OpenPlans Operations

Come next week — save President’s day — we’ll be testing a new release of NYCStreets. Although there will always be new bugs to find, it is sometimes useful to look backward toward the ones that have already been found.

Enter Testing Regression Days.

As you might expect, Testing Regression Days give us a chance at verification — making sure bug fixes actually fix bugs — but for us, they’ll be more than that. Testing Regression Days will give us a chance to make sure old fixes didn’t cause new errors.

We’ll be doing this using two methods: Bug regression and charter regression.

• Bug regression is where we look at old bugs, verify they’re actually fixed, and then test around the fix to see if more errors have been made. This is verification++.
  
• Charter regression is simply taking an old testing task — say, testing the joining process — and re-exploring that area. This lets the tester cover that area of code again to see what turns up.

We plan on doing this after every new release, the first day of each week. The dates are:

1. Tuesday, Feb. 19th.
2. Monday, Feb. 25th.
3. Monday, Mar. 3rd.
4. Monday, Mar. 10th.
   

Notice that the last Testing Regression Day falls on our big release, March 10th. That Testing Regression Day will count as a “post-deployment” testing day to make sure everything goes smoothly.

Tim