysis to be precise in practice—it must model essential characteristics of the language accurately enough to understand the difference between idiomatic code and unintentional mistakes.

Precision also has other desirable consequences. When types are trustworthy, developers tend to rely on them to structure their code and reason about it, leading to cleaner and more efficient code with fewer dynamic checks. When type errors are trustworthy, developers can focus on what their code does rather than thinking about how to rewrite their code to satisfy (or work around) the type system.

Finally, precision enables useful developer tools to be built. In particular, the quality of results reported by Flow when the developer asks for the type of an expression, the definition reaching a reference, or the set of possible completions at a point through an IDE is correlated with the precision of Flow’s analysis.

Speed

Precision usually comes at the cost of speed. But while a precise analysis is desirable, most of the appeal of JavaScript is lost if we slow down the edit-refresh cycle by making developers wait as we compile.

In other words, we must engineer Flow’s analysis to be extremely fast—it must respond to code changes without noticeable delay, while still being precise enough in practice.

Like precision, speed also has other significant effects. When bugs are reported as the developer makes changes to code, they become part of the editing process—the developer doesn’t need to run the code to detect bugs, and tracing bugs back to the code becomes simpler. Similarly, when the IDE can show the type of an expression, the definition reaching a reference, etc. as the developer is coding, we have observed that productivity can improve dramatically.


What makes Flow precise?

One of the main contributors of Flow’s precision is path-sensitivity: the way types interact with runtime tests. The essence of many JavaScript idioms is to put together ad hoc sets of runtime values and to take them apart with shallow, structural (in)equality checks. In Flow, the set of runtime values that a variable may contain is described by its type, and a runtime test on that variable refines the type to a smaller set. This ability turns out to be quite powerful and general in practice.

What makes Flow fast?

The key to Flow’s speed is modularity: the ability to break the analysis into file-sized chunks that can be assembled later.

Fortunately, JavaScript is already written using files as modules, so we modularize our analysis simply by asking that modules have explicitly typed signatures. (We still infer types for the vast majority of code “local” to modules.) Coincidentally, developers consider this good software engineering practice anyway.

With modularity, we can aggressively parallelize our analysis. Furthermore, when files change, we can incrementally analyze only those files that depend on the changed files. Together, these choices have helped scale the analysis to millions of lines of code.

Under the hood, Flow relies on a high-throughput low-latency systems infrastructure that enables distribution of tasks among parallel workers, and communication of results in parallel via shared memory. Combined with an architecture where the analysis of a codebase is updated automatically in the background on file system changes, Flow delivers near-instantaneous feedback as the developer edits and rebases code, even in a large repository.