2012-01-06 08:12:41 |
Robert Collins |
description |
Manual QA is error-prone and time-consuming. We need a script that will set up and dispatch package, recipe and a translations builds in large numbers and monitor their results. Some builds should also be set up to fail. |
While there are unit tests in the tree (including ones that run up the slave software), this doesn't exercise things as closely to production as integration tests in the datacentre could do.
Some key things to test are:
* set up and dispatch package, recipe and a translations builds
* in large numbers and
* monitor their results.
* Run some expected failure cases too. |
|