James,

If we want to understand how it may happen, and we have only one program that may help to reproduce this, then it is the only choice to use the program to get the additional information.

You think that "the program which encountered it is not what caused it to happen", but I cannot see any evidence of this.

In any case I see that we have a little bit different approach and logic in investigating this particular problem; and because of this (probably) we cannot work together on this particular issue.

Let us do it separately and exchange opinions when and if we have new facts.

>Yuri --
>
>I was trying to understand how it may happen as well. However, the program which encountered it is not what caused it to happen, so understanding that program is fruitless.
>
>\
>>James,
>>
>>You have already answered your own question. Yes, it is possible because you met the condition where you know exactly that columnorder is zero.
>>
>>Is it expected behavior? No, as far as I know.
>>Anybody met this? No, as far as I know.
>>Is it documented anywhere? No, as far as I know.
>>
>>Probably I go too far and out of scope of your post when trying to understand how it may happen.
>>Sorry.
>>