And in the “Oow, I’m glad I’m not in that department” department.
The latest Woody’s Office Watch reports that Excel 2007 can’t format 2^16 or 2^16-1 properly, instead deciding to show it as “100,000”
Heh. It’s not too bad. According to Microsoft, the bug only affects 12 numbers (or is it 2, I’m not sure, but the quote says 12) out of a virtually infinite number space. From Woody’s; “They (Microsoft)point out that Excel can use ” 9.214*10^18 different floating point numbers ” but the bug only affects 12 of that incredibly large number of possibilities”.
Oh man. Can I use that one with my next off-by-one mistake? (hey, I don’t make ’em much anymore but still…).
4 Comments
By this time next week new viruses will be written using these 12 magic numbers, rendering them INVISIBLE to Excel! Somebody alert Homeland Security…
I’m quite amazed this stuff doesn’t happen more often. Testing an app like Excel must be incredibly difficult.
I talked once with one of the guys on the VB6 testing team at a VBits.
Damn, I would not want that job.
But he was as gung ho as anyone I’ve ever met about what he did. Absolutely loved being hated by the development team.
It takes all kinds!
Testing is an art form. It’s easy to identify something as "broken". It’s hard to document how you did it in such a way that it’s reproduceable. The better testers can also suggest a possible fix. I’ve had relationships with good testers that bordered on dependency, and relationships with the bad ones that might have resulted in blood on the keyboard had we continued.
And that’s testing a "simple" application that collects data from the user and does something. I cannot conceive how testing a programming language would work.
It’s a blind spot I am quite willing to live with.
I tend to liken testers to book editors.
You certainly don’t want your authors to be editing their own work. And a good editor can make or break a book.
Tough to find and not respected nearly enough.