I used to rely on mendeley for working with BibTeX related things. After switching mostly to Emacs, I find mendeley a bit confining. This page documents my Emacs based setup as of now. I will be updating it with new things that I try.

There are a few pieces in the process of managing references.

  1. Searching/adding papers/documents
  2. Using the BibTeX entries
  3. Working with pdf annotations/notes

1 Searching

Mendeley and similar tools have decent support for step 1 and 2. The way I had been using mendeley was to use its capture feature and make it output a single ~/library.bib file which can then be read by Emacs.

The literature search feature from mendeley can be replicated if you get an interface to Google Scholar and few other aggregators. Turns out there is a package exactly for that, gscholar-bibtex, which lets you search documents and add BibTeX entries in a single file as you would have been doing with mendeley.

2 Using references

Not much to say here. org-ref is a very feature complete solution for working with references in Org files. I am managing the following three locations using it:

  • The main bibliography file, org-ref-default-bibliography. The helm functions from org-ref make it really easy to search and add citations.
  • Notes file, org-ref-bibliography-notes. I am starting to keep a notes file (lives here) which will help me trace/organize ideas and also make the citations more meaningful rather than just pushing a [1], [2], [3], [4] list.
  • PDFs in org-ref-pdf-directory. PDFs kept here are picked up by org-ref allowing you to open/send email etc. Since pdf download features requires more metadata (at least DOI, if I read it right), I am keeping it a bit manual as of now.

Cross linking items is easy since they all are keyed by their cite-keys. Plus there are functions to extract document specific bibliographies so that I can keep pages here self contained.

3 Annotations

Although Emacs has a really decent pdf reading facility in pdf-tools now, it still provides a brittle experience. Text selection and scrolling doesn't feel as good as your regular pdf reader.

I am now using okular for reading all non plain text documents. It comes with really good support for file extensions and has complete annotation tooling (plus I am a KDE refuge). For seamlessly working with Emacs, I found openwith really useful.

4 Missing pieces

Here are a few things that I would like to setup when I have some time. Most likely there already are tools for these which I am unaware of.

  1. Browser capture

    Capturing papers from browser one of the most useful features of mendeley. Shouldn't be that hard to cook considering the many utilities for working with different websites available in org-ref.

  2. PDF management

    I tried papis but it doesn't provide a lot if you already have a bib thing going. Since most of the pdfs will be (somewhat) curated, I think a few basic functions would do just fine.