There are several problems with todays PartnerLink visualization.
1) It will not scale well for PatnerLinks that contain ports with many operations.
2) The individual WS activity nodes in the PartnerLink are not selectable. This
hinders the ability to do DnD from PL to Process.
Recommendation - revisit this visualiztion.
Scaling issue may be partly addressed by making PL expand/collapse capable.
But this probably will not be enough. For intance, if there are 20 operations
but our process, it will still be uncomfortably large when expanded.
We may need to consider following:
Also o using smaller visualizations/icons for the operations in the PL.
Or we may need some "hide / show all operation" toggle. This would hide/show
operations that are not connected to our process. The connected operations would
always be visible (assuming PL is expanded).
We had a design comment "Pinch / close swimlane to help horizontal space mgmt."
Not sure if this is still a valid or worthwhile idea.
TAF1: Prototyping or storyboarding is probably the way to go here--it's too hard
to react to descriptions. In general, I like the idea of collapsable
swimlane(s), but I'd need to see it to understand it.
TAF2: BTW, one aspect of that proposal that I really like is having a zone on
the left margin of the diagram that denotes "partner area". The current lack of
this is confusing, especially with the current algorithm that puts partners
outside of the viewport. I'm left thinking, "why is there this big ugly gap here?"
TAF3: I also want to mention that the averaging algorithm that Alexey used to
place the partner between two related activities results in a lot of excessive
lines and general confusion. Objections: 1) As a user, I want to be able to sort
the list of partners according to my own ideas of emphasis or importance. 2)
There is never a case where there is a straight line from a partner to the
activity, which results in excessive bent lines, which are hard to follow. The
vertical placement of partners is (or should be) indicative of where they become
relevant to the process, so it's frustrating to have the algorithm work against
that. I would rather have the algorithm produce one, good, straight line to the
first relevant activity in the process and then screw up the rest of the lines
than have the algorithm mess up all the lines in an attempt to average things out.
I think most achievable enhancement for 5.5.1 is addition of Filter to "Show All
Operations" (the opposite effect would be to only show those operations which
are connected to process). This would allow user to reduce the size of the
partner links with many operations.
Since the only outstanding issue here is item 2 from the original comment, degrading to P3 and postponing.