Hacker News Reader

Comment by karanbhangui | Hacker News Reader

the main issue I have with native PDF annotation is that it mutates the file, which changes the hash. I've been thinking about building a collaborative PDF reader (using native PDF toolkits, not pdf.js) that fingerprints the file locally and pulls/pushes annotations to a central repository (either self hosted by individuals, research teams) or globally. Would love to see annotations on popular research papers or textbooks.

View on HN

What you're suggesting would be amazing.

For what it's worth, the popular "Skim" PDF reader on OS X stores annotations in filesystem extended attributes, so the hash of the raw file data doesn't change (but of course this comes with the drawback of more limited interoperability with other tools).

There's Fermat's Library [1] which allows for collaborative annotation on pdfs. The only issue I have with them is that they lack highlighting and aren't open source.

With PDF annotation, why not separate the content from the content representation? Have an editor for the original content, and export to pdf. If people want to annotate, give the original content or a copy. Easier to version control this way as well.

[1] https://fermatslibrary.com/

Hey, author here! Polar Bookshelf does exactly what you want, I recommend trying it out. It's mentioned in the article here https://beepb00p.xyz/annotating.html#polar

At least in principle it is possible to annotate a PDF file while leaving the original file as a prefix of the end result. The directory of objects is at the end of the file, so the annotation software can build a new tree of objects, referencing any unchanged parts of the original, and append its own directory which the reader software will find. I don't know of any PDF editors that do this, though.

this really depends on the annotation app. ISO-conformant PDF viewers support XFDF annotations, which can either be burned into the PDF, or stored separately. the behaviour is entirely viewer-specific.

hypothes.is does exactly this, except it uses pdf.js

It doesn't matter where you load a PDF from (locally or any online source) the public annotations are overlaid.

Reply on news.ycombinator.com