Hi all, I was looking at the current tests implemented for Pyglet. You all might know they are quite cumbersome to run due to the manual work. There is some functionality to run as an automatic regression set, but there is no default reference (which is understandable given the challenges). The result is that it can be quite hard to apply a patch while being sure you are not breaking existing functionality.
I want to propose to make the following changes to how Pyglet is tested: - Split off the non-interactive tests and convert them to unit tests. Unit tests - Determine coverage of the unit tests - Write unit tests until full coverage is obtained, while keeping test time short Interactive tests - Create a verified set of reference images for each supported platform (for now not considering hardware differences) - Show a second window with the reference image, so you do not need to read that much, but you can directly compare - Change tests that require additional keypresses to have a default run sequence, with reference images for each step - Automatically compare screen results to reference image and only show comparison if they do not match - Try to generate input events (keyboard, mouse) for Linux platform (maybe some tools are also cross-platform) As there are more interactive tests, I think I will focus on that first. Functionality for doing regression checking is present, but needs some polishing. I might also borrow some inspiration from Cocos's auto test set. Rob -- You received this message because you are subscribed to the Google Groups "pyglet-users" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To post to this group, send email to [email protected]. Visit this group at http://groups.google.com/group/pyglet-users. For more options, visit https://groups.google.com/d/optout.
