[PaulHoule]

Personally I love reading PDF files on my iPad (or cheap Android tablet for that matter.) I can load 100 books and even more technical papers and read them on the bus, spinning at the gym, in bed, curled up on the couch, as a passenger in my car, etc.

Posting papers from arXiv to HN I have a choice of pointing to the HTML abstract or the PDF and overall people express preference for the PDF and I believe I get more upvotes for the PDF.

I can see a phone being a little too small though. PDF does have features that make it possible to make a PDF which is reflowable like an HTML document for the sake of accessibility and better UX, but those features are complicated as hell and rarely used.

[a2800276]

_all_ PDF features are complicated as hell apart from the "Export to PDF" feature of your word processor :D

[PaulHoule]

My #1 complaint is that for all of the complexity there is no primitive to draw a circle in PDF, instead any ‘circle’ you see in a PDF is an approximation based on Bézier curves. (I thought I was uniquely untalented because I couldn’t draw anime characters with Bézier curves but when I looked at the resource packs for my favorite game I found my favorite illustrator couldn’t do it either.)

I did a deep dive into PDF because I was pitching and sketching out the ultimate test extraction system for corporate use and figured PDF was the most important format to support and I could say boy is it tricky to get the correct text out with the structure properly mapped unless some of those obscure features were in use and it would still be pretty hard then.

[a2800276]

We tried to extract pseudo-code test procedures out of an ISO Standard and ended up developing heuristics of how to piece the text together based on the color of the syntax highlighting. Terrible, terrible format.

At the moment I'd like to add OCR text overlays onto scanned documents and just can get myself to deal with the format for longer than ten minutes until my brain gives up.