lage VS large-monorepo

But let's do some more real "apples-to-apples" comparison of Lerna's speed with useNx enabled. For benchmarking Nx we have set up a repo in the past which we regularly use to measure the speed of new Nx releases with other similar tools on the market such as Lage and Turborepo: https://github.com/vsavkin/large-monorepo. We now added Lerna+Nx (Lerna with useNx enabled) to that repo to measure the impact.

The last commit on that docs page in Lage was in August 2020, months before Turborepo was announced. So clearly, it wasn't Lage copying Turbo. https://github.com/microsoft/lage/blob/master/docs/guide/pipeline.md

The Nx daemon has seen significant enhancements, notably through the use of Rust to calculate file hashes behind the scenes. This improvement not only speeds up the start-up times but also optimizes performance even without the daemon, especially on CI environments where the daemon isn't used. The benchmark results at this repo showcase the remarkable speed improvements, making Nx competitive with native code solutions while maintaining the accessibility and flexibility of Node.js. Nx is still Node-first, so contributions are easier and only the most performance-critical parts of Nx are native code.

At Nx, we’ve heavily embraced Typescript from the beginning and we’ve been very happy with that decision. Nx also stands as the fastest JS monorepo tool available, demonstrating that adopting TypeScript does not necessarily compromise speed. However, we don't stop here. To push the boundaries further, we started to rewrite the most performance critical and computationally intensive parts of the Nx core in Rust.

You can see our results and the details of the benchmark - and even run the benchmarks for yourself in this repo.

Performance is at the core of what we do at Nx. Hence it isn't surprising that Nx is the fastest JS-based monorepo solution out there. We've shown that a couple of times. But every millisecond counts! As such, we decided to experiment with Rust to see whether we could further optimize our project graph creation as well as the hasher function that is used for the computation cache.

Speed - we drastically improved the speed of Nx, making it the fastest monorepo solution in the frontend space.

Up until Lerna v4, either the p-map or p-queue npm packages have been used to delegate the task scheduling. With v5.1 we introduced nx as an additional mechanism to schedule tasks. The advantage? Nx has caching built-in, which also gives Lerna caching support, making it lightning fast. A recent benchmark test resulted in Lerna being 2.5x faster than Lage and around 4x faster than Turbo (as of Oct 2022; test it out by yourself).

Nx is faster than most of the current available alternatives. See the corresponding benchmark repository

Delegating task scheduling to Nx allows to speed up any Lerna workspace in the range of 2-10 times. Check out our public benchmark which compares Lerna with other popular JS based monorepo tools.

(as always, feel free to reproduce it here)

But let's do some more real "apples-to-apples" comparison of Lerna's speed with useNx enabled. For benchmarking Nx we have set up a repo in the past which we regularly use to measure the speed of new Nx releases with other similar tools on the market such as Lage and Turborepo: https://github.com/vsavkin/large-monorepo. We now added Lerna+Nx (Lerna with useNx enabled) to that repo to measure the impact.