0.9.2-Beta

One of the biggest wins of this release is a massive improvement in our ability to parallelize and batch linter executions. Previously, the linting of every file ran in an isolated process, this meant that linters with high startup costs would drag when executing on a large number of files. We solved that by coalescing work into batches and the results are amazing. For example, previously running rubocop on the rubocop repo took over 10 minutes. We shaved that down to 38s - a nearly 20x speed improvement.

We’ve always aimed to deliver a completely hermetic toolchain because it leads to highly reproducible results, but some repos already manage some of their runtimes (Ruby is a popular one) and want to leverage/reuse their existing runtime deployments. Now you can! In the root of your trunk.yaml file you can now specify the runtimes you want to explicitly enable and the versions to deploy. When you use a range expression >=3.1.0 we will first check if your machine has a matching instance of ruby available. If the system already has it - we won’t bother installing a hermetic version of ruby. If you don’t have ruby - don’t worry - we’ll take care of installing a version that matches that requirement. Lot’s of other knobs to tinker with here (see docs for full story)

We love hearing from our users about all the ways we can make trunk better. In order to allow us to better prioritize asks and let people know when the new feature they asked for landed we've started gathering feedback here - features.trunk.io