Wrong interpretation of MAX_VISIBLE_LIBRARY_HIERARCHY

[sebastian-brunner]

For the statemachine: unit_test_state_machines/deep_libraries/library_middle2

It has 33 States (library state instances count as two states, one for the library wrapper and one for the state_copy).

If MAX_VISIBLE_LIBRARY_HIERARCHY is set to 3 than all 33 Models for the states are loaded, which is correct. 3 levels of hierarchies are shown if a show-flag is enabled, which is also correct.

If MAX_VISIBLE_LIBRARY_HIERARCHY is set to 2 also all 33 Models for the states are loaded as well, which is wrong! Only 2 levels of hierarchies are shown if a show-flag is enabled, which is actually correct. Thus, a lot of models are created, but not shown using the show flag.

For bigger state machines, this drastically increases the loading time by a factor of FOUR!
E.g. for the unit_test_state_machines/deep_libraries/sm with deepest libraries!

For our LRU state machines, this is even worse!

[sebastian-brunner]

We should also refactor the MAX_VISIBLE_LIBRARY_HIERARCHY variable: currently it is some mixture between hierarchy and library levels.

My desired behavior would be the following:

• If MAX_VISIBLE_LIBRARY_HIERARCHY is set to 0 and I click the show flag, nothing should be shown.
• If MAX_VISIBLE_LIBRARY_HIERARCHY is set to 1 and I click the show flag, I can see the content of this library but the content of its child libraries is hidden.
• If MAX_VISIBLE_LIBRARY_HIERARCHY is set to 2 and I click the show flag, I can see the content of this library, the child libraries, but NOT the grand-child libraries.
  etc.

In fact, setting the MAX_VISIBLE_LIBRARY_HIERARCHY to 3 (in my new proposed convention 2) is the main use case in my experience.

[sebastian-brunner]

I am currently working on the internal issue 558. See branch feat/decrease_loading_time_internal_558

As this issue is not directly related, it can be analyzed in parallel.