Hello Mike An option I like is to put the main project files in a top level directory called src, so your organization would look like this: ./src/Algo/Fux.hs ./src/Basics/CSound.hs ./src/Basics/Node.hs I usually have a running example which has imports for all the files in the project inside a top level folder called demo... ./demo/ImportAll.hs For a GHCi session, I cd to $PROJECT/demo then
ghci
Prelude> set -i../src This is the useful bit of having all the project files under the src hierarchy - the search path is very simple. During development adding the files needed for cabal (Setup.hs, project.cabal...) is still worthwhile - you can then generate Haddock docs very easily. If you used Haddock in standalone mode you would have to supply a few flags to tell which files to document and where to put the output. Each time you add or remove a module in the src tree, you should re-run configure...
runhaskell Setup.hs configure runhaskell Setup.hs haddock
but as your GHCi session is always using the interpreted project source you don't need to run 'runhaskell Setup.hs build' and 'runhaskell Setup.hs install'. Best wishes Stephen