aganea wrote: > This enables more efficient implementation of the caching VFS in #88800. Thanks @jansvoboda11 for working on this, I appreciate. One question, what do you mean by "more efficient"? https://github.com/llvm/llvm-project/pull/102540