[new-hs-indexer #6] Workarounds to handle #included source files

[simonmar]

1. Ignore Names that come from another source file (TODO)
2. Ignore Names that come from the current source file but have a
   line number outside the valid range. No idea why this happens, I
   didn't investigate.

Note: stacked on #518

[netlify]

✅ Deploy Preview for fb-oss-glean canceled.

Name	Link
🔨 Latest commit	2c83bdb
🔍 Latest deploy log	https://app.netlify.com/projects/fb-oss-glean/deploys/6843e3b8e3080200084b249f

[CatherineGasnier]

Happy to review but not sure I have enough context here. Happy to also test in phabricator and approve if you think that's safe enough.

[simonmar]

@CatherineGasnier let me know if there's anything I can do to help.

[CatherineGasnier]

Ok some context questions I have:

• What are #included files in haskell? A quick search didn't yield anything
• Why do we need them? In what way is the proposed solution a workaround? (I guess I don't quite understand the description because I don't understand #included files)

Not sure it really matters for me to understand the details though. Meanwhile I'll go ahead an "import to fbsource" in the hope that it gets tested in phabricator.

[facebook-github-bot]

@CatherineGasnier has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.

[simonmar]

@CatherineGasnier

* What are #included files in haskell? A quick search didn't yield anything

In Haskell you can use the C preprocessor by adding {-# LANGUAGE CPP #-}, and then you can use #include to include other files at compile-time, like in C/C++. This causes an issue for the Haskell indexer because the .hie file generated by the compiler only has the original source file name and contents, it doesn't have any #included files.

Fortunately it's rare to use #include in Haskell, but I did come across this in the wild, in GHC itself: https://gitlab.haskell.org/ghc/ghc/-/blob/34e352173dd1fc3cd86c49380fda5a4eb5dd7aef/compiler/GHC/Builtin/PrimOps.hs#L147

* Why do we need them? In what way is the proposed solution a workaround? (I guess I don't quite understand the description because I don't understand #included files)

The solution is a workaround because it's just ignoring any symbols and references that come from a file that is not the primary source file. The right solution would be to track the other source files correctly.

Hope this helps!

[simonmar]

@CatherineGasnier note that this PR is stacked on #518 so you probably want to import the earlier PRs in the stack first.

[facebook-github-bot]

@bochko has imported this pull request. If you are a Meta employee, you can view this diff on Phabricator.