Comment by raxxorraxor
Comment by raxxorraxor 8 days ago
In my opinion this isn't even too relevant. I am no python expert but I believe defining a logger at the top for the average one file python script is perfectly adequate or even very sensible in many scenarios. Depends on what you expect the code to do. Ok, the file is named utils.py...
Worse by far is still the ability of AI to really integrate different problems and combine them into a solution. And it also seems to depend on the language. In my opinion especially Python and JS results are often very mixxed while other languages with presumably a smaller training set might even fare better. JS seems often fine with asynchronous operation like that file check however.
Perhaps really vetting a training set would improve AIs, but it would be quite work intensive to build something like that. That would require a lot of senior devs, which is hard to come by. And then they need to agree on code quality, which might be impossible.
This is a logging setup being done top-level in an auxiliary module "utils": you might import it into one command and not another, and end up surprised why is one getting the logging setup and the other isn't. Or you might attempt to configure it and the import would override it.
As for getting a lot of code that was vetted by senior engineers, that's not so hard: you just have to pay for it. Basically, any company could — for a price — consider sharing their codebase for training.