Hello,
I’d like to keep my library of scientific articles in orgmode, along
with notes, links to external files (mostly PDF), etc. This has been
discussed repeatedly on this list, for example in the recent thread
http://thread.gmane.org/gmane.emacs.orgmode/78983.
Most solutions seem to be based around a central BibTeX file and take
advantage of RefTeX to navigate between citations to articles (in LaTeX
or org files), the BibTeX file, related entries in an org-file, and
linked external files. Often the key that connects the various items is
a unique label (in LastnameYear format, for example). This key is used
as label when citing and in BibTeX, as orgmode CUSTOM_ID, and as the
filename of an associated external file.
This seems to work well for people who have complete control over the
articles they write. But what about articles with co-authors? These
must be self-contained, so one needs a separate BibTeX file for each
article project. Let’s say that a co-author adds a new reference to a
common project, but the cited paper is already in my database under a
different label. Maybe that very paper is already cited in an older
article with different co-authors using a different \cite label?
Does anyone have a solution that handles such cases nicely? Perhaps
something along these lines: My (own, i.e. controlled by myself) library
of articles lives in one/several org files. Each entry there contains
links to external files, URLs, DOIs, etc. and enough information to
create a BibTeX record. (Such entries could be generated almost
automatically from .bib files found on the internet.) A custom Emacs
function allows to create BibTeX records for use in project-specific
.bib-files from the orgfile-entries. Now the last missing piece is to
connect things also in the other direction: allow to jump to the related
org-entry for a paper from a \cite{} in a paper, even if the reference
has been added to the .bib file by a colleague.
This last piece is the most difficult to realize. It would require
either a rather smart search, or a robust CUSTOM_ID (Perhaps
LastnameYearPagenr would be unique and robust enough?).
I’m interested in discussing these issues.
