Kyle Meyer writes: > All right, so we can't get by without using eldoc--make-callback here? > Relying on a symbol marked with "--" makes me uneasy, and I'd like to > avoid it if possible. Yeah, good point. I was doing that so I could honour the setting of eldoc-documentation-strategy. I don't know if that's going to be a concern in practice though. > Does your cider test case above break if we use > eldoc-print-current-symbol-info without relaying the callback? That is, > this squashed into your patch: My concern with using the eldoc-print-current-symbol-info is that it's now somewhat subverting the actual eldoc documentation function -- i.e. the invocation of org-eldoc-documentation-function now "fails" and instead it prints out the actual documentation as a side-effect. Indeed, applying that patch makes the eldoc for python code blocks not work correctly. The below patch which essentially just inlines the definition of eldoc-documentation-default, so it's not messing around with any private variables in eldoc, although it now won't honour the documentation strategy. It remains to be seen if that will be an issue in practice, but if necessary we could just check the value of eldoc-documentation-strategy and behave appropriately.