Optimising Haskell developer tool performance using OpenTelemetry

The Haskell developer tool ecosystem, while greatly improving in recent years, has a performance problem. Haskell IDE Engine (HIE), the most popular implementation of the language server protocol for Haskell, is a great example of this. While it is an incredibly useful tool, its variable performance interferes with the productivity of Haskell developers.

Profiling tools is another area where the Haskell ecosystem is lacking. While some tools do exist (for example ThreadScope), they are not as polished as similar tools for other languages, and do not allow for remote monitoring, which can be incredibly useful in distributed applications.

I believe these two problems have a common solution: OpenTelemetry is a set of APIs and protocols made for instrumenting code, gathering traces and metrics, and analysing all that data, in a language-agnostic way, on a variety of independent clients.

As my Google Summer of Code Project, I would like to first improve the newly-created haskell-opentelemetry library, use that to instrument HIE, and then optimise it based on my findings.