df.itemimprovements.xml:written_content.ref_aux comment may be wrong

[PatrikLundell]

I'm looking at a codex whose contents that has 4 references. The first one of those has a ref_aux value of 0, while the other ones have a value of 5.
Comparing the references with the DF display of the codex contents indicates that the 3 non zero ref_aux entries correspond to the codex' 3 chapters.

I've only seen 5 as a non zero ref_aux value (chapter?), but haven't looked at it in any kind of depth. My guess is that the zero reference is the work overall, while non zero ones are parts of the work.
My codex thus holds info about a goblin civ on the top level, with two chapters on the acceptance of peace treaties and one of the appointment of a general.

[PatrikLundell]

I've found the following ref_aux values in my world (without any idea what they mean, apart from possibly 0 and 5 as per the above):
0, 5, 6, 7, 8, 9

using the following script:

function w ()
  local aux_values = {}
  
  for i, work in ipairs (df.global.world.written_contents.all) do
    for k, aux in ipairs (work.ref_aux) do
      aux_values [aux] = true    
    end
  end
  
  for i, aux in pairs (aux_values) do
    dfhack.println (i)
  end
end

w ()```