Confirmed users
3,816
edits
Calendar:QA Home: Difference between revisions
→Finding a Regression Range on a Bug
| Line 42: | Line 42: | ||
An example using a [http://en.wikipedia.org/wiki/Binary_search_algorithm binary search]: | An example using a [http://en.wikipedia.org/wiki/Binary_search_algorithm binary search]: | ||
Let's say that test X worked in version 0.3.1 and did not work in the current nightlies. You find out that 0.3.1 was released on February | Let's say that test X worked in version 0.3.1 and did not work in the current nightlies. You find out that 0.3.1 was released on February 15, 2007. So, you start looking at builds to find out when test X stopped working. | ||
# Check 0.3.1 just to be sure that it did in fact work | # Check 0.3.1 just to be sure that it did in fact work | ||
# Check a nightly build created on a date halfway between now and February | # Check a nightly build created on a date halfway between now and February 15, 2007. | ||
* If test X works on this build, then check it with a nightly that is halfway between this date and the current date. | * If test X works on this build, then check it with a nightly that is halfway between this date and the current date. | ||
* If test X fails on this build, then check it with a nightly that is halfway between this date and the date of your last known working build (in this case, February | * If test X fails on this build, then check it with a nightly that is halfway between this date and the date of your last known working build (in this case, February 15, 2007) | ||
Keep halfing like this until you find the date where in the nightly before this date test X worked, and in the nightly built on this date, test X fails. | Keep halfing like this until you find the date where in the nightly before this date test X worked, and in the nightly built on this date, test X fails. | ||