Le lundi 19 juillet 2021 à 20:47 -0400, John Kitchin a écrit : [ Snip... ] > It is also possible to use  \nocite{*} as a cite, which includes all > references from a bibliography, and yet contains no key. Hence your reluctance to work from a set of larger databases, which would include irrelevant reference. AFAIUI, your workflow is based on the creation of a per-publication database. > Even funnier in a way is \nocite{key} which just adds entries to the > bibliography, but does not cite them in the body of a document. Do you mean that thes empty entries will be added to the original .bib database ??? Or simply that these entries will be added to the (unseen) list of \bibitem entries that will be typeset as the reference list ? > Footnotes are not always allowed in publications, and for various > reasons not worth defending, in proposals one might want to put this > in the references because of space limitations. FWIW, \citetext will not add anything to the reference list (see enclosed example...). BTW : what should it add ? > I count at least 10 examples of such personal communications in the > references in my library of ~1800 pdfs, so they aren't very common, > but certainly they exist in the wild. Whether people should do it or > not, they do. I certainly agree with that (I have used such unpublished information myself). My point is rather that, since there is nothing to refer to, they have no place in a reference list ; unless someone creates something to refer to (i. e. an entry in the bibliography database...). > > ISTR that at least CSL and BibLaTeX have types appropriates for a > > manuscript or a letter. You may also consider your own notes as > > documents and reference them (properly). > > > > > I don't see a way to get something like that in org-cite, since > > it > > > seems that a key is always required. > > > > Indeed : the key is, in relational algebra terms, the primary key > > of > > the bibliographic relation... > > > > > I think of it more like a lambda function, but for a cite reference, > where you > define what you want inline.  The fly in the ointment being that you have to "create on the fly" what will be in the text and what will be in the reference list... > It is pretty common in scientific papers > and proposals to see that.  > > It may not make sense to make an @misc bibtex entry for that purpose, > since it is a one time citation for that document, and is like a > lambda reference. The problem with that approach is that the reader has to somehow reverse your function (which is by no way guranteed to be bijective...). -- Emmanuel Charpentier