Since we're covering `pathmap` now, we can end up with more interesting scenarios where it "conflicts" with `serverpath`.

Since we're covering `pathmap` now, we can end up with more interesting scenarios where it "conflicts" with `serverpath`.

We already had an issue here where `serverPath` was effectively "first one wins" rather than "longest prefix wins" and this extends that if `pathMap` includes a path that's a subfolder of some user specified `serverPath` (making it less obvious)

Since it's pre-existing, we can probably handle it in a follow up PR; but I figured it's worth calling out.

---

There might also be some normalization considerations, particularly with the keys ending with `\` but URLs not strictly ending with `/` and with Roslyn's robust handling/normalization of paths vs the more naive handling htmlgenerator has.

_Originally posted by @tannergooding in https://github.com/dotnet/source-indexer/pull/296#discussion_r3863391249_

No source file or test is named. Start by tracing how pathmap and serverpath are handled, then compare the existing behavior with Roslyn path normalization and htmlgenerator URL handling. Done means the conflict and path-normalization expectations are explicitly determined and verified.

Written by the indexing model from the issue text.