On Wed, Sep 28, 2011 at 04:38:30PM -0400, Hans-Christoph Steiner wrote:
I think that audio tests would be the best way to start. Basically generate some audio to an table/array, then compare it to the expected result. There are a number of objects that do type-punning to speed things up. Modern compilers with their vectorization and other optimziations don't like type punning, so its a common source of errors.
Interesting. Before I entered the field of software development, I spent some time working as an audio mastering engineer, and we used to do "null tests" to validate the integrity of digital audio copies which involved flipping polarity on the copy and confirming that adding that to the original resulted in a flatline. Writing these tests will be nostalgic. :)
After some research today, I think it's going to take me a little while to work out the testing framework, though. Perhaps I'll stop by the IRC channel for some brainstorming.
Marvin Humphrey