qrilka / haskell-navigation

Tools to ease Haskell source code navigation

Haskell navigation

This repository is an attempt to provide tools for easier Haskell source code navigation. It uses haskell-indexer project as a way to get information about source code.

Index a package

To index a package you need to do the following:

1. Build this project with stack build, so you'll get query-tool and kythe-to-call-graph

2. Build ghc_kythe_wrapper with stack build haskell-indexer-pipeline-ghckythe-wrapper (Stack currently doesn't build Github project packages by default)

3. Index your package with

   ./index.hs <PATH-TO-YOUR-PROJECT> <CALL-GRAPH-FILE-NAME>
   

   where <PATH-TO-YOUR-PROJECT> is a path pointing to your package directory containing a Stack project (i.e. stack.yaml) and <CALL-GRAPH-FILE-NAME> should point to a file where you'll store the resulting call graph

Please note that package indexing takes some time and requires a bit of space as currently it rebuilds a package "from scratch" starting from an empty $STACK_ROOT so any of required Stackage dependencies will get indexed.

Having done all of that you should be able to run query-tool with the call graph you've created.

haskell-navigation's Issues

CLI UI tool to navigate call graphs

Incremental indexing

Currently index.hs indexes all dependencies from scratch and this doesn't make sense when indexing multiple projects sharing some dependencies and also for reindexing later code versions

Output filnames for call locations

<Line 17, Column 8> should become /path/to/file.hs:17:8

index.hs leaves STACK_WORK directories in project package directories

The one in the project root gets deleted but we also need to take care of the ones in package directories

Check out importing call graph data from haskell-code-explorer

https://github.com/alexwl/haskell-code-explorer is the project.
Probably the main current counterargument is that the tool indexes only per package and including package dependencies requires extra work.