This is the blog of John Dulaney, a hacker of Fedora, SCAdian, player of Music, blacksmith, sailor, and consumer of Bacon.
So, I have not been sleeping well for the past couple of months. In fact, two to three hours of sleep is the norm, and only one hour is not uncommon. This is not a new phenomenon; I go through basically the same thing every summer. As it happens, I am sitting here at 2 AM, writing this and listening to Neil Young.
So, I wind up being rather productive.
There is something substantially different about this year, though. I am no longer living where I grew up, but rather in what was my Grandfather’s house prior to his passing. So, this means no more sitting on a lifeguard stand at the beach and pondering Life, the Universe, and Everything.
Anyway, my big project of late is testing AutoQA.
What is AutoQA, you ask? It is the software suite in development by Fedora’s Quality Assurance team that is designed to test Fedora. You can read about it here: https://fedoraproject.org/wiki/AutoQA
Any way, in order to test the test software, several things are needed. Firstly is a mock-up of the Fedora build system (Koji and Bodhi), although the entire thing is not needed; just the back-ends that actually do the work. I am using the actual back-ends and interfacing with them as necessary. Secondly, one needs an instance of AutoQA running to be tested. This is pointed to our mock-up of the build system as if it was the build system. No new code needs to be integrated into AutoQA itself to handle this. And third, test cases with predictable output are needed.
So, how does this mock-up work precisely?
1. Read test case from file (plain text) that describes location of packages to be run through the AutoQA process and the predicted output of AutoQA if those packages were put through the actual build system. It should be noted that for some tests, almost the entirety of (default) Fedora may be required. From here, it is not difficult to simulate the hooks that AutoQA uses, although different ones are required for different tests. Some of them need to be updated after AutoQA is running, which triggers certain tests.
2. Start up an AutoQA instance and pass as options the test we want to test and point it to our packages in our mock build system.
3. Wait for output to be generated and then parse it. Compare it to expected results.
So, none of this is really difficult. The hard part is in creating test cases that thoroughly check AutoQA. We need tests that will pass AutoQA tests, tests that fail them, tests that fail because of one dependency error, tests that test to see what happens if there is a failure in the build system.
Want to help? We could use it with test creation, and Quality Assurance in general. It’s a fairly easy way to contribute to a Good Thing (Okay, maybe not so easy with test creation, but testing software in general is not that difficult). Go to https://fedoraproject.org/wiki/QA to get started.
P.S. Since I wrote this this morning, I’ve checked my email. Tim Flink has also been working on this project and has got some of the back end stuff I hadn’t done yet up on Git Hub: