I think CMOS is pretty clear about your question: they recommend no formatting outside of the citation text that comes from the bibliography file in pandoc-citeproc, but it's flexible and up to the author/publication: "A footnote or an endnote generally lists the author, title, and facts of publication, in that order. . . . The notes allow space for unusual types of sources as well as for commentary on the sources cited, making this system extremely flexible." See http://www.chicagomanualofstyle.org/16/ch14/ch14_sec002.html, http://www.chicagomanualofstyle.org/16/ch14/ch14_sec014.html, http://www.chicagomanualofstyle.org/16/ch14/ch14_sec015.html, and http://www.chicagomanualofstyle.org/16/ch14/ch14_sec018.html.

Citation texts can come in three forms: full form (first citation), short form (last name, short title only), and "ibid." What follows the citation text can be punctuation (most often . , ; : but possibly also ? ! ] or otherwise—but this is left up to the author) or a space. As you know, in the short form of citation text, when the title is styled in quotation marks (e.g., journal article), a following comma needs to change places with the end quotation mark. All other following punctuation and spaces just remain as is (except with ibid. and a period; see next sentence). As ibid. comes with its own period, an following period needs to be suppressed (all other following punctuation remains; see http://www.chicagomanualofstyle.org/16/ch06/ch06_sec117.html). The only variable on the front end is with "ibid." and whether to cap or not cap the letter i—this is determined by whether the citation is starting a new sentence. I believe I saw elsewhere John writing that determining a sentence ending period and therefore a sentence beginning citation is difficult, so this would be a sticking point. 

So, as far as I can tell, if pandoc-citeproc expands a pandoc citation in a md note to the appropriate citation text (full, short, or ibid), with the above few conditions (and of course I may have missed some—but again, it's supposed to be flexible for the author/publication, if they want pp. or not, for instance), that would work. Maybe this is so ignorant of the challenges as to be offensive—but it seems like pandoc-citeproc is most of the way there. From my end, it seems like perhaps there would still be a use for the bracket elements of pandoc citations in order to indicate, for instance, where citations are within notes or to help facilitate making the above conditional changes based on non-citation sentence elements.

Another reason I would advocate for splitting the functions of pandoc-citeproc is the nice ability of md to stand alone as easily readable plain text. It would be great for footnotes to actually be md notes (even with pandoc-style citations). Perhaps this isn't so important for others.

And thanks, nickb, the line break works in my limited testing!

On Saturday, September 27, 2014 12:20:07 AM UTC-4, John MacFarlane wrote:

+++ hgv [Sep 26 14 14:43 ]:
>A larger solution might be in giving up the ability of pandoc-citeproc to
>produce either inline citations or note citations. I understand how useful
>this is, but if it doesn't actually work for one side of it (notes), I
>don't see the value. Of course, I only work with Chicago note-bibliography,
>which is where my bias comes from. But it seems to be designed primarily
>for those who work with Chicago author-date (and other inline styles), not
>really those who use both extensively. But as I'm sure this would entail a
>fair amount of work to just get back to where it is now, I understand the
>downsides.

The question is this:  in footnote styles, how SHOULD a citation that
appears inside a note be formatted?  Clearly not as a footnote, but
unfortunately beyond that the style won't give us guidance.  Should
it be a separate sentence?  In parentheses?  In brackets?  These are
all stylistic variations, but the style can't help us here because
it's a note style.

If this question could be answered, perhaps progress could be made.