QA Hates You

I was trying to use the online interface for one of my credit cards and wasn’t doing anything too naughty, I thought, when I got the Java Null Pointer Exception:

Click for full size

I only use this application once in a great while (this was my second try with it), and I only spent a couple minutes with it as a user, not as a tester. Yet, here it is, an asplosion.

Am I just talented, or does the world of applications just suck that badly?

Whenever I’m planning or estimating for a financial application of any sort, I always plan for weeks of testing. Then, I don’t get called back. No doubt development shops the world over can find lower bidders (or none at all!)

And here you go.

You know what happens when you try to delete that entry?  All hell breaks loose:

Click for full size

You know how you find that error?  You just test.  It stems from a basic oversight in hooking up the database to the interface, where the application lets you delete something you’re not supposed to delete.

You should never see this sort of stack trace.  It leaks all sorts of insight into your database structure.

You should never see this sort of stack trace.  But you probably will.

Reader Max G. sends in this stack trace, the wailing and lamenting of a Web server lacking a little size where it matters most:

Click for full size

You know how to check for that?  Load testing, perhaps, where you pile in a whole lot of records, I suppose, or overloading an uploading sort of application using your 100Gb test files.

But brace yourself: when you encounter this in a test environment, your developers will tell you it’s due to the limitations of test environment (and you can’t try it on the production server because it’s the production server, and your test might successfully break it on the Internet).  But you have to push back.  Otherwise somebody like Max might find it in production.

Okay, this one does not occur in nature, but man, if only:

Click for full size

Here’s a simple entry into the Gallery of Stack Traces: a simple timeout that occurs when your Web server taps its foot and looks at its watch while the database isn’t showing up.  Finally, the Web server (badly) leaves a note that says, impertinently, that it waited as long as it could and it had other things to do.

Click for full size

You should really handle that programmatically.  Especially if your entire site hides within a Content Management System, and this little timeout makes it look like your site is irrevocably broken.

Sometimes, your content management systems just cannot get what your database is telling them. The result: The Parser Error:

Click for full size

In this case, it looks like it’s supposed to dish up a banner ad that the application cannot handle, resulting in havoc.

How do you find these things? Well, you just have to try to load the page. If you’ve got some sort of banner ad rotation algorithm going on, you have to reload the page. A lot.

A very basic automated test could do this for you; simply have the script load the page and use some navigation on the page. If the automated test bombs out, it could be because the navigation wasn’t there. Heck, even Empirix can handle that.

Wait a minute, someone just told me that Gopher was the Pacific Princess’s purser, whatever the heck that means. Oh, well, too late now, the gag’s done.

Don’t you hate it when you’re doing timeout tests, and the application shows the login page, but when you log back in you get .BARF?

Click for full size

When the application says, “The return URL specified for request redirection is invalid.”, it really means “Dude, why did you put all those nasty, ill-encoded characters in the URL when you know I can’t handle them?”

To find this particular beauty, you need to watch for your application dumping a lot of data onto the querystring between transactions. Then, wait for it to time out in the middle of the operation. Your application can then recognize you’ve timed out and direct you to the login page with a return URL on the querystring (if your application does this, of course). When you try to log back in, the application chokes on its own URL.

Beautiful!

For more information, see this piece on ASP.NET migration and this bit entitled “Silent Breaking-Change to FormsAuthentication::RedirectFromLoginPage.”

Hey, what can happen if you click .Net 2.0 controls before the page finishes loading? Sometimes, this:

Click for full size

Build an application where you force the interface through the pipe along with the actual logic, and you’re asking for trouble. Have some.

Developers are from Venus, and QA isn’t from Mars; it’s from a higher plane of existence than mere code monkeys as demonstrated in this stack trace:

(more…)

What do you get when you try to put a string where the application wants a double? Well, some developers would call it a training issue, but we here at QA Hates You call it a beautiful, beautiful stack trace:

Click for full size

Following the heels of yesterday’s crash, today BugTracker.NET crapped out on me:

Click for full size

This stack trace occurs when the Jedi Master developers play mind tricks on the weak-minded code:
(more…)

Ever feel like you’re stuck in The Matrix?

Spoon boy: Do not try and bend the row at position 0. That’s impossible. Instead… only try to realize the truth.
Neo: What truth?
Spoon boy: There is no row at position 0.
Neo: There is no row at position 0?
Spoon boy: Then you’ll see, that it is not the row at position 0 that bends, it is only yourself.

(more…)

As you know, you should always test your edit boxes to make sure that it can handle things like HTML and XML tags within it, particularly the dreaded </html> and </xml> or </table> or <!DOCTYPE html PUBLIC "-//W3C//DTD XHTML 1.0 Transitional//EN" "http://www.w3.org/TR/xhtml1/DTD/xhtml1-transitional.dtd">. At best, the application will handle it through any sort of mechanisms, such as:

• Accepting the value, but using escaped characters so that the application, browser, and database know this isn’t supposed to render.
• Disallowing use of the < or > characters.
• Performing client- or server-side validation to tell user to try again without the markup.

What the application should not do is this:

(more…)

As you know, stack traces provide QA with the joie de vivre and esprit de corpse that get us through the day. As part of an ongoing series, I am going to present a series of some of my favorite stack traces that never fail to bring a smile to my face and a Dosso double-snap to my fingertips. From the Gallery today, we have two extra fun .NET stack traces that occur, frequently, when you do naughty things with dates and times.
(more…)