Still nothing better than BibTeX format? [LWN.net]

Still nothing better than BibTeX format?

Posted Apr 5, 2026 16:18 UTC (Sun) by ceplm (subscriber, #41334)
Parent article: Pandoc: a workhorse for document conversion

> @article{fsfs,
> author = {Lee Phillips},
> journal = {LWN},
> month = {June},
> title = {{T}he {I}mportance of {F}ree {S}oftware to {S}cience},
> url = {https://lwn.net/Articles/1023299/},
> year = {2025}
> }

Can we, please, all agree that this is completely ridiculous to use in this year of Our Lord 2026? Using TeX groups for preserving capitalisation was a bad hack in 1985 and it is not any better now. When will Pandoc be finally able to read directly from Zotero?

to post comments

Still nothing better than BibTeX format?

Posted Apr 7, 2026 13:38 UTC (Tue) by leephillips (subscriber, #100450) [Link] (1 responses)

It may be ugly, but this is the format that can be reliably used by the most tools, from LaTeX to Typst. And, unless I have to make a manual adjustment (or write an article) I never need to look at it with my bare eyes.

Still nothing better than BibTeX format?

Posted Apr 7, 2026 14:07 UTC (Tue) by daroc (editor, #160859) [Link]

I believe Zotero can export BibTeX information as well. So adding a direct connection is "just" a matter of smoothing out the workflow, and not of enabling some new possibility, which might make it a less urgent prospect for developers contributing to Pandoc.

Still nothing better than BibTeX format?

Posted Apr 7, 2026 22:42 UTC (Tue) by jschrod (subscriber, #1646) [Link] (4 responses)

This usage of braces is not a property of the BibTeX format.

It's a habit caused by a relict, the behavior of old default BibTeX formatting styles. Other styles exist that implement capitalization different.

Please note that in the TeX world, almost nobody uses BibTeX, the program, any more. In BibLaTeX, one realizes capitalization with LaTeX macros, as one wants it.

Still nothing better than BibTeX format?

Posted Apr 9, 2026 17:52 UTC (Thu) by ejr (subscriber, #51652) [Link] (3 responses)

Actually... The rules for title capitalization depend on publication venue, locale, and other things. It's completely nuts and not at all designed for an acronym-heavy field. Or when peoples' names may coincide with articles that won't be capitalized (e.g. The). Everything that "must" preserve the acronyms needs some form of the brace-quoting. CSL JSON uses brace quotes as well.

Still nothing better than BibTeX format?

Posted Apr 9, 2026 18:25 UTC (Thu) by jschrod (subscriber, #1646) [Link] (2 responses)

I don't dispute the need for acronym markup, also not for markup of person names in titles (which is a rare thing) and for markup of titles in languages that don't follow English capitalization rules. I'm from Germany after all. Braces have themselves established as good readable markup for these cases.

I dispute the statement that it's necessary (and/or sensible) to put the 1st character of any word in the title in braces, which is a completely other horse.

> The rules for title capitalization depend on publication venue, locale, and other things.

Being the author of xindy, I know more about different rules for capitalization than I ever wanted. They are among the reasons why I invented merge rules in xindy.
See, I wrote my first BibTeX style file in 1986 (Oren published BibTeX in 1985...) -- I know exactly what you mean. These problems contribute to my statement that usage of BibTeX doesn't make sense any more. BibLaTeX is more flexible. (Although correct sorting is often still a problem.)

Still nothing better than BibTeX format?

Posted Apr 9, 2026 19:35 UTC (Thu) by ejr (subscriber, #51652) [Link]

Hah! xindy is incredibly useful, thank you.

I mentally lump BibTeX into BibLaTeX, although so many processors still rely on ye olde BibTeX. I've been trying to move to CSL JSON + citeproc. It isn't trivial. Staying with BibTeX *is* trivial for lots of people in computing. I don't have great ideas on how to move the needle, or at least I don't have time to really delve into bib -> json conversion.

Although... The "document understanding models" may make much of this irrelevant. If you just cite the *document* and let the agent gather the rest of the information... hmm. I suspect many of us already to that manually via BibTeX exports, etc. But given the success in LLM-ish OCR systems in responding to tax form queries (as a timely example), they entire input format problem may go away-enough. Then there's the human feedback loop of making sure papers *can* be cited that way. We'd close that pretty quickly, I suspect.

In some of my fields, we can rely on incredibly well-curated BibTeX files maintained by a few. But, as people may notice, the few keep growing... fewer. And few are stepping up to continue their efforts.

Still nothing better than BibTeX format?

Posted Apr 9, 2026 21:21 UTC (Thu) by ceplm (subscriber, #41334) [Link]

I don’t write anything significant anymore (twenty years I switched from academia to being a programmer/packager), so just from distance I am fascinated that there is not greater traction in the LaTeX world for BibTeX replacements like Zotero or for example papis (https://github.com/papis/papis). Any thoughts?