The following is information about what major changes have gone into each release.
Now builds on GHC 7.10.*.
Attribute changes:
Added the NoTranslate attribute.
Add support for overlines in HTML labels.
Both of these are available as of Graphviz 2.38.0, but were somehow overlooked in previous releases of this library.
Uses new fgl-arbitrary library for testing rather than custom instances for FGL graphs.
Migrated to git; central repository now on GitHub.
Augmentation tests are no longer run by default. This avoids problems where Graphviz is not installed/configured properly.
Use .gv for temporary files instead of .dot.
transformers-0.4.*, by Sergei Trofimovich.Add support for Graphviz 2.32.0, 2.32.1, 2.34.0, 2.36.0 and 2.38.0:
WARNING: at least as of Graphviz-2.32.0, dot -Tcanon no longer produces Dot code that is in the format expected by the Canonical Dot representation in this library. As such, unless you’re very sure of your sources you should always use the Generalised representation for parsing (see also the new FromGeneralisedDot class mentioned below).
New attributes:
XDotVersion (as of 2.34.0)
InputScale (as of 2.36.0)
OverlapShrink (as of 2.36.0)
Changed attributes:
Aspect no longer available (as of 2.36.0)
New ModeType values for use with sfdp: SpringMode and MaxEnt.
Weight now takes a value of type Number, that explicitly distinguishes between Doubles and Ints.
FixedSize and Normalize now have their own types.
New Shapes: Star and Underline (as well as Square which seems to have been omitted from previous versions).
Other relevant changes:
XDot now takes an optional version. Note that this doesn’t have any effect on how graphviz works.
The default extension for Dot-graphs is now .gv rather than .dot to reflect Graphviz’s changed conventions.
Other changes to the API:
Add the FromGeneralisedDot class, which provides a semi-inverse to fromCanonical in DotRepr.
GraphID now uses Number rather than separate Int and Double constructors; this only matters if you manually constructed or de-constructed GraphID values (ToGraphID still works).
Add the ability to parse a Dot graph “liberally”: that is, if an Attribute doesn’t match the specification, then let it fall back to an UnknownAttribute. This is still experimental, and requires more manual usage than the in-built commands (e.g. it isn’t supported in the default round-tripping).
Now using the definition of bracket from polyparse >= 1.9.
Monadic representation now has Functor and Applicative instances to satisfy the up-coming changes in GHC 7.10.
Compilation time has been reduced in two ways:
The Data.GraphViz.Attributes.Complete module has been split up (but still exports the same API, so no need to import more modules). Whilst I haven’t measured it, this should also reduce memory requirements for compilation.
The testsuite now uses the library explicitly, and thus no longer needs to re-compile half the library.
Bug-fixes:
Double values are now longer parseable without double quotes if they have an exponential term (to better match the definition).
It is no longer assumed when round-tripping that dot -Tdot generates canonicalised Dot graphs.
The TestParsing script is now directly buildable by Cabal using the test-parsing flag (the resulting executable is called graphviz-testparsing). This is not made an actual test-suite as not all files found will be actual Dot graphs, and it’s known that it fails on some. Instead it’s meant to be used as an indication of how well this library parses “real-world” Dot code.
Add support for Graphviz-2.30.0:
New attributes:
AreaHead_LPLayerListSepLayerSelectTail_LPXLPBgColor, Color and FillColor now take a list of colors with optional weightings.
Layer handling now supports layer sub-ranges.
Added the voronoi-based option to Overlap.
Added the Striped and Wedged styles.
Updates to attributes and their values:
The following attributes have had their values changed to better reflect what they represent:
FontPath takes a Path value.
Layout takes a GraphvizCommand (and thus GraphvizCommand has been moved to Data.GraphViz.Attributes.Complete).
Added MDS to Model (which had somehow been overlooked).
Various attributes now have defaults added/updated/removed if wrong.
Removed the following deprecated attributes:
ShapeFile
Z
Now any value that has a defined default can be parsed when the Dot code just has attribute="" (which dot -Tcanon is fond of doing to “reset” attributes).
GraphvizCommand now defines Sfdp (which had somehow been overlooked up till now).
The canonicalise and related functions have been re-written; whilst there might be some cases where their behaviour does not fully match how dot -Tcanon and tred behave (due to the interaction of various attributes), the new implementation is much more sane.
Use temporary files rather than pipes when running dot, etc.
Makes it more portable, and also avoids issues where dot, etc. use 100% CPU when a graph is passed in via stdin.
Original patch by Henning Thielemann.
renderCompactDot now outputs to just one line (which means it’s shorter, but more difficult to read for a human).
The QuickCheck instances are a bit smarter now to avoid having such ugly huge outputs when a test fails.
More documentation to help you understand what certain data types are for, etc.
canonicalise and related functions did not properly deal with attributes of a single node or edge within a sub-graph.The repository is now hosted on hub.darcs.net, and using the bug-tracker there.
Updates to various Attribute definitions:
The list of available shapes has been expanded to take into account the new synthetic biology shapes.
The Size and FontNames attributes now have specified data types.
The optional start and end points for Splines were previously the wrong way round; this has now been fixed.
Explicitly only print 2-dimensional Point values for Rect and DPoint (previously only 2-dimensional values where parsed, but it was possible to print a 3-dimensional value).
HTML-like labels previously disallowed empty textual label components when parsing; spotted by Daniel Hummel.
GraphvizParams now lets you specify whether a “cluster” is actually a cluster or a sub-graph. Requested by Gabor Greif.
Fixed an error where printing edges whose attributes contained a ColorScheme attribute, that attribute would stay in the state and the old color scheme would override the parent’s state.
Previously, some malformed attributes were accepted when being parsed as they silently became parsed as an UnknownAttribute. Now, attributes where the attribute name and the equal sign are successfully parsed but the value of the attribute is not successfully parse will throw an unrecoverable error.
Note: this does mean that some Dot graphs that are accepted by Graphviz (as they separate tokenizing from parsing; as such something like "0.1" is successfully accepted as an integer, specifically 0) are no longer accepted when parsing them in.
Miscellaneous parsing improvements:
Whilst not specified anywhere, Graphviz seems to treat empty quotes as values for attributes expecting a number as their default; as such, this is now taken into account when parsing.
DPoint values can now parse an optional + prefix.
Whitespaces after commas in HSV colors are now accepted.
Error messages from parsing are improved to help you track down where the parsing error occurred (in that it’s easier to find which attribute failed to parse, etc.).
Some parsing combinators are no longer publicly accessible, but shouldn’t be needed by end users anyway (as they were there for parsing the Attribute data type).
isGraphvizInstalled and quitWithoutGraphviz for programs to test whether Graphviz has been installed (as previous attempts at doing so via actual calls to dot, neato, etc. weren’t working and would be too verbose anyway).Labellable for strict Text values, as requested by Erik Rantapaa.String instance for ToGraphID requires the TypeSynonymInstances pragma (at least on GHC < 7.4). Spotted by Gregory Guthrie.Added support for the osage and patchwork visualisation tools, available as of Graphviz-2.28.0.
Updated attributes as of Graphviz-2.28.0:
SVG colors are now supported, and the support for different colors has been revamped.
overlap=false is now equivalent to overlap=prism and the RemoveOverlaps option has been removed.
LabelScheme and Rotation are new attributes for use with sfdp.
Scale is a new attribute for use with twopi.
Add new italics, bold, underline, superscript and subscript options for HTML-like labels.
LHeight and LWidth for getting the height and width of the root graph and clusters.
Updated attributes from the current development branch of Graphviz (i.e. 2.29.*). Please note that these will probably not work yet, but are implemented now for future-proofing.
A new style for edges: Tapered.
XLabel allows you to specify labels external to a node or edge. ForceLabels allow you to specify that these should be drawn even when they will cause overlaps.
ImagePath allows you to specify where to search for images.
HTML-like labels now support ID values as well as horizontal and vertical rules.
BgColor and FillColor now take a list of colors: this allows gradient fills for graphs, clusters and nodes. The Radial style and GradientAngle are also used for this purpose.
FillColor is now used by edges to set the color of any arrows.
WebP output support added.
Other attribute changes:
Use a specified data type for the Ordering attribute rather than an arbitrary Text value, and provide a documented wrapper in Data.GraphViz.Attributes.
Bb has been renamed BoundingBox.
ID now only takes EscString (a type alias for Text) values rather than arbitrary Labels.
The Data.GraphViz.Attributes.HTML module has had all values re-named and is now meant to be imported qualified. It is also no longer re-exported from Data.GraphViz.Attributes.Complete.
The ToGraphID class provides a common wrapper to help create cluster identifiers, etc.
Cabal’s Test-Suite functionality is now used. As part of this, the Data.GraphViz.Testing module and sub-modules are no longer exported by the library.
The new Benchmark support in Cabal-1.14 is now used for the benchmark script.
Dropped support for base-3.
The Data.GraphViz.State module is no longer exposed, as there’s no need for users to use it.
Bugfixes:
Some corner cases in canonicalisation prevented it from being idempotent.
The TestParsing script will no longer crash and refuse to continue if an IO-based error (e.g. unable to successfully call dot) occurs.
A typo was spotted by Gabor Greif.
Parsing error messages have been cleared up, especially when parsing Dot graphs. This came about from trying to help Uri J. Braun with an error in his code (which took a while to diagnose as a problem with the node type).
Made more clear to people looking in Data.GraphViz who want to create Dot graphs by hand to look in Data.GraphViz.Types (came about when Rustom Mody stated he got confused trying to work out how to do this).
Fixed up augmentation; bug and fix spotted by Max Rabkin.
Fix up the TestParsing script to actually use the new type setup. It can also now take a single directory as an argument, and will try to parse all (non-recursive) files in that directory.
String -> Text conversion or adding new modules and not checking their documentation thoroughly enough).Many large-level changes were done in this release; in rough categories these are:
Can now more easily create Dot graphs from other graph-like data structures with the new graphElemsToDot function, which takes a list of nodes and edges.
It is now no longer possible to use graphToDot, etc. to create Dot graphs with anonymous clusters; all clusters must now have explicit names (note that uniqueness is not enforced, and it is still possible to directly create Dot graphs with anonymous clusters).
The canonical graph representation has been moved to its own module: Data.GraphViz.Types.Canonical.
The generalised representation has had all its “G” and “g” prefixes removed.
Two new representations:
Data.GraphViz.Types.Graph allows graph-like manipulation of Dot code.
Data.GraphViz.Types.Monadic provides a monadic interface into building relatively static graphs, based upon the dotgen library by Andy Gill.
The DotRepr class has been expanded, and three pseudo-classes have been provided to reduce type-class contexts in type signatures.
Pure Haskell implementations of dot -Tcanon and tred are available in Data.GraphViz.Algorithms.
A new module is available for more direct low-level I/O with Dot graphs, including the ability to run custom commands as opposed to being limited to the standard dot, neato, etc.
Data.GraphViz.Attributes now contains a slimmed-down recommended list of attributes; the complete list can still be found in Data.GraphViz.Attributes.Complete.
The charset attribute is no longer available.
Functions for specifying custom attributes (for pre-processors, etc.) are available.
Now uses Text values rather than Strings. Whilst performing this migration, the improvements in speed for both printing and parsing improved dramatically.
prettyPrint and prettyPrint' functions have been removed.Now uses state-based printing and parsing, so that things like graph directedness, layer separators and color schemes can be taken into account.
Parsing large data-types (e.g. Attributes) now uses less back-tracking.
Now has a benchmarking script for testing printing and parsing speed.
Uses a custom exception type rather than a mish-mash of error, Maybe, Either, exception types from used libraries, etc.
Usage of UTF-8 is now enforced when doing I/O. If another encoding is required, the Text value that’s printed/parsed has to be written/read from disk/network/etc. manually.
The Rects Attribute should be able to take a list of Rect values; error spotted by Henning Gunther.
In some cases, global attribute names were being printed without even an empty list (which doesn’t match what dot, etc. expect).
Addition of the Labellable class (and its method toLabel) to make it easier to construct labels.
Backslashes (i.e. the \ character) are now escaped/unescaped properly (bug spotted by Han Joosten). As part of this:
Dot-specific escapes such as \N are now also handled correctly, so the slash does not need to be escaped.
Newline ('\n') characters in labels, etc. are escaped to centred-newlines in Dot code, but not unescaped.
Point values can now have the optional third dimension and end in a ! to indicate that that position should be used (as input to Graphviz).
LayerList uses LayerID values, and now has a proper shrink implementation in the test suite.
Fixed a mistake in one of the source files that was made just to make haskell-src-exts happier.
Fix the Arbitrary instance for Point in the testsuite (since there’s only one constructor now).
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 DotNodes are, etc.
graphNodes now returns DotNodes defined only as part of DotEdges, and will also merge duplicate DotNodes 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 Ints; this turns out not to be the case.
SortV and PrismOverlap now only take Word16 values rather than Ints, 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.
graphviz now has an FAQ and an improved README as well as its own homepage.
Canonicalisation of DotRepr values is now available with the canonicalise function.
Add support for record labels; values are automatically escaped/unescaped. The Record and MRecord shapes have been added for use with these labels. Requested by Minh Thu and Eric Kow.
Add support for HTML-like values (this replaces the wrong and completely broken URL datatype). Strings are automatically escaped/unescaped.
Named PortPos values are now accepted (as required for record and HTML-like labels).
GraphID no longer allows HTML-like values (since Graphviz doesn’t seem to allow it).
RankSep takes a list of Double values as required.
Attribute has a new constructor UnknownAttribute for use when parsing deprecated Graphviz attributes in old Dot code.
Various parsing fixes; of special note:
Statements no longer need to end in a semi-colon;
Anonymous sub-graphs are now supported.
Edge statements can now handle node groupings (e.g. a -> {b c}) as well as portPos values (e.g. a:from -> b:to).
Unquoted Strings containing non-ASCII characters are now parsed properly (though they are assumed to be encoded with UTF-8). Thanks to Jules Bean (aka quicksilver) for working out how to do this.
More specifically: almost all Dot files that ship with Graphviz, as documentation in the Linux kernel, etc. are now parseable.
A new script to assist in testing whether “real-world” Dot graphs are parseable.
Slight performance increase when parsing: whereas parsing is done case-insensitively, the “correct” case is now checked by default which has a moderate affect on parsing times.
Split lines are now able to be handled when parsing.
Added support for generalised DotGraphs; this optional representation removes the restriction of ordering of Dot statements. This allows for greater flexibility of how to specify Dot code. As an offshoot from this, most relevant functions now utilise a new DotRepr class that work with both DotGraphs and the new GDotGraphs; this shouldn’t affect any code already in use.
With the prompting of Noam Lewis, the augmentation functions have been revamped in two ways:
Now contains support for multiple edges.
The ability to perform “manual” augmentation if greater control is desired.
Add a preview function to quickly render and visualise an FGL graph using the Xlib canvas.
Added a pseudo-inverse of the FGL -> Dot functions (a graph is created, but the original node and edge labels are unrecoverable).
The Printing and Parsing modules have been moved (from Data.GraphViz.Types to Data.GraphViz).
Reworked file-generating commands such that they return the filename of the created file if successful.
Updated and extended functions to interact with the Graphviz tools. This now includes:
a better listing of available output types;
distinguishing file outputs from canvas outputs;
ability to automatically add the correct file extension to file outputs
Return any errors if calling Graphviz failed rather than just printing them to stderr
Improved Color support:
fixed ColorScheme values
explicitly named X11 colors
conversion to/from values from the colour library
Removed OrientationGraph; problems with distinguishing it when parsing from node-based Orientation values; its functionality is duplicated by Rotate.
By default, the generated Dot code is now no longer indented; you can now use the prettyPrint functions in Data.GraphViz to produce readable Dot code.
Added a testsuite; this is buildable by building with --flags-test. Numerous printing and parsing bugs were picked up with this.
Remove some Shape aliases and change capitalisation of others.
Properly parse and print IDs of clusters.
Allow NodeCluster values to have node types different from the LNode they come from.
node, edge, graph, digraph, subgraph, and strict need to be quoted. Spotted by Kathleen Fisher.Potentially fixed the graphvizWithHandle bug; correct approach spotted by Nikolas Mayr.
Fixed up Parsing of various Attribute sub-values, especially Point and Spline (and hence Pos, which uses them).
Pre-process out comments and join together multi-line strings before parsing.
Properly parse Doubles like ".2".
A major re-write occured; these are the highlights:
Re-write parsing and printing of Dot code using the new ParseDot and PrintDot classes. This should finally fix all quoting issues, as well as leaving Show as the code representation for hacking purposes. As part of this, the Data.GraphViz.ParserCombinators module has been moved to Data.GraphViz.Types.Parsing.
Re-write the various Dot* datatypes in Data.GraphViz.Types. Sub-graphs/clusters are now their own entity rather than being part of DotNode and the Node ID type is now a type parameter rather than being just Int. Sub-graphs/clusters can now also be parsed.
The various conversion functions in Data.GraphViz now come in two flavours: the unprimed versions take in a Bool indicating if the graph is directed or not; the primed versions attempt to automatically detect this.
Add cluster support for the graph augmentation functions, as requested by Nikolas Mayr.
Allow custom arrow types as supported by GraphViz; as requested by Han Joosten.
Fixed a bug in HSV-style Color values where Int was used instead of Double; spotted by Michael deLorimier.
Properly resolved the situation initially spotted by Neil Brown: Matthew Sackman was following Dot terminology for an edge a -> b when using head for b and tail for a (this is presumably because the head/tail of the arrow are at those nodes). DotEdge now uses from and to avoid ambiguity; the various Attribute values still follow upstream usage.
Color values were double-quoted.Color Attribute should take [Color], not just a single Color.Either for composite Attributes and use a custom type: this avoids problems with the Show instance.Fixed a bug where the Show instance and read function for DotEdge had the from/to nodes the wrong way round. This was not immediately noticed since the Graph to DotGraph functions created them the wrong way round, so for those users who only used these this was not apparent. Spotted by Neil Brown.
Greatly improved Attribute usage: almost all attributes are now covered with allowed values.
Extend DotGraph to include whether a graph is strict or not and if it has an ID. Also move the directedGraph field.
Make Dot refer to the actual dot command and DotArrow refer to the ArrowType (rather than DotCmd and Dot as before).
Make the Data.GraphViz.ParserCombinators module available to end users again, but not re-exported by Data.GraphViz; it has a warning message up the top not to be used. It is there purely for documentative purposes.
Use extensible-exceptions so that base-3.x is once again supported.
New maintainer: Ivan Lazar Miljenovic.
Support polyparse >= 1.1 (as opposed to < 1.3)
Require base == 4.* (i.e. GHC 6.10.*) due to new exception handling.
Include functions from Graphalyze-0.5 for running GraphViz commands, etc.
Module re-organisation.
The Data.GraphViz.ParserCombinators module is no longer available to end users.
Improved Haddock documentation.
Differentiate between undirected and directed graphs (previously only directed graphs were supported).
Clustering support was added.
For versions of graphviz older than 2008.9.20, the exact differences between versions is unknown.