With my university’s announcement of having two exams per subject, along with a final assessment that accounts for the entire coursework, most of my time has been spent trying to keep up with each subject’s unique demands. I read Tom’s post on the forums, and found time to tackle the two failures he more or less solved.
I had already been thinking about those failures beforehand: the serialized_contents failure was an obvious one, with the solution being the same as how we maintained
test_names in the initial GSoC post. It does look much cleaner than the test_names one because it’s a dictionary, and it reuses the same aliases we have defined in our settings, making it a bit elegant and concise.
The autoreload failure insight pointed out a very simple solution to a seemingly unknown problem to me.
There’s another failure I’m not sure how to tackle yet due to the nature of how it’s written: the check_framework failure. I initially thought about using a SharedMemory object from the multiprocessing library to attempt to synchronize the boolean check. I tried it out for a couple of hours, but the implementation was too complicated for a simple problem. Nonetheless, I’ll try to consider other ways of doing this. Of course, it might be something that has to be solved with said complicated process, but I want to exhaust other options before revisiting that one. There’s also an advantage that this solution would be applicable for other future syncing problems; it would become the standard way of solving it.
On the Jenkins build, I had two new failures on contenttypes_tests. I didn’t have them on my local branch so I pulled them to check them out. Currently, I don’t have an idea why they’re failing, but my ballpark estimate is that something changed about the management command itself or spawn is messing up the settings it uses. I’m going to check this out for next week.
Starting from next Monday, I’ll be able to start tackling Django full-time, and I expect my progress on Django to accelerate. I hope I can kick it into full gear starting from next week to tackle the remaining failures and to start testing on PostgreSQL and MySQL.