Isn’t the irony of this that LLMs will only have enough training data to generate meaningful code for popular frameworks?

Less popular platforms will get more isolated rather than less as the wave of generated code that “works out the box” only works with stuff like react…