Generated stable keys with better-bibtex

edited 7 days ago
Hello to all

Background:

I currently use Zotero, better-bibtex and emacs. This question can be answered without knowing emacs.

To insert a citation, I just need to do the command "org-cite-insert", which allows me to insert something like this :

[cite:@a-key]

Here, the key obviously refers to the right reference in zotero.

Unfortunately, my key generation depends on the author, a date or whatever: it is modified if I modify the reference. In my documents, it will always be marked "[cite:@a-key]", and not "[cite:@a-new-key]" (because I change a field in the reference). And so, org-mode doesn't recognize my key anymore.



So my question is:

How to randomly and stably generate the keys, even if I change a field of my reference (name, date etc) ?

I didn't find a clear explanation about this, the only method I can use is to put a key generation that gives an empty result, so I would have keys with this : zotero-1, zotero-2 etc. I find the solution not very clean.

Does anyone have a better solution?

(maybe I missed something in the documentation, and I didn't pay attention)

PS : the keys can be anything, it could simply be the date with a precision to the second, or why not a uuid (a bit long maybe ?). It should not be a meaning, but unique and it should never change.

Thanks in advance for your answer.
  • In BetterBibTeX you can pin citation keys. It does not look like you can default this in the settings, but you can bulk edit.

    Select all the items you want to remain the same, then right click, go to Better BibTeX, and select Pin BibTeX Key.

    You can undo this in the future with Unpin, but if you changed the name or date while it was pinned, it will not automatically change it when you unpin it. For that, you will have to Refresh the key.
  • edited 5 days ago
    Thanks you for yours answers

    I saw the possibility of pin, the problem being that it is a manual step, I find it surprising that there is not an option(I read the second message, and there is a practical option). I also have a problem: in the options, it says: On conflict with a pinned key, not pinned keys will be "kept" or "postfixed". So, if I understand correctly, if a key is pinned, when adding a reference, there can be a conflict. One more manual step. Isn't there a way to generate these keys otherwise?

    Moreover, I have another question: why do you need to have control over the generation of your keys? They must be unique, so why not generate a simple uuid? Is it to know which reference they are linked to ? (I don't have this problem with my software, I click on it and it shows me all the fields of the reference)
  • edited 5 days ago

    I also have a problem: in the options, it says: On conflict with a pinned key, not pinned keys will be "kept" or "postfixed". So, if I understand correctly, if a key is pinned, when adding a reference, there can be a conflict. One more manual step. Isn't there a way to generate these keys otherwise?

    You can have BBT pin the existing, generated key, and there should be no conflicts, but you can pin the key to whatever you buy manually typing the key. If you enter a key that way, or if you import a bib file that has a key that you already have in your library, you have to tell BBt what to do with these existing, non-pinned keys. If BBT generated heyns2022, and you import or manually type a key heyns2022 for a different entry, you now have two items with the same key. BBT can leave it that way, or it can postfix the existing heyns2022 so you have unique keys.

    But whatever you pin stays pinnned. You could have every item in your library pinned to heyns2022, and this option would do nothing.

    Moreover, I have another question: why do you need to have control over the generation of your keys? They must be unique, so why not generate a simple uuid? Is it to know which reference they are linked to ?

    That just tells me you don't use LaTeX for document authoring. Keys are meaningfully part of the text to LaTeX users, and between

    As \citet{heyns2022purism} argues, philosophy is the purest of sciences, 

    and

    As \citet{b117f0ca-2801-4d2d-be53-f1476a2d3a0c} argues, philosophy is the purest of sciences, 

    the first is easy to type by hand (which is what many LaTeX users like me do, without opening the bib file), and allows me to meaningfully review my text, where the second is gibberish, and I'd have to look it up every time I wanted to add a reference.

    (I don't have this problem with my software, I click on it and it shows me all the fields of the reference)

    then you don't meaningfully use bibtex, and bib files are just a transport mechanism in your toolchain.

  • Okay, thanks for all this very useful information. I understand a lot of things better now. Thank you again!
Sign In or Register to comment.