The new file sort_20220821.html.txt is out, with some interesting progress, such as a stack sharing the space of the main working area. More stress cases are available with the new buttons.

The algorithm so far is quite simple, built upon the previous iteration that detects increasing or decreasing sequences. While the input is monotonous, the values are stacked. If the value becomes increasing, the stack is flushed to the top of the working area and the next data are accumulated there until the end of the run.

Otherwise the data get pushed on the stack while they are monotonous or decreasing. When the run ends, the current run is popped in reverse order and moved to the working area, to make an increasing run.

After each run, the position and length of the run should be pushed on the stack but that will be for the next version... and once we have this list of elements on the stack we can merge-sort them as they should.