Grasping Deno's limitations
As with everything, choosing solutions is a matter of dealing with trade-offs. The ones that adapt best to the projects and applications we're writing are what we end up using. Currently, Deno has some limitations; some of them due to its short lifetime, others because of design decisions. As it happens with most solutions, Deno is also not a one-size-fits-all tool. In the next few pages, we'll explore some of the current limitations of Deno and the motivations behind them.
Not as stable as Node.js
In its current state, Deno can't be compared to Node.js regarding stability for obvious reasons. Node.js has more than 10 years of development, while Deno is only nearing its second year.
Even though most of the core features presented in this book are already considered stable and correctly versioned, there are still features that are subject to change, and live under the unstable flag.
Node.js's years of experience made sure it is battle-tested and that it works in the most diversified environments. That's something we're hopeful Deno will get, but time and adoption are essential factors.
Better HTTP latency but worse throughput
Deno keeps performance on track from the beginning. However, as seen on the benchmarks page (https://deno.land/benchmarks), there are topics where it is still not at Node.js' level.
Its ancestor leverages the direct bindings with C++ on the HTTP server to amplify this performance score. Since Deno resisted to add native HTTP bindings and builds on top of native TCP sockets, it still suffers from a performance penalty. This decision is something that the team plans to tackle after optimizing TCP socket communication.
The Deno HTTP server handles about 25k requests per second with a max latency of 1.3 milliseconds, while Node.js handles 34k requests but has a latency that varies between 2 and 300 milliseconds.
We can't say 25k requests per second is not enough, especially since we're using JavaScript. If your app/website needs more than that, probably JavaScript, and thus Deno, is not the correct tool for the job.
Compatibility with Node.js
Due to many of the changes that have been introduced, Deno doesn't provide compatibility with existing JavaScript packages and tooling. A compatibility layer is being created on the standard library, but it is still not close to finished.
As Node.js and Deno are two very similar systems with shared goals, we expect the latter to execute more and more Node.js programs out of the box as time goes on. However, and even though some Node.js code is currently runnable, that is not the case currently.
TypeScript compiler speed
As we mentioned previously, Deno uses the TypeScript compiler. It reveals itself as one of the slowest parts of the runtime, especially compared to the time V8 takes to interpret JavaScript. Snapshots do help with this, but this is not enough. Deno's core team believes that they will have to migrate the TypeScript compiler to Rust to fix it.
Due to the extensive work required to complete this task, this is probably not going to happen anytime soon, even though it's supposed to be one of the things that would make its startup time orders of magnitude faster.
Lack of plugins/extensions
Even though Deno has a plugin system to support custom operations, it is not finished yet and is considered unstable. This means that extending native functionality to more than what Deno makes available is virtually impossible.
At this point, we should understand Deno's current limitations and why they exist. Some of them might be resolved soon, as Deno matures and evolves. Others are the result of design decisions or roadmap priorities. Understanding these limitations is fundamental when it comes to deciding if you should use Deno in a project. In the next section, we will have a look at the use cases we believe Deno is the perfect fit for.
