Hello Bonitasoft Team,

I have executted my process several time and I would like to extract my event logs for process analytics but don't know how to. Could you give a step by step guide? Please

From data create semantic bpmn

Is your feature request related to a problem? Please describe.
With v1.0.2, outgoing edges of a gateways overlap incoming edges.

Examples generated with https://github.com/process-analytics/bpmn-visualization-pm4py:

Describe the solution you'd like
For gateway, the layout of outgoing edges could always start with an horizontal line, then when needed to top or bottom then horizontal lines.

Example with diagram.bpmn.txt:

current	proposal

Describe alternatives you've considered
#58 may help

Additional context
This is the solution that is used by https://github.com/bpmn-io/bpmn-auto-layout
They apply the same to incoming edges of gateways as well

ref: 3.6 paper page 395

The paper mentions on p 398 the orxyx-editor as a bpmn vendor for integration of the algorithm

The next step is integrating the algorithm with the afore-
mentioned Oryx Editor in cooperation with the Hasso-
Plattner-Institut. This will allow everyone to evaluate the
algorithm or download its source code as part of the Oryx
source.

Original googlecode home: https://code.google.com/archive/p/oryx-editor/
This project served as base for the Signavio implementation

The new home for the source code is https://github.com/tiku01/oryx-editor (as of 2020-04-15, latest commit occurred on July 2015)
Let's review if the algorithm has been finally implemented there

This would allow to generate something like in A.4.1.bpmn file from BPMN-MIWG

This could be an option of the API/CLI: choose horizontal (default) or vertical layout

See https://www.eclipse.org/elk/

• Bonita Studio
• Camunda Model API, see #3

Compare with the solution that we are going to implement, in particular how they manage waypoints:
https://github.com/bpmn-io/bpmn-auto-layout/

waypoint for edge to link element

Is your feature request related to a problem? Please describe.
We already manage our R implementation with such a bridge. Having the same for nodejs will avoid having to reimplement the generator in JS.

Additional context
check the following resources as starting point and others

• https://www.npmjs.com/package/java
• https://www.geeksforgeeks.org/how-to-run-java-code-in-node-js/
• http://creatale.github.io/webdevbbq-java-nodejs/#/
• call java via spawn

Is your feature request related to a problem? Please describe.
In some complex layout, the edges overlap the shapes, which makes the diagram hard to read.

Example with the "Repair Example" generated after performing Process Discovery.

Describe the solution you'd like
Find a way to ensure that edges don't overlap edges.
#58 may help.

We already compute font size for each shape/edge label in our internal display model but don't use it in the BPMN export.

Need sorting elements to be able to position on the grid.

Is your feature request related to a problem? Please describe.
#43 will consider pools/participants. Then collaboration message flows won't be displayed.

Describe the solution you'd like
So, consider message flows when computing way points of the edges (sequence flows only for now).

The current implementation

• hard code the number of waypoints considered for the svg generation instead of using them all
• generate multiple svg lines. We could have a single svg path or a svg polyline

See

Hello, the algorithm is pretty good but it lacks considering some elements like pools, lane, and messageFlow. Is there any plan to update the algorithm to include these types of elements?

Currently, we only manage the subprocess as a single element because
For now, we don't even transform the elements from BPMN file into our internal model

In case a label is set on edge, we should position it. Usually, BPMN visualization put the label on the edge by default if no label bounds are provided.

In particular, decide if we want to match BPMN naming

BPMN

• Diagram
• Shape
• Edge

Current #4

• Diagram
• Node
• Edge

edge without waypoint.

We need to have something other than human task when importing data from the CSV

Right now all elements named "task" are imported as human task. It should be Abstract task instead and we should handle more speccific types

see Readme for references

See 3.8.1 of the paper

https://github.com/FHOOEAIST/bpmn-viz

BPMN visualization provides a functionality to generate the diagram interchange information for a given BPMN model.

The BPMN semantic usually use the 'merge' term.
So we should update the layout and waypoints implementation to use this term instead of 'join'.

Use https://github.com/process-analytics/bpmn-visualization-R when a public version is available.
See process-analytics/bpmn-visualization-R#10
In the meantime, a development version can be used by installing the package from github

Implement the logic to pass from the BPMN model to our internal model (used in the layout generator algorithms) and from our internal model to the BPMN model.

There is an issue when layouting cycles, we need to revert more the a single edge to break the cycle.
We also need to revert other edges to be able to layout elements correctly.
This is briefly described in the paper.

add "gateways" "automatic task" etc...

#52 introduced edge display in the svg export, but there is currently no arrow.
We probably will need the edge direction to know how to paint the arrow

The whole code in in the AlgoToDisplayModelConverter class which is very large.
The code is also procedural and could be improved for simplification.

See

package installation: the usage of a gh token is probably not needed

The 0.1.4 package documentation explains that a github auth token must be pass to install the package.
I don't understand why we need the token as all resources used by the package are public. If this is required, please document why.

package installation failure

devtools::install_github("process-analytics/bpmn-layout-generators", ref="bpmnLayoutGeneratoR-0.1.4", subdir="R/bpmnLayoutGeneratoR")

The package requires JDK to be installed on the machine.
Otherwise the rjava package installation failed

Example on Ubuntu 20.04

checking Java support in R... present:
interpreter : '/usr/lib/jvm/default-java/bin/java'
archiver    : '/usr/lib/jvm/default-java/bin/jar'
compiler    : '/usr/lib/jvm/default-java/bin/javac'
header prep.: ''
cpp flags   : '-I/usr/lib/jvm/default-java/include -I/usr/lib/jvm/default-java/include/linux'
java libs   : '-L/usr/lib/jvm/default-java/lib/server -ljvm'
checking whether Java run-time works... ./configure: line 3765: /usr/lib/jvm/default-java/bin/java: No such file or directory
no
configure: error: Java interpreter '/usr/lib/jvm/default-java/bin/java' does not work
ERROR: configuration failed for package ‘rJava’

rJava installation: http://rforge.net/rJava/
Java installation

• AdoptOpenJDK: https://adoptopenjdk.net/installation.html
• Adoptium: https://adoptium.net/installation.html

Once JDK is installed, don't forget to run R CMD javareconf. See https://cran.r-project.org/doc/manuals/R-admin.html#Java-support

Usage example in the readme --> NullPointerException

Note: confirmed on Ubuntu 20.04 by @tbouffard and on Windows 10 Pro by @oanesini

With package 0.1.4, R 3.6.3, AdoptOpenJDK 11.0.11

    Error in .jcall("RJavaTools", "Ljava/lang/Object;", "invokeMethod", cl,  :
      java.lang.NullPointerException

Same issue when running

• diagram <- bpmnLayoutGeneratoR::generateBpmnLayout(flow_node, sequence_flow, "BPMN")
• diagram <- bpmnLayoutGeneratoR::generateBpmnLayout(flow_node, sequence_flow)

as we are using the Java API and not the CLI, we should add a try catch bock when calling the 'layout generation from csv' method and at least print the stack trace

http://rforge.net/doc/packages/rJava/Exceptions.html

 tryCatch( diagramAsString <- bpmnLayoutJava$generateLayoutFromCSV(flow_node_as_csv, sequence_flow_as_csv, type), Exception = function(e){
     e$printStackTrace() 
 } )

When using the stack trace, we got

java.lang.NullPointerException
	at io.process.analytics.tools.bpmn.generator.input.CSVtoBPMN.assignIncomingAndOutgoingReferences(CSVtoBPMN.java:113)
	at io.process.analytics.tools.bpmn.generator.input.CSVtoBPMN.readFromCSV(CSVtoBPMN.java:51)
	at io.process.analytics.tools.bpmn.generator.BPMNLayoutGenerator.generateLayoutFromCSV(BPMNLayoutGenerator.java:48)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
	at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:62)
	at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
	at java.base/java.lang.reflect.Method.invoke(Method.java:566)
	at RJavaTools.invokeMethod(RJavaTools.java:386)

CSVtoBPMN.readFromCSV received string parameters as

nodes

"","id","node","type"
"1",1,"task1","task"
"2",2,"task2","task"

edges --> the issue is here, from and to id must be number, not string, because in nodes, they are defined as number. Both must match.
We need to add robustness to the Java implementation, when the referenced id in from/to doesn't exist in nodes. See #95

"","id","from_id","to_id"
"1","1","1","2"

Paper: https://www.researchgate.net/publication/335360755_Sketching_Process_Models_by_Mining_Participant_Stories
Demo:

• https://www.bpmn-sketch-miner.ai/# (js lib 1.17.0 on 2020-06-15)
• https://design.inf.usi.ch/bpmn-sketch-miner/# (js lib 1.15.4 on 2020-06-15)

Additional resources

• https://github.com/bpmn-sketch-miner
• https://github.com/USI-INF-Software/RESTfulConversationMining
• https://github.com/ivanchikj/BPMNTransformations

IMO it should contain:

• shapes (ids of bpmn elements)
• edges
• position of shapes
• waypoints ( to be added during implementation)

Right now we have:

• Diagram
• SortedDiagram
• Grid

ref: 3.6 paper page 395.

• change repository settings
• ensure the GH Actions are still working

Already covered by #97 and https://github.com/orgs/process-analytics/projects/1#card-69368599

Is your feature request related to a problem? Please describe.
Currently, the boundary events are considered as a isolated elements for the layout instead of being attached to the activity.

Describe the solution you'd like
A Boundary events should be attached to its parent activity

Additional context
Example with A.3.0.bpmn generated with PR #52 (the escalation boundary event is not displayed here as the screenshot has been done with [email protected] which does not support it)

original layout

release process

For now, we have remaining branches and git tags
we should be able to only keep the git tags as the package installation using install_github can use a git tag with the ref parameter (https://www.rdocumentation.org/packages/remotes/versions/2.4.0/topics/install_github)
If so, update the install procedure in the README

DESCRIPTION

The version, title, description and license are not or incompletely set.
license should be License: Apache License (== 2)

Notice that due to the current release process, the DESCRIPTION file is never updated in the master branch

development

The jar file is not available in the master branch because we don't want to mess the repo.
We should explain how to get it when developing the package. There is no git ignore for the jar that MUST not be commited during dev (only during release): error prone
Makefile: review PACKAGE_NAME=rJavaPackageExample

As for the release, run make install from the R/bpmnLayoutGeneratoR folder

see issue shown in #52

Describe the bug
See #91
We have a NPE without explanation, making the debugging very hard to to.

Expected behavior
Decide if we want to silently ignore the unknown ids, or consider as an error.
If we decide to throw an Exception, we should provide context for the end user.

ref 3.6 paper page 395

The build is currently run on Travis CI, we will easily manage this new requirement with GH Actions.
In addition, Travis CI limited the number of build minutes available even for Open Source projects.
So I suggest we move to GitHub Actions

• Configure the repository to use GH Token with readonly permissions by default
• Introduce GitHub workflow for java build
• Remove the Travis CI build
• Remove the Travis CI application from the organization (bpmn-layout-generators is the only repository that needs it). Done on 2022-01-06 around 11:12 UTC
• Update maven wrapper to use the Apache maven wrapper released on 18 Dec 2021: http://mail-archives.apache.org/mod_mbox/maven-announce/202112.mbox/%3c3167727.44csPzL39Z@giga%3e

Not managed here
Test the java build with java 17. This may require a maven update. See maven 3.8.3 release on 2021-10-04
--> compilation error due to lombok. A new lombok version exists but it has not been tested.

This is the way we are building the graph under the hood: The Manhattan distance function computes the distance that would be traveled to get from one data point to the other if a grid-like path is followed.

resources

• https://docs.lib.purdue.edu/cgi/viewcontent.cgi?article=1039&context=ecetr
• https://stackoverflow.com/questions/1285623/manhattan-layout-algorithm
• https://en.wikipedia.org/wiki/Organizational_chart
• http://www.improvedoutcomes.com/docs/WebSiteDocs/Clustering/Clustering_Parameters/Manhattan_Distance_Metric.htm

mxgraph java implementation

• https://github.com/jgraph/mxgraph/tree/v4.2.2/java/src/com/mxgraph/layout

used by bpmn.io for autolayout (there are refs to this algorithm described in diagram-js)

• https://github.com/bpmn-io/diagram-js/blob/develop/lib/layout/ManhattanLayout.js#L197
• https://github.com/bpmn-io/bpmn-auto-layout/blob/fa13b219c470a46ea0d1044d04d19d9183c7ac40/lib/DiUtil.js#L48

Activity and event label positions should be adjusted