Message11253

@Jeff I think making a Java bug report for this is worthwhile, just to see what they think. Although it's not clear it's a Java bug. At the very least it should be in the doc for Collections.sort, that in the failure case it can result in the list elements being changed, as I don't think that's what users of the API would expect. It is also kind of a regression caused by TimSort, as that wouldn't happen before. I will clean up the Java example and try to get that done tonight.

If I get chance I will have a go at the idea you suggest, I think it can work. We will need to be careful though as I think other test rely on different exceptions during sorting, but hopefully the tests cover them all. Even this approach will cost performance though as we will need a try catch for every comparison.

Thanks for also looking at this.