I got myself an issue assigned on Friday, but did not start working on it until today, but fret not, I've spent a lot of time with Kiwi's codebase today and I think I have a preliminary understanding of how to write most of the other testcases. I'll try to remember as much of what I did as I can.
The issue I took up is of writing automated testcases for one of kiwi's submodules (issue, for reference). The issue wants us to take the testexecutionstatussubmodule and write tests for it. Luckily for me the file only has a single function. So this is what I'm supposed to do:
- Create a test file for
testexecutionstatus.py. - Choose one funtion.
- Make a test class for it.
- Initialize all the pre-requiste objects.
- Check all the possible use cases of the function.
Which sounded pretty easy at a first glance, as it usually does, but ain't, especially if you have no idea what's happening in the code. I used the code written for a similar issue (different module) as reference, and there were plenty of other tests for different modules in the testssubdirectory.
While I had understood what to do, I didn't know how to do it. Digging around a little more, I started seeing a common pattern when it came to initializing all the pre-requisite objects. All the objects being initialized were using factories. I found the factories.pyfile, and sure enough this is where the objects being as prerequisites were being defined. I tried looking for a class for TestExecutionStatus, but there wasn't one. Hmm, couldn't figure out what to do. Then I remembered that the project maintainers had told me to look at the individual tables and also how all of them related to each other. I started looking for something along the lines of TestExecutionStatus and found it quickly as you can see below.
The image tells me that TestExecutionStatus is a foreign key in the TestExecution model, which there exists a factory for, Bingo!.
Now to writing the actual tests, which I'll write about in the next post.
Other than that I spent a lot of time yesterday trying to figure out how to run the test file faster. Whenever I want to run my test file I do
./manage.py test --pattern="test_testexecutionstatus.py"
This does three things:
- Create a new test database
- Run all the migrations on it
- Run the test
This works, but each test run takes about 62 seconds, which is a little too long. I run whatever I write pretty often and this impedes progress and concentration severely, so I was looking for ways make it faster. First I used the --keepdb flag, this reuses the test databaase, but no significant difference in the total time. Then I tried the -v 3flag, to have a verbose looking output and it was clear that there were too many migrations, so it was the second step that was causing the delays. If I could make the migrations on the test database presist for each test run, I could significantly reduce the test time. Sadly, I haven't been able to find a solution to it yet. I've tried switching databases from SQlite to Postgres, turning off the [Migrate](https://docs.djangoproject.com/en/3.1/ref/settings/#std:setting-TEST_MIGRATE) flag, but have either run into problems, or it hasn't helped, I'll update if and when I find a fix.
In dotfile news, I have started using ripgrep(using vim-ripgrep) to search for text in files. It has been very useful and I'd highly recommend it for people trying to parse and understand massive codebases, it's very quick and accurate. If you're interested in why ripgrep is so fast, you can read the blog post by the author himself (haven't read it myself :| it's just too big and complicated). I've also switched out Delimitmate in favour of Auto-Pairs after reading this comparision, and using vim-closetag for having auto html tag completion.
That about sums it up! Yeah, long day :)