Sam w/o Max

Well, since I’m not updating much, I thought I’d point out this great new blood in the game qa blogging department, Sam Kalman’s Climactic Ave.

So far there aren’t many posts, but things are looking good with his diatribe on Session Based Testing (SBT, I promise you won’t get a disease if you click the link)

It sounds like a good idea, but I’m not sure if the actual bug reporting from it will be all that helpful. Yes, at some point you need this kind of functionality testing, but you’ve poisoned the repro steps. Or at the very least for any issue, say the grunt falling inside the other grunt’s body, the repro steps will end up looking like:

  1. Played Halo for 45 minutes
  2. Reached Library
  3. Shoot the grunt
  4. Grunt dies, falls

Expected result: Grunt’s corpse collapses to the floor

Actual result: Grunt’s corpse fell to the floor and passed into another dead grunt’s corpse.

Well, as you can see there is that mystical #1, and before a developer ever hears about this you may need better repro steps.

This kind of testing still needs to be done, but there has to be some trust there between you and your coworkers for them to believe you’re not just screwing around with this game. The charter and all are nice, but when it comes time to report the bug you need to have as simple a test case as possible to reproduce the issue, which invalidates all bugs you see whilst playing: you’re going to have to go back and reproduce them more simply, attempting to expose the error on its own. Capture and playback software, or a vcr, are really useful for all of this, as usual.

I admit that I’ve never actually done this kind of charter and whatnot myself, however, I’d certainly like to give it a try just to find out if my issue with it is correct.

No food for test team unless you make with the clicky-clicky:
  • Digg
  • Technorati
  • del.icio.us
  • Facebook
  • Slashdot
  • Mixx
  • Google
  • E-mail this story to a friend!
  • Print this article!

One Response to “Sam w/o Max”

  1. Sam Kalman Says:

    Hey Zach, thanks for the links! I hear your reservations about SBT & repro steps. One of the tricky balances of SBT is deciding the amount of time you spend testing versus the amount of time you spend reproducing bugs. The amount of time in each category is tracked as a key metric, so if you do end up spending a lot of time trying to get two grunts fall into each other, your lead/dev/management will know that.

    However, I should point out that SBT in no way promotes bad repro steps. Regardless of the time you spend in-session before finding a bug, the write-up should always be isolated. Especially for a bug like this one, which is really independent of location. Identifying and supplying good repro steps is really a separate skill from running Session-Based Testing.

    Plus, I love your post title L.O.L.

Leave a Reply