That would allow a lot of caching optimisations, as you can "index" each file in isolation.
[1]: https://github.blog/2021-12-09-introducing-stack-graphs/
[2]: https://github.com/github/stack-graphs