C Document Error: Return type taking precedence over first parameter. about cldoc HOT 2 CLOSED

To give some more detail, the responsible code is here: https://github.com/jessevdk/cldoc/blob/master/cldoc/tree.py#L305

Normally what happens is that functions returning a particular struct type are moved under that struct's page. This is loosely inspired by godoc, but is obviously wrong for C functions. There are actually two bugs in that code. The first assumes that every typedef is a struct-like thing. The second is to give priority to the return type over the first argument type. Maybe a better heuristic for the behavior is that only functions Type* type_new(...) are moved under Type.

Having said all that, I'm not sure about using any kind of non-trivial heuristic, since there will always be cases for which the heuristic doesn't work. Open to ideas.

from cldoc.

Would it make sense to check for common alloc idioms (init, alloc, create, new) and give functions with those precedence for the return type, and associate other functions with the first parameter?

Type* type_new(Foo *param);

This would correctly associate the previous function with the right type, but prevent functions that return types but are part of another "type".

Of course this wouldn't hit everything, but might catch a good amount.

from cldoc.