PM73657: Auto-Arrange Left to Right not working as expected

APAR status

• Closed as Permanent restriction.

Error description

• Auto-Arrange Left to Right not working as expected
  

Local fix

Problem summary

• ****************************************************************
  * USERS AFFECTED:                                              *
  ****************************************************************
  * PROBLEM DESCRIPTION:                                         *
  ****************************************************************
  * RECOMMENDATION:                                              *
  ****************************************************************
  The Auto Arrange algorithm places two Process symbols too
  far to the left.
  

Problem conclusion

• Working as designed. One of the Process symbols as 5
  incoming flow lines. The client would like one of the flows
  to dictate where the symbol will go, but in fact the other 4
  dictate.
  
  About the design:
  1. It's left-to-right, that is it starts from far left Event
  or Process symbols and follows flow lines rightward.
  2. There is no precedence; that is, all flow lines get equal
  weight.
  3. The start-off layout is not taken into account. This
  means that the user may have laid out the above cluster, but
  there is no attempt to remember it and try to preserve it.
  4. The design: Think of a spreadsheet with rows and columns,
  and think of the algorithm as placing each symbol into a
  cell within the spreadsheet.
  5. Somehow the "problem" Process symbols wound up 5 "cells"
  to the left of where the client wants them to be.
  
  Some history: Auto Arrange was created as part of a 3-part
  solution for users of IBM's WebSphere Business Modeling
  tool.
  
  Part-1 was a macro to import a Business Process diagram into
  SA.
  ? Problem: WBM provided NO location information. Thus all
  symbols came in as a hopeless jumble.
  
  Part-2 was Auto Arrange. Its purpose was to make sense of
  the WBM import as a first approximation.
  ? Comment: Auto Arrange was meant to be a brute-force
  one-time operation with no particular sophistication.
  
  Part-3 was the user making final adjustments to his
  satisfaction.
  ? Comment: We didn't expect a user to return to Auto Arrange
  after his manual adjustments. Hence no "memory" of logical
  clusters was built in.
  
  
  Auto Arrange has a random-walk generator built in, which is
  why different tries might get different results.
  
  The first step in the algorithm is to do a few hundred
  random walks that follow different paths through the flow
  lines from start to finish. It tries to find one start-off
  path that shows the most promise for resulting in a coherent
  diagram.
  
  My guess is that the algorithm came up with two or three
  different start-off paths that all seemed about equal, and
  chose one of them at random during one run and a different
  one during a different run.
  
  The original complaint was about a cluster of 4 Process
  symbols that the client wanted to keep clustered. Did he
  consider putting them inside a Group Box to preserve them as
  a cluster?
  
  
  All that said, bottom line is this is working as designed,
  and if the user is not happy, a Request for Enhancement
  needs to be submitted.
  

Temporary fix

Comments

APAR Information

• APAR is sysrouted FROM one or more of the following:

• APAR is sysrouted TO one or more of the following:

Fix information

Applicable component levels