insert_overlap does not sometimes merge intervals

[lailoken]

So it seems that insert_overlap merges the inserted item to a interval in the tree if there is overlap, but it does not merge to all overlapped intervals.

Not sure if this is by design:

  insert_overlap [0-9]
  insert_overlap [20-29]
  -> [0-9] [20-29]

  insert_overlap [10-19]
  -> [0-19] [20-29] // should have been merged.

  Deoverlap
  -> [0-29] // this is what I expected above

Same if that last insert_overlap was clearly larger (obviously, but I tested anyway):

  insert_overlap [9-20]
  -> [0-20] [20-29] // should have been merged.

For now I always do a deoverlap() after insert_overlap (I only use insert_overlap in my app), which seems redundant.

[lailoken]

To be clear, I've worked around / fixed all bugs in my own code, so no rush.

[5cript]

I just simply missed thinking about this case. Its a good question whether that should be by design or not.
In any case, an option or another function should provide this functionality.

Thank you very much for reporting your findings and helping to improve its quality.

[5cript]

Implemented with #50.

[5cript]

Please rebase your changes onto the current master and try again with insert_overlap(interval, false, true).
(Last true = recursive).

Please close the issue if I solved this issue. (Two unit tests say yes)

[lailoken]

Tested it as well, I think the recursice parameter is the best option.

Without it, the only thing that bugs me is that I don't know with which range it will merge. I think I've seen it merge to the earliest range, but with a red/black tree and non-deoverlapped it could be nondeterministic.

(I would vote for making the default true)

Tested the parameter and it worked. Closing.