[canselza] Reviewing exercise submission #25
Description
Link to Repository
Link to PR
What it did nicely:
- RAI nicely split its first response into multiple codeblocks where i could easily copy the scripts out of
- It created multiple functions that each had a single purpose and used them in the script instead of doing everything in one place
- I really liked the comprehensive README it created with explanations and descriptions of values
- The initial check for correct values and usage of default values when reading the config
Where it failed
- In the README it talked about a LICENSE or requirements.txt even tho they did not exist
- The test it wrote failed out of the gate
- It used unicode as variable names to have a better look instead of thinking about readable and usable code
AI Review vs Human Review
It seems like I mainly focused on improvements to the code for using the script itself.
I also focused on usability for developers. I also looked through the README and talked about things that seemed off for me as a developer which i would like better if i need to use that code.
Copilot focused on the correct "message" the content of a sentences or other texts give, e.g. the name of th etest function or some statements in the README that it did not deem correct. It also found many typos (mismatch between britisch and american english) and ut showed inconsistent usage of those. It also mentioned the wriong usage of en-dashes instead of hyphens.
Still it also found the missing LICENSE and requirements.txt files and the additional check that could be done in the reading of the config.
It hardly focused on improvements on the script itself and did not talk about the unicode variable name.