graphviz
Bindings to Graphviz for graph visualisation.
https://github.com/ivan-m/graphviz
Version on this page: | 2999.20.1.0@rev:3 |
LTS Haskell 22.37: | 2999.20.2.0 |
Stackage Nightly 2024-10-10: | 2999.20.2.0 |
Latest on Hackage: | 2999.20.2.0 |
graphviz-2999.20.1.0@sha256:023942db07ae923d654fb093614ab07ee0fb82eb4c7bc39daf26fd611cb7559b,7935
Module documentation for 2999.20.1.0
- Data
- Data.GraphViz
- Data.GraphViz.Algorithms
- Data.GraphViz.Attributes
- Data.GraphViz.Commands
- Data.GraphViz.Exception
- Data.GraphViz.Internal
- Data.GraphViz.Internal.State
- Data.GraphViz.Internal.Util
- Data.GraphViz.Parsing
- Data.GraphViz.PreProcessing
- Data.GraphViz.Printing
- Data.GraphViz.Types
- Data.GraphViz.Types.Canonical
- Data.GraphViz.Types.Generalised
- Data.GraphViz.Types.Graph
- Data.GraphViz.Types.Internal
- Data.GraphViz.Types.Internal.Common
- Data.GraphViz.Types.Monadic
- Data.GraphViz
% Haskell bindings to the Graphviz toolkit % Ivan Lazar Miljenovic
The graphviz Library
The graphviz library provides bindings to the Graphviz graph visualisation suite of tools for the purely functional programming language Haskell. It can be downloaded from HackageDB or - if you have cabal-install - installing it is as simple as:
cabal update
cabal install graphviz
Library features
Main features of the graphviz library include:
-
Almost complete coverage of all Graphviz attributes and syntax.
-
Support for specifying clusters.
-
The ability to use a custom node type.
-
Functions for running a Graphviz layout tool with all specified output types.
-
The ability to not only generate but also parse Dot code with two options: strict and liberal (in terms of ordering of statements).
-
Functions to convert FGL graphs and other graph-like data structures to Dot code - including support to group them into clusters - with a high degree of customisation by specifying which attributes to use and limited support for the inverse operation.
-
Round-trip support for passing an FGL graph through Graphviz to augment node and edge labels with positional information, etc.
graphviz is free software licensed under a 3-Clause BSD License.
(C) 2008 Matthew Sackman
(C) 2008 - onwards Ivan Lazar Miljenovic
For more information, feel free to email me.
Changes
% Changelog % Ivan Lazar Miljenovic
Release History and Changelog
The following is information about what major changes have gone into each release.
Changes in 2999.20.1.0
-
Add
MonadFix
instance forDotM
(thanks to George Wilson) -
Fix exception catching for missing executables (thanks to Kostas Dermentzis)
-
Dependency bumps.
Changes in 2999.20.0.4
- Dependency bumps.
Changes in 2999.20.0.3
- Dependency bumps.
Changes in 2999.20.0.2
-
Fix Haddock issue (thanks to Moritz Kiefer)
-
Bump HSpec upper bound (thanks to Moritz Kiefer)
-
Make Hackage happier with Cabal-Version field
Changes in 2999.20.0.1
- Allow building with temporary-1.3.*.
Changes in 2999.20.0.0
-
Can now create subgraphs using the Monadic representation.
-
Allow unescaped
-
and'
in HTML labels (thanks to Andrey Kartashov) -
Support for strict
Text
instances for printing/parsing. -
Creating a Graph representation with
mkGraph
was not adding edges correctly (reported by Joshua Chia). -
Test suite now uses HSpec, making it a lot easier to add tests for specific issues.
-
Builds with GHC 8.4.* (thanks to Tony Day).
-
Monoid and Semigroup instances for Monadic representation (thanks to Chris Martin).
-
Be more lenient in parsing some attributes (e.g. allow
top
instead of justt
forVerticalPlacement
). -
Add new HTML attributes:
Columns
,GradientAngle
,Rows
,Sides
andStyle
. -
Improve/update the TestParsing executable to work on being able to parse all the sample
Dot
graphs shipped with Graphviz. -
Bump dependencies.
Changes in 2999.19.0.0
-
Roll back change in 2999.18.1.0 for Monadic graphs, as they turn out to not actually work in practice (reported by Lennart Spitzner).
-
Add a
quickParams
value to help with testing graphs in ghci (requested by Ian Jeffries). -
Fix parsing of edge chains (reported by Jonas Collberg).
-
Fix how seemingly numeric text literals are quoted (reported by Joey Hess).
Changes in 2999.18.1.2
-
Allow dlist-0.8, thanks to Sean Leather.
-
Properly allow QuickCheck-2.9.
Changes in 2999.18.1.1
- Bump allowed version of QuickCheck for the test suite.
Changes in 2999.18.1.0
-
Monadic graphs can now have edges to and from lists of nodes, as requested by Harold Carr.
-
Compatible with GHC-8.* pre-releases.
- Please note that there is currently a warning emitted on compilation by GHC-8.*; this has been reported here.
Changes in 2999.18.0.2
- Match new behaviour of
dot
, etc. in supporting edge specifications likea, b -> c, d
(spotted by Dilawar Singh).
Changes in 2999.18.0.1
- Loosened bound on
fgl-arbitrary
.
Changes in 2999.18.0.0
-
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.
Changes in 2999.17.0.2
-
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
.
Changes in 2999.17.0.1
- Allow usage of
transformers-0.4.*
, by Sergei Trofimovich.
Changes in 2999.17.0.0
-
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 newFromGeneralisedDot
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 withsfdp
:SpringMode
andMaxEnt
. -
Weight
now takes a value of typeNumber
, that explicitly distinguishes betweenDoubles
andInts
. -
FixedSize
andNormalize
now have their own types. -
New
Shape
s:Star
andUnderline
(as well asSquare
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 tofromCanonical
inDotRepr
. -
GraphID
now usesNumber
rather than separateInt
andDouble
constructors; this only matters if you manually constructed or de-constructedGraphID
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 anUnknownAttribute
. 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
frompolyparse >= 1.9
. -
Monadic representation now has
Functor
andApplicative
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 thetest-parsing
flag (the resulting executable is calledgraphviz-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.
Changes in 2999.16.0.0
-
Add support for Graphviz-2.30.0:
-
New attributes:
Area
Head_LP
LayerListSep
LayerSelect
Tail_LP
XLP
-
BgColor
,Color
andFillColor
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
andWedged
styles.
-
-
Updates to attributes and their values:
-
The following attributes have had their values changed to better reflect what they represent:
-
FontPath
takes aPath
value. -
Layout
takes aGraphvizCommand
(and thusGraphvizCommand
has been moved toData.GraphViz.Attributes.Complete
).
-
-
Added
MDS
toModel
(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=""
(whichdot -Tcanon
is fond of doing to “reset” attributes). -
GraphvizCommand
now definesSfdp
(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 howdot -Tcanon
andtred
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.
Changes in 2999.15.0.1
- Fixed bug where
canonicalise
and related functions did not properly deal with attributes of a single node or edge within a sub-graph.
Changes in 2999.15.0.0
-
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
andFontNames
attributes now have specified data types. -
The optional start and end points for
Spline
s were previously the wrong way round; this has now been fixed. -
Explicitly only print 2-dimensional
Point
values forRect
andDPoint
(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, specifically0
) 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).
Changes in 2999.14.1.0
- Add
isGraphvizInstalled
andquitWithoutGraphviz
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).
Changes in 2999.14.0.0
- Added an instance of
Labellable
for strictText
values, as requested by Erik Rantapaa.
Changes in 2999.13.0.3
- Allow usage of transformers-0.3.*. Spotted by Peter Simons.
Changes in 2999.13.0.2
- The
String
instance forToGraphID
requires theTypeSynonymInstances
pragma (at least on GHC < 7.4). Spotted by Gregory Guthrie.
Changes in 2999.13.0.1
- Fixed Haddock typos.
Changes in 2999.13.0.0
-
Added support for the
osage
andpatchwork
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 tooverlap=prism
and theRemoveOverlaps
option has been removed. -
LabelScheme
andRotation
are new attributes for use withsfdp
. -
Scale
is a new attribute for use withtwopi
. -
Add new italics, bold, underline, superscript and subscript options for HTML-like labels.
-
LHeight
andLWidth
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
andFillColor
now take a list of colors: this allows gradient fills for graphs, clusters and nodes. TheRadial
style andGradientAngle
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 arbitraryText
value, and provide a documented wrapper inData.GraphViz.Attributes
. -
Bb
has been renamedBoundingBox
. -
ID
now only takesEscString
(a type alias forText
) values rather than arbitraryLabel
s. -
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 fromData.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, theData.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 calldot
) occurs. -
A typo was spotted by Gabor Greif.
-
Changes in 2999.12.0.4
-
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 inData.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.
Changes in 2999.12.0.3
- Fixes various mistakes in the Haddock documentation that slipped
through (either in the
String -> Text
conversion or adding new modules and not checking their documentation thoroughly enough).
Changes in 2999.12.0.2
- Forgot to explicitly list the module for Arbitrary instance for the graph representation.
Changes in 2999.12.0.1
- Fix a bug that prevented graphviz from being built with GHC 7.2.
Changes in 2999.12.0.0
Many large-level changes were done in this release; in rough categories these are:
Conversions from other types
-
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).
Dot graph representations
-
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.
Using and manipulation Dot graphs
-
Pure Haskell implementations of
dot -Tcanon
andtred
are available inData.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.
Attributes
-
Data.GraphViz.Attributes
now contains a slimmed-down recommended list of attributes; the complete list can still be found inData.GraphViz.Attributes.Complete
. -
The
charset
attribute is no longer available. -
Functions for specifying custom attributes (for pre-processors, etc.) are available.
Implementation
-
Now uses
Text
values rather thanString
s. Whilst performing this migration, the improvements in speed for both printing and parsing improved dramatically.- As part of this, human-readable Dot code is now produced by
default. As such, the
prettyPrint
andprettyPrint'
functions have been removed.
- As part of this, human-readable Dot code is now produced by
default. As such, the
-
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.
Bug-Fixes
-
The
Rects
Attribute
should be able to take a list ofRect
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).
Changes in 2999.11.0.0
-
Addition of the
Labellable
class (and its methodtoLabel
) 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
usesLayerID
values, and now has a propershrink
implementation in the test suite.
Changes in 2999.10.0.1
-
Fixed a mistake in one of the source files that was made just to make haskell-src-exts happier.
-
Fix the
Arbitrary
instance forPoint
in the testsuite (since there’s only one constructor now).
Changes in 2999.10.0.0
-
Conversion of
FGL
-style graphs toDotRepr
values is now achieved using the newGraphvizParams
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 thesetDirectedness
function.There are three default
GraphvizParams
available:-
defaultParams
provides some sensible defaults (no attributes or clustering). -
nonClusteredParams
is an alias ofdefaultParams
where the clustering type is explicitly set to be()
for cases where you don’t want any clustering at all (whereasdefaultParams
allows you to set your own clustering functions). -
blankParams
sets all fields to beundefined
; 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 theDotNode
s are, etc. -
graphNodes
now returnsDotNode
s defined only as part ofDotEdge
s, and will also merge duplicateDotNode
s together. -
graphNodes
andgraphEdges
also returnGlobalAttributes
that apply to them.
-
-
The
Point
type now only has one constructor:Point Double Double
. TheInt
-only constructor was present due to historical purposes and I thought that thePos
value for aDotNode
would always be a pair ofInt
s; this turns out not to be the case. -
SortV
andPrismOverlap
now only takeWord16
values rather thanInt
s, as they’re not meant to allow negative values (the choice of usingWord16
rather thanWord
was arbitrary, and because it’s unlikely those values will be large enough to require the larger values available inWord
). -
NodeCluster
has been generalised to not have to take anLNode
for the node type; the type aliasLNodeCluster
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.
Changes in 2999.9.0.0
-
graphviz now has an FAQ and an improved README as well as its own homepage.
-
Canonicalisation of
DotRepr
values is now available with thecanonicalise
function. -
Add support for record labels; values are automatically escaped/unescaped. The
Record
andMRecord
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 ofDouble
values as required. -
Attribute
has a new constructorUnknownAttribute
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 asportPos
values (e.g.a:from -> b:to
). -
Unquoted
String
s 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.
Changes in 2999.8.0.0
-
Added support for generalised
DotGraph
s; 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 newDotRepr
class that work with bothDotGraph
s and the newGDotGraph
s; 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 theXlib
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
andParsing
modules have been moved (fromData.GraphViz.Types
toData.GraphViz
). -
Reworked file-generating commands such that they return the filename of the created file if successful.
Changes in 2999.7.0.0
-
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-basedOrientation
values; its functionality is duplicated byRotate
. -
By default, the generated Dot code is now no longer indented; you can now use the
prettyPrint
functions inData.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.
Changes in 2999.6.0.0
-
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 theLNode
they come from.
Changes in 2999.5.1.1
- When used as labels, etc., the Dot keywords
node
,edge
,graph
,digraph
,subgraph
, andstrict
need to be quoted. Spotted by Kathleen Fisher.
Changes in 2999.5.1.0
-
Potentially fixed the
graphvizWithHandle
bug; correct approach spotted by Nikolas Mayr. -
Fixed up
Parsing
of variousAttribute
sub-values, especiallyPoint
andSpline
(and hencePos
, which uses them). -
Pre-process out comments and join together multi-line strings before parsing.
-
Properly parse
Doubles
like".2"
.
Changes in 2999.5.0.0
A major re-write occured; these are the highlights:
-
Re-write parsing and printing of Dot code using the new
ParseDot
andPrintDot
classes. This should finally fix all quoting issues, as well as leavingShow
as the code representation for hacking purposes. As part of this, theData.GraphViz.ParserCombinators
module has been moved toData.GraphViz.Types.Parsing
. -
Re-write the various
Dot*
datatypes inData.GraphViz.Types
. Sub-graphs/clusters are now their own entity rather than being part ofDotNode
and the Node ID type is now a type parameter rather than being justInt
. 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 aBool
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 whereInt
was used instead ofDouble
; 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 forb
and tail fora
(this is presumably because the head/tail of the arrow are at those nodes).DotEdge
now uses from and to avoid ambiguity; the variousAttribute
values still follow upstream usage.
Changes in 2999.1.0.2
- Fix a bug spotted by Srihari Ramanathan where
Color
values were double-quoted.
Changes in 2999.1.0.1
- The
Color
Attribute
should take[Color]
, not just a singleColor
.
Changes in 2999.1.0.0
- Stop using
Either
for compositeAttributes
and use a custom type: this avoids problems with theShow
instance.
Changes in 2999.0.0.0
-
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 theGraph
toDotGraph
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 andDotArrow
refer to theArrowType
(rather thanDotCmd
andDot
as before). -
Make the
Data.GraphViz.ParserCombinators
module available to end users again, but not re-exported byData.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. -
Follow the Package Versioning Policy rather than using dates for versions.
Changes in 2009.5.1
-
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.
Changes in 2008.9.20
-
Differentiate between undirected and directed graphs (previously only directed graphs were supported).
-
Clustering support was added.
Older versions
For versions of graphviz older than 2008.9.20
, the exact differences
between versions is unknown.