So after a bit of fiddling with Hugo, Netlify, and Jekyll, I’ve finally set up this blog. I’ll be dedicating this entire portion of the blog to documenting my Google Summer of Code progress and associated thoughts or plans.
Suffice to say, I’m absolutely thrilled to work with my two mentors, Adam Johnson and Tom Forbes, on bringing this project to fruition.
Working on Django for an entire summer as part of Google Summer of Code is a dream come true to say the least especially with Adam and Tom, but enough about how excited and ecstatic I am because I’d run out of room to write anything else if I didn’t stop.
One of my mentors, Adam Johnson, has pinged me on two pull requests so far; both of them being ones I wanted to submit a patch for but didn’t have the time to undertake.
One of them is about Django’s faulthandler
not working properly in parallel; this ticket had caught my eye from early on because I knew I’d have to implement it during the summer. The faulthandler
works by being enabled in the parent process and getting inherited through fork
. This breaks down if we’re using spawn
instead. The solution is simple: add a call to faulthandler.enable()
during worker initialization if faulthandler
was not disabled initially and our multiprocessing
method is spawn
. I imagine I’ll have to add this during my final clean up of the project since this feature will be added later in Django 3.2.
The second one concerned output buffering which was an interesting ticket to track out. It’s a patch that’ll simplify test output to make it less cluttered and more efficient by putting successful test output behind a buffer and only showing failures since they’re useful information.
I wanted to work immediately on setting up my blog and picking up where I left off a month and a half ago on my project, but my computer’s bootloader cracked down after I attempted fiddled with one of my operating systems. It cracked down because I was had multiple operating systems sharing the same physical hard drive. I managed to fix all of this four days ago and I’ve quickly set up this blog along with a main blog.
I’ll be making a forum post within the week to summarize my benchmarking of the initial proposed methods in my proposal. I first have to fix the SQLite errors I’m getting during the test run. I haven’t been able to figure out the cause behind them yet, but I’m fairly certain it’s related to constraint checking being enabled on cloned databases. I also want to finalize discussions of goals and milestones to ensure I’m on track with my mentors.
I submitted a pull request for an easy ticket, and added comments on other tickets mainly 7835 and 22079.
I’m looking forward to tackling 26822 in a way that’ll work for both spawn and fork during Google Summer of Code. There’s no official implementation yet but the workaround suggested by Simon works. It would be nicer if we had that implementation at the create_test_db level though as mentioned by him.