[Haskell-cafe] cabalizing "legacy' source

[Henning Thielemann]

Vasili I. Galchin wrote:
> Hello,
> 
>     I have cabalizing some "legacy" Haskell source. By "legacy", I 
> totally don't mean in a pejorative sense ... code has been designed and 
> written well. I am making good progress ... I have succeeded in building 
> an archive file(library) on a POSIX platform, i.e. Linux. The original 
> author intended to build several executables from this .a file. In 
> addition there are many pairings of .hs/.lhs along with "main" test 
> files. Currently I have only one .cabal file. This is inadequate for the 
> author's intent:
> 
> 1) I need to build "n" test executables.
> 
> 2) I need to build 1/2 CLI executables.
> 
> I need advice on the cabal front. E.g. should I first run the library 
> .cabal and then run several subsidary .cabal files, e.g. "n"test .cabals 
> and also 1/2 CLI executable .cabals??? I am looking for advice so that I 
> don't have a clumsy cabal build process/tree!

You can build several executables from one cabal file. I propose to 
enable compilation of test programs only by user request. See for example:
    http://code.haskell.org/~thielema/tagchup/tagchup.cabal