List all of them in the comment.

2) Use the recently stated relative error of _mpd_qln() to generate the
   interval for the exact value of ln(x). See also the comment in mpd_qexp().