My understanding is that PDF is a monster of a document format, and it's clearly not (usually and historically) meant to be reflowed. Even copy/pasting from PDFs can be very disconcerting because the viewer may not have a good idea of where blocks of text start and end (or even what the characters really are).
I can empathize with the feeling that the web is incredibly bloated, but that's IMO throwing the baby with the bath water. Simple HTML with some optional CSS would do the job much better IMO (and can be easily downloaded, mirrored or offlined with tools like wget).
And if you really don't like writing HTML (I won't blame you) then there's always formats like markdown, org-mode and friends which can easily be converted to pretty much anything.
Dealing with PDFs (as in, coding a system that can import/export/display them) is more obnoxious than dealing with excel spreadsheets.
Unless your system is a PDF library (as in, you make the black-box dependency that other systems use to handle PDF exports), everything you do with PDFs will be through some annoying black-box dependency that is a pain to use.
Even relatively complex HTML is much more fun to work with than PDF.
I can empathize with the feeling that the web is incredibly bloated, but that's IMO throwing the baby with the bath water. Simple HTML with some optional CSS would do the job much better IMO (and can be easily downloaded, mirrored or offlined with tools like wget).
And if you really don't like writing HTML (I won't blame you) then there's always formats like markdown, org-mode and friends which can easily be converted to pretty much anything.