Testing framework rewrite

ValentijnvdBeek commented 3 years ago

Rewrites the testing framework from shell into Python 3 using the Jinja templating engine.

Overview:

Test code is executed in a different function preventing abstractions leaks
Fixes an issue where memory issues were not detected by the testing framework
An error for the test now also causes the test to fail when memory tracing is enabled
Tests are now organized in test groups, or suites, based on the file location of the test
Tests are generated at compile time and stored in a different build directory
Library for printing ANSI escape codes
Entire test suite is now written in Python 3
Adds a bunch of extra assert macros which show what the expected and actual results are

TODOs:

Source of random information in the kernel, rather at compile time
Selection of a test suite using the debugging console
Implementing float, double, and long printing for kprintf

Known bugs:

Getting the testing object files to be linked correctly using Make has been a bit of a struggle, so I have decided to not to do that and instead work on rewriting the build system into a modern build system, scons, as discussed in one of the meetings

ValentijnvdBeek commented 3 years ago

I made a mistake and accidently pushed onto this branch instead. The last commit contains an initial version of using meson rather than Make, but the kernel doesn't run and I can't quite figure out why. @jonay2000 got any ideas or way to debug it?

jdonszelmann commented 3 years ago

I made a mistake and accidently pushed onto this branch instead. The last commit contains an initial version of using meson rather than Make, but the kernel doesn't run and I can't quite figure out why. @jonay2000 got any ideas or way to debug it?

Which part? The pushing on the wrong branch or using meson? I haven't ever used meson I must confess so I don't think I'll be much used but if you want to talk about it (call maybe) that's alright

ValentijnvdBeek commented 3 years ago

Which part? The pushing on the wrong branch or using meson?

Pushing the wrong branch

I haven't ever used meson I must confess so I don't think I'll be much used but if you want to talk about it (call maybe) that's alright

It is more that I knew that you used a different build system yourself so maybe you had a similar problem or know in which direction I should look at it. Basically, it compiles and seems to use the same arguments but it just doesn't start the kernel when running qemu

rellermeyer / course_os

Testing framework rewrite #83