Fixes the exclusion filter, so that the java files are not included in the main jar. Added a sources jar generation. Fixed formatting (mixed spaces and tabs -> tabs). Specified plugin versions for a repeatable build. Optional javadoc generation (by an explicit mvn javadoc:jar command). Added a .gitignore for Maven/Eclipse output & Eclipse project files.

I've migrated build to Maven, and added Travis configuration. I've removed doc/api and lib. lib/jgraphx.jar can be shared via Maven Central. javadoc could be served either via Maven Central (regular javadoc integration in IDEs) or via separate branch.

Note: maven coordinates are com.github.vlsi.mxgraph:* as I'm testing it via my own account. Note: readme should be improved to account maven build instructions, release instructions, example execution instructions.

You can see Travis CI here: https://travis-ci.org/vlsi/jgraphx (current javadoc fails under java 8)

Add targets to create artifacts, install them to local repository and deploy them to remote repository. Those targets are not called by the default build.

Requires Maven Ant Task (http://maven.apache.org/ant-tasks/), copy maven-ant-task jar in lib directory of Ant installation.

This adds an Automatic-Module-Name entry with value com.mxgraph.jgraphx to the jar manifest. This provides a stable name upon which Java 9 modules can depend in modular projects.

See http://blog.joda.org/2017/05/java-se-9-jpms-automatic-modules.html

Fix #91

This builds on top of https://github.com/jgraph/jgraphx/pull/82 Adds more logging with adequate levels. We're using the build with these changes in a server environment, it's fine - it doesn't spam the log or something. Also tested with the java GraphEditor example, it's fine too.

Added a shape style called mxLoadedImageShape, that allows the use of an image that is loaded into memory. Instead of mxImageShape, which requires the path to an image stored on file.

Hey guys,

What's up. I did a minor contribution adding a new graphical shape (double rectangle). I did it because I need in my project. Please take my contribution if you think it is interesting.

Cheers Anderson

The Nullpointer also happens in other cases. To trace it checking the canvas to be null is a first start to have a handle for debugging and be able to work on the problem in surrounding code without having to catch a NullPointer Exception. The check for the svgdoc to be null seems more appropriate to me.

Changes were ported from mxgraph/javascript/src/js/layout/hierarchical/mxHierarchicalLayout.js. I don't really understand this layouting algorithm, so please double-check my code.

Fix is mainly located in traverse function, but I also found a difference in filterDescendants function between Java and JS versions.

You can find sample XML graphs that are causing endless loop in issue #71 description

I'd like to validate a dash pattern that I'm giving to JGraph, and it makes most sense for this validation to be done inside JGraph in case the number separator (currently space) or number type (currently float) changes in the future (the former is not available as a style constant).

It looks like com.mxgraph.canvas.mxGraphics2DCanvas could also be refactored to use this mxUtils.parseDashPattern https://github.com/jgraph/jgraphx/blob/774aa8d63b178ea7995875902ac018600f4e554f/src/com/mxgraph/canvas/mxGraphics2DCanvas.java#L518 https://github.com/jgraph/jgraphx/blob/774aa8d63b178ea7995875902ac018600f4e554f/src/com/mxgraph/canvas/mxGraphicsCanvas2D.java#L355

I was not sure what behavior mxUtils.parseDashPattern should take on when the dash pattern string is null or empty string, whether it return the JGraph default dash pattern (https://github.com/jgraph/jgraphx/blob/1c4000edd211923136ea4f11e8554ed1696cc030/src/com/mxgraph/canvas/mxGraphicsCanvas2D.java#L1650), an empty float array, null, or throw an exception.

I am sorry I did not test my changes from the command line. Since jgraphx uses a non standard source directory "src" instead of "src/main/java" it is not possible to keep the test sources under the standard "src/test/java" - i have moved them to test/java instead and added the directory declaration in the pom file. Also the test case would fail with the null assignment - since that is only a workaround I made it succeed for the time being until the underlying problem is really fixed.

This pull request is based on the original master i forked from i do not know whether you can merge it directly. Please let me know if i should update my fork for a pull request that only has the differences to the already merged version.

-------------------------------------------------------
 T E S T S
-------------------------------------------------------
Running com.mxgraph.util.TestMxCellRenderer
Tests run: 1, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 0.508 sec

Results :

Tests run: 1, Failures: 0, Errors: 0, Skipped: 0