Mapping lanes and pools as organisation hierarchy

[Ainsley Haslett]

Hello,

If a process model has been defined with many standard lanes and pools, it is required to document/produce a diagram which displays the relationships between those pools and lanes. This would appear as an organisation hierarchy.

e.g. the hierarchy of pools and lanes might be:
1.0 organisation pool
-1.1road office lane
--1.1.1 CTO actor lane
--1.1.2 Road office worker lane
2.0 Customer Pool
3.0 Supplier pool

Is there a way that the lanes and pools can be dragged onto a diagram type that would allow the lanes and pools to be linked like an ERD or class diagram format? Or could an ERD be made and the lanes and pools linked into the ERD elements in the same way that requirements can be linked into activies in a process diagram?

The purpose would be to show the context in the organisation that each lane and pool fits in to.

Alternatively is there an option to do the same thing using the actor element? This could also work as a solution although not all the lanes and pools are yet defined as actors so it wouldn't immediately be as easy to produce.

Any help would be appreciated, cheers!  

[bioform]

Anytime I create an actor in my model, I go ahead and create a partition "owned" by the actor (located directly below the actor in the project explorer) that is named the same as the actor.

My BR is that every "partion" must be owned by an actor, and a actor will ONLY have 1 partition.

I then drag the actor's partition onto the diagram that uses it...

Hope that helps,

David

[KP]

create a partition "owned" by the actor (located directly below the actor in the project explorer) that is named the same as the actor.

Yes, this is good. An alternative approach would be to create the partition and set its classifier to be the actor (Ctrl+L).

[Ainsley Haslett]

Thanks for the suggestions!

I have been using the BPMN tools for my activity/process diagrams so am interesting in linking the lanes/pools into the org hierarchy.

The model I have is very big, has multiple instances of the same lane or pool so as to allow for the elements to be ordered correctly in the project browser and hence come out in the right order in the reports.

So I think the best thing to do is to follow the classifier suggestion and firstly create a diagram with actors linked to represent the org hierarchy. I can then link each lane or pool so that the classifier is the relevant actor.

[bioform]

Thanks for the suggestions!
... The model I have is very big, has multiple instances of the same lane or pool so as to allow for the elements to be ordered correctly in the project browser and hence come out in the right order in the reports...

You really want to avoid having multiple occurrences of the same conceptual thing... If you’re planning on linking requirements/supporting traceability, then ALL requirements should be linked to the one unique object.

REUSE... REUSE... REUSE...

IMO - This is one of the issues that separates a UML model from just a collection of diagrams... one is a living/engineered construct used to support analysis and the project lifecycle, the other may be a semi-static model that will not support accurate and repeatable analysis...

[Ainsley Haslett]

I hear what you are saying, but...

The RTF orders elements according to the order in the project browser right. If you have a pool with activites within the activities sit in under the pool in the tree, and then in the rtf report all come out in order correctly.

When you add a second process diagram in a different area in the tree, if you reuse the pool then it starts to all go wrong. As you add activities into the second diagram the activities are added to the tree but you can't see the pool in the tree under the second diagram. The pool will not be included in the rtf report for the second diagram as it is not in that are of the tree.

With the main goal being able to generate document of the process for communication and signoff, would duplication then be required to make sure the documents are correct?

What are your thoughts?

[Ainsley Haslett]

How would the SQL be for a custom search that returns the element name and notes as well as the classfier name?

I am stuck on trying to join the t_object table to the t_attribute table to return the classifier name...

Cheers

create a partition "owned" by the actor (located directly below the actor in the project explorer) that is named the same as the actor.

Yes, this is good. An alternative approach would be to create the partition and set its classifier to be the actor (Ctrl+L).

[KP]

How would the SQL be for a custom search that returns the element name and notes as well as the classfier name?

You need to open t_object twice and join on Classifier id:

Code: [Select]

SELECT t1.Name, t2.Name FROM t_object AS t1, t_object AS t2
WHERE t1.Classifier = t2.Object_ID

[Ainsley Haslett]

KP you are a legend, thank you!  

Here is my final SQL script just in case anyone else might find it helpful:

SELECT
t1.ea_guid AS CLASSGUID,
t1.Object_Type AS CLASSTYPE,
t1.Name as Object,
t2.Name as Classifier,
t1.object_type as Type,
t1.stereotype as StereoType,
t1.scope as Scope,
t1.status as Status,
t1.version as Version,
t1.createddate as Created,
t1.modifieddate as Modified,
t1.author as Author
FROM
t_object t1 left outer join t_object t2 on
t1.Classifier = t2.Object_ID
where t1.stereotype in ("Pool","Lane")

[bioform]

... When you add a second process diagram in a different area in the tree, if you reuse the pool then it starts to all go wrong...What are your thoughts?

The method that I use assumes there is one primary activity diagram that shows all of the possible flows. When the flows are too complex to contain in one diagram, I re-factor them into multiple child diagrams using Composite activities.

Doing it that way I believe would prevent the problem you are encountering, as the child diagram contains the child elements.

Let me know if that seems to work for you...
David

Side Note:
I use the UC's scenarios to describe each of the different flows through the activity diagram. Not all, but the ones that are sponge.. oops, I mean analysis worthy

I run code that extracts pathways, and creates a "scenario script" using the partition's name (e.g., User and System) and the text from the "notes" section of each activity or action encountered on that thread...

Example:
USER selects a report to be run.
SYSTEM prompts user to supply report parameters.
USER specifies and submits the report's parameters
SYSTEM validates the report parameters
SYSTEM etc...

One thing to be aware of... an action or activity is NOT associated with a partition in the database, but the upper right and lower left X/Y coordinates are specified for all elements that are "on" the diagram (actions/activities, partitions, etc.) so I run a quick function in the code to determine WHAT partition the element "belongs" to (accounting for elements that may touch more than one... ID the primary as the one that contains the "most.")

[Ainsley Haslett]

Thanks for your note, interesting stuff. I see your approach and it sounds robust. The model we are working with has about 12 'mega-processes' which each need their own set of lanes and pools. For a smaller model I will take your approach, and I def take your advice on board, but what I think we will do is:

Set up an actor for each in a distinct list of lanes and pools. Link each lane and pool into a stereotype of the corresponding actor name.
Then map the actors on a diagram to represent the organisation hierarchy.
We will be able to monitor ongoing to make sure that each lane/pool has a relevant stereotype linked and that the stereotype, actor and lane/pool all have the same name.

That script sounds useful! Is it SQL or some sort of application code?

Cheers, Ainsley  

[bioform]

It sounds like you have a good approach mapped out.

The code I referred to was part of a Proof-Of-Concept that I previously developed and tried out on several projects.

I am currently trying to get up to speed in C# (I wrote everything in VBA using MS Access and Word) and create a simple add-in that could be used and explored. I have a private wiki that I am trying to work on that is specific to requirements engineering and implementing approaches (UP/UML) using EA. IF your or anyone else are interested send me a post. I intend to open the wiki eventually and to create an open source project from the effort. (feralrequirements at - some gibberish - gmail dot and with a COM. Trying not to get the address scraped...

The EA community really needs to start sharing code and solutions IMO and I am going to try and do just that..

David