[Haskell-cafe] ANNOUNCE: graphtype — A simple tool to illustrate dependencies between Haskell types
While developing applications which deal with complex data it is crucial to know how exactly you manipulate this data. Haskell provides excellent tools for expressing a data scheme you work with: ADTs, `type` and `newtype` declarations, type classes and much more is hidden in rich Haskell's type system. Obviously, when types of data in your domain you work with grow — all declarations grow, and it becomes hard to grasp all dependencies, to change them and to remove them deliberately. graphtype was developed to visualise type declarations in you Haskell source files. It produces .dot-file for subsequent processing with graphviz. Results for example file bundled with graphtype: http://i.piccy.info/i4/00/90/bfa07290012c2d3b455696bdaa86.png To play with it, you can use hackage: http://hackage.haskell.org/package/graphtype or hack some code: http://github.com/explicitcall/graphtype Visualisation of dependencies in complex type class hierarchies is still on the way. It isn't obvious how do to this nicely, as in most cases type class declarations are imported from other libraries, and you don't always have source files for them. Anyway, graphtype is fairly usable. Leave here your questions, suggestions and have fun looking at type dependencies in your code. WBR, Max ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
Re: [Haskell-cafe] ANNOUNCE: graphtype — A simple tool to illustrate dependencies between Haskell types
2009/8/24 Max Desyatov explicitc...@googlemail.com: graphtype was developed to visualise type declarations in you Haskell source files. It produces .dot-file for subsequent processing with graphviz. Anyway, graphtype is fairly usable. Leave here your questions, suggestions and have fun looking at type dependencies in your code. Neat. You could probably get some leverage from the GHC API for reading .hi files to find out information about imported types. It looks to me like you generate one image file for the whole graph. It could get quite big. I think dot supports hyperlinks, and so do some image formats (SVG I believe). Maybe you could split it up into pieces with hyperlinks between them. Browser support for SVG appears to be getting better these days. I've sometimes mused about the idea of graphing the static function call graph of programs and annotating arcs with type information. For that the GHC API would be the way to go (might need to do a little type reconstruction along the way to figure out the concrete types at which polymorphic functions are used.) Cheers, Bernie. ___ Haskell-Cafe mailing list Haskell-Cafe@haskell.org http://www.haskell.org/mailman/listinfo/haskell-cafe
