How does lage work?

So how does lage make builds faster? To fully appreciate how lage gives you the best build performance compared to other monorepo task runners, take a look at this example. Here we have a repo with this dependency graph:

Level 1: Typical Lerna or Workspace Runners

First, let's take a look at the typical workspace runners. Lerna, pnpm recursive, rush and wsrun all will run one task at a time. This creates a sort of "build phase" effect where test scripts are not allowed to run until build.

Level 2: Scoping

One of the first way to speeding up build jobs is to use "scoping". Usually a change only affect a subset of the graph. We can get rid of the builds of FooCore, FooApp1 and FooApp2 if the only changes are inside BarCore. However, we'll note that BarPage is still affected, resulting in this.

Level 3. Caching

To further improve build times, we can take advantage of build caches. If we had previously built certain packages, we should be able to speed up the build with a cache. Here, the BarCore packages have already been built and tested, and so

Level 4. Pipelining

Finally, the last thing we can to speed things up is to break down the wall between build phases from the task runner. In lage, we define the relationship between scripts in the pipeline configuration.