as far as namespace is concerned what is division between a module's name and it's associated .cabal with it's Hs-Source-Dirs directive? This is kinda' absolute vs relative path I think.
Vasili, First, let's talk about Haskell modules, without mentioning cabal. When Haskell 98 standard came, modules had just a name, like, say, 'Complex'. You can still see those names in GHC library for compatibility: http://www.haskell.org/ghc/docs/latest/html/libraries Another interesting think you can see in that link is the separation of modules into 'packages'. Right to the name of every module there's a package name, like 'haskell98', 'base', 'stm-2.1.1.2' etc. GHC and others adopted the convention of using dot in module names to have a standard way of showing hierarchy between modules. There's also a convention used in GHC (and others?) of allowing only one module per file and, when looking for such file (as in --make) compose the search path with the module name, replacing dots by the start of a sub-directory name. So: suppose your search path (a colon separate list of directories) is '.:/hs:sub' and GHC wants a file for module Data.Our.Test. Then it looks for: ./Data.Our.Test /hs/Data.OurTest ./sub/Data.OurTest You can find details on that (like how to add to the search path) here: http://www.haskell.org/ghc/docs/latest/html/users_guide/separate-compilation... Now about .cabal file. When you use it, it helps you generate a package. In 'hs-source-dirs' you say what is the search path. In 'exposed-modules' you say wich modules are going to be seen by users of your package. You can also use 'other-modules' to name modules your package needs but are not going to be visible to other packages. Just say if anything is not clear. Best, Maurício