H. Verbeet wrote:
On 26/09/06, Ivan Gyurdiev ivg231@gmail.com wrote:
change I'm interested in. However, I'm not sure why that would be better
- it would accomplish the exact same thing in twice the amount of time.
Well, there are two things I can think of right away:
- It makes thing easier to review
- It helps with tracking down regressions
I will make the required changes, as discussed with julliard on IRC.
...but this is really a theoretical argument. This is a framework mostly written by me to begin with, so the chances of breaking it are essentially zero. If it does get broken, the damage factor will be very low - it won't test as well as it used to.
If you want to help prevent regressions, please help write more state tests using this framework, rather than focusing on unimportant details. We need more test coverage! In fact, part of the motivation behind the changes being made is to make the framework a bit easier to understand and more flexible, so that others can contribute. If you don't like this framework, feel free to redesign it - but we have a number of obvious bugs in state management that need fixing, and test verification is the first step to doing that.
Ivan