Last week, Dan Blanchard, the maintainer of chardet—a Python library for detecting text encodings used by roughly 130 million projects a month— released a new…
I find “from scratch” a questionable claim as well. The model was probably trained on the code it is supposed to replace. Interfacing with the concrete tests and interfaces makes the result even more likely to be and puts it conceptually closer to the original code fed as training data.
What was the test suite licenced under? If it was in the same repo, then it was probably LGPL code as well.
If the MIT rewrite uses the LGPL licensed test cases, including them in the repo, then it probably must be LGPL as well.
I find “from scratch” a questionable claim as well. The model was probably trained on the code it is supposed to replace. Interfacing with the concrete tests and interfaces makes the result even more likely to be and puts it conceptually closer to the original code fed as training data.