I'm pleased to announce a new version of my graphviz [1] library that provides bindings to the Graphviz [2] suite of tools for visualising graphs. [1]: http://hackage.haskell.org/package/graphviz [2]: http://graphviz.org/ Changes in this release: * I followed my own advice and put in bounds on the version of FGL it used... * Conversion of FGL-style graphs to `DotRepr` values is now achieved using the new `GraphvizParams` configuration type. This allows us to define a single parameter that stores all the conversion functions to pass around rather than having to pass around several functions. This also allows all the non-clustered and clustered functions to be collapsed together, whereas what used to be handled by the primed functions is now achieved by using the `setDirectedness` function. There are three default `GraphvizParams` available: - `defaultParams` provides some sensible defaults (no attributes or clustering). - `nonClusteredParams` is an alias of `defaultParams` where the clustering type is explicitly set to be `()` for cases where you don't want any clustering at all (whereas `defaultParams` allows you to set your own clustering functions). - `blankParams` sets all fields to be `undefined`; this is useful for situations where you have functions that will set some values for you and there is no sensible default you can use (mainly for the clustering function). * Expansion of the `DotRepr` class: - More common functions are now defined as methods (`getID`, etc.). - The ability to get more information about the structure of the `DotRepr` graph, as well as where all the `DotNode`s are, etc. - `graphNodes` now returns `DotNode`s defined only as part of `DotEdge`s, and will also merge duplicate `DotNode`s together. - `graphNodes` and `graphEdges` also return `GlobalAttributes` that apply to them. * The `Point` type now only has one constructor: `Point Double Double`. The `Int`-only constructor was present due to historical purposes and I thought that the `Pos` value for a `DotNode` would always be a pair of `Int`s; this turns out not to be the case (and accounted for some bugs in older versions of my SourceGraph tool; it's taken me this long to remember to fix this...). * `SortV` and `PrismOverlap` now only take `Word16` values rather than `Int`s, as they're not meant to allow negative values (the choice of using `Word16` rather than `Word` was arbitrary, and because it's unlikely those values will be large enough to require the larger values available in `Word`). * `NodeCluster` has been generalised to not have to take an `LNode` for the node type; the type alias `LNodeCluster` is available if you still want this. * Several documentation typos fixed, including one spotted by Kevin Quick. * The test-suite now allows running individual tests. -- Ivan Lazar Miljenovic Ivan.Miljenovic@gmail.com IvanMiljenovic.wordpress.com