Context
Improved evaluation and build throughput and effectivenes translates to improved IDE responsiveness and saved costs on lab builds. So there is a clear ROI.
One way of achieving those goals is via redesigns (evaluation caching; moving from x-process parallelisation to in-process parallelisation; extracting long lived stateful caches - msbuild server; RAR service ...) - all those are tracked separately.
In this epic we want to focus on isolated targeted improvements via a "measure - analyse - improve" loop.
Docs