The thought of having to make a “TimeSignature_wrap”, as opposed to a “Grob_wrap”, is antithetical to the LilyPond core principles. We insist that, as much as possible, internals documentation be created from executable code, rather than
comments about the code, because we don’t want the documentation to diverge from the actual code. Thus, the interfaces listed as affecting BarLine don’t come from somebody keeping a list of the interfaces related to BarLine, they come from the bar_line_engraver
declaring (in the C++ code) the interfaces actually used. And there is virtually no static mapping.
This is not what I meant. In my idea, you don't have to touch, replace or modify the "Grob_wrap" principle of LilyPond. You just have to *add* a static wrap, and not necessarily manually: for example you can use the tool that Jean just provided. I thought that the tool was meant for formatting the doc, because this thread belongs to the *user* lists, but once Jean clarified that "it is _not_ intended to become something outputting documentation for on-the-fly usage in editors" , then you can generate an additional *redundant* wrap with it, so to feed the autodoc tool that you want to use and that you (user) can keep aligned with the previous wrap whenever you (user) want (given that the objects don't change so often).