No, I mean unfortunately for the persons who find themselves in the situation of making an argument before a court that (1) runs contrary to dozens of precedents, (2) against the whole tenor of copyright law for at least a century, and (3) without at least a theory of why the law requires such a conclusion.
The GPL is just fine. The only issue is about where and when the difference between the GPL and the LGPL can actually be enforced. If there is joint distribution of both the GPL component and the proprietary component, the developers of the former have significant leverage that they do not have if there is not, because the distributors must comply with the license to have the legal right to distribute the GPL component. No specious theories about derivative works required.
"Do you have examples of the former case?"
No. But as far as I know there is no legal dispute about the proposition that including substantive creative material from another source into a resulting product, whether literally or by translation, and without technical necessity creates a legally restricted derivative work. C++ templates and inline functions are usually raised as an example, although to some degree the use of both might be allowed on the grounds of technical necessity, i.e. to the degree necessary to make a compatible interface. cf. Baystate v. Bentley Systems (1996).
"And whether 110-years old decisions all make sense for software is an interesting question."
That is an issue Congress would have to address. The situation with software patents is similar. The latter are probably a counterproductive drag on the economy and the progress of science and the useful arts, but courts aren't generally in the business of making such determinations. That is what legislatures are for.