Chips these days have monster technical reference manuals. I usually use Okular (KDE) for large PDFs, but text search is slow - as in minutes - and each subsequent search is slow.
So I tried Evince (Gnome), and it only took about 10s to search an 8000-page document. It also showed the search results in the side panel so you can quickly jump through them. Very efficient.
Okular seems to have better annotation and markup tools.
Different tools are often better at different things.
Is it acceptable to feed the PDF to some “indexer” and search/prompt it instead of plain PDF search?
Asking as I have been working with relatively extensive documentation (it does not seem as large though), and chose to feed it to an indexing model. I can now ask basic questions and get specialized answers (and just scratched the surface what can be done with the many tools out there across vector DBs, agent harness like OpenCode and complementary software like OpenViking).
That is a great idea! Do you have a recipe for how to set this up at a personal level?