Strange mergesort behavior

Hi all.

I have encountered a strange issue with mergesort objects.

In the attached model, there are 7 identical lines, where a source with an arrival time of uniform(0,2) feeds a queue, which in turn feeds a mergesort. The mergesort feeds 6 processors. The mergesorts have the default "pull anything" pull criteria.

If I run the model, some of the lines work, while others don't. On the lines that don't work, the queue has status:releasing, and the mergesort has status:empty, but no flow items are transferred from the queue to the mergesort.

Each time I reset and rerun the model, the pattern of working and non-working lines changes. A line that worked on a previous run may not work, and vice versa.

I can force the line to run by executing a openip(Mergesort,1) command, but not with an openinput(Mergesort).

Can anyone offer an explanation of why the lines do not run?

Best regards
Christian

This appears to be a bug with the MergeSort's wrap-around feature.

Because all of your exit points have a blocking value of 0, when flowitems reach the end of the conveyor, they start at the beginning of the conveyor again. The state of the conveyor when you stop the model is determining whether you see this issue or not. That's why this bug is inconsistent; it is based on the timing of when you stopped the previous run.

If you set the last exit point's blocking value to 1, then you can avoid this issue.

You can also put the following code into the reset trigger of each of your MergeSorts to avoid this issue. This code will reset the values in their input table that aren't getting reset properly:

treenode inputtable = getvarnode(current,"inputtable");
for (int r = 1; r <= gettablerows(inputtable); r++)
	settablenum(inputtable, r, 2, 0);

Thanks, Phil.

I will work around it, then. Do you expect this to be fixed in a future release?

I will fix this bug for the next release.