Error Handling by Microsoft

[czardas 1,126]

A bug in one of my scripts had somehow gone unchecked by me: simply because it was not causing any problems. I seem to remember having doubts about it at the time, and perhaps I had dismissed the possibilty of a divisor becoming zero during a particular operation which involved the Mod function. Amazingly, the script always seemed to work in any case. So apart from now making myself feel like a clown, I was surprised to receive a Microsoft bug report when I tried to run the script with the latest AutoIt interpreter. It didn't take me long to track the problem, and after running tests I discovered the following.

AutoIt Version: 3.3.0.0

Mod(1,0) => Returns -1.#IND

AutoIt Version: 3.3.6.1

Mod(1,0) => Send Error Report to Microsoft

I couldn't find any reference to changes in error handling. It could be relevant to the Devs. Both tests were run on XP pro service pack 3. This is the first time I have encountered a Microsoft Bug Report while using AutoIt, otherwise I would not have mentioned it.

Edited May 31, 2010 by czardas

[jchd 967]

It's either (or both, why not?) related to this ticket (where as posted as anonymous because I almost always forget to re-enter jchd in this little box) or that post.

I posted a workaround.

[czardas 1,126]

Thanks for the responce. I really have no idea if it's a bug, related to any other issues, or not. I am used to getting errors from the handler, but not from the OS. It's really a consequence of an error in my code. But it seemed a bit out of the ordinary to get that message. I don't know if it helps at all.

[Richard Robertson 186]

It is a bug that it would hard crash like that.

[jchd 967]

Granted.

[Mat 364]

It's fixed in the next version according to the ticket.

[czardas 1,126]

It's fixed in the next version according to the ticket.

Presuming it is the same issue.

[jvanegmond 297]

Presuming it is the same issue.

It's not. The ticket says: Fixed by revision [5432] in version: 3.3.1.7

Looks like that change has probably created this bug since it's not in 3.3.0.0 and in my 3.3.4.0 it does crash.

[czardas 1,126]

I decided I ought to create a new ticket.

[Valik 471]

You are correct, the code was fixed so that integers are treated as integers and not converted to floats. This introduces a divide-by-zero possibility.