5 ms·

I'd like to see you try to have a conversation on a tech & startup news aggregator built in PDF, see how quickly your reader loads it then. You're talking about

I'd like to see you try to have a conversation on a tech & startup news aggregator built in PDF, see how quickly your reader loads it then. You're talking about PDF like the only documents you've seen are printed from LaTeX / Chrome, but PDF supports forms, javascript, 3D models and more. PDF is an atrociously bad format, and I don't know what "multi-hundred page PDF loads in the blink of an eye" for you but even a 100 blank page PDF takes nearly a second to fully load on my beefy rig (I did the test a few months back to prove a point). [Edit: Other commenters made the clarification below, but single page render time is not the same as document render time] Clearly extracting text from a PDF is nearly as difficult as extracting it from a photo. Digitally extracting information from PDFs in general is awful, which makes the format awful for the various things it's used for. Not to mention that many uninformed users today still install the garbage / malware PDF readers such as Acrobat because they don't know any better.

gspr 6y ago
> I don't know what "multi-hundred page PDF loads in the blink of an eye" for you but even a 100 blank page PDF takes nearly a second to fully load on my beefy rig (I did the test a few months back to prove a point). The manual for PGF/TikZ [1] is a huge PDF I frequently open. It's more than 1300 pages and has lots of graphics. It opens and navigates in the blink of an eye on my 3 year old laptop (with the Okular reader). PDFs aren't perfect, but they sure feel spiffy compared to modern webpages. I do agree with some of the article's complaints, but not this one. [1] http://mirrors.ctan.org/graphics/pgf/base/doc/pgfmanual.pdf http://mirrors.ctan.org/graphics/pgf/base/doc/pgfmanual.pdf
- athriren 6y ago
  What reader do you use?
  - gspr 6y ago
    Okular [1]. It's strange; I'm a KDE user and big fan of the core DE, but I find almost all the KDE software outside of that core DE nearly unusable. Except Okular – it's by far the best PDF reader I know. I guess credit goes to Poppler for the heavy lifting [2]. [1] https://okular.kde.org/ https://okular.kde.org/ [2] https://poppler.freedesktop.org/ https://poppler.freedesktop.org/
    - leephillips 6y ago
      Try zathura, if you haven’t. Super fast, and keyboard oriented.
      - savingsPossible 6y ago
        thanks! really cool
    - andrepd 6y ago
      I use qpdfview and it works very well.
- tehabe 6y ago
  That is created using LuaTeX and I'm sure the sources behind that PDF document are carefully crafted and LuaTeX works really well. But if you would do the same document with the same amount of images in Microsoft Word and create a PDF document is would be much much bigger and it won't load that quickly. I will take the last part back, if someone can prove that I'm wrong about Word and PDF documents.
  - gspr 6y ago
    In that case it sounds like a problem with Word and not with PDF. I wouldn't know – most PDFs I consume are generated by some variant of TeX. I gave a random 300-page datasheet I have lying around a go. It says it was made with Acrobat Distiller and "C2 Rendition". Feels just as spiffy as the PGF/TikZ manual.
    - tehabe 6y ago
      All that I wanted to say is: not all PDF documents are created equal, some are really well and some are just awful.
- 7952 6y ago
  The speed depends a lot on how the PDF is structured. If you export a complex CAD drawing you may have a ridiculous amount of detail that has to be fully rendered before the page can be viewed. Or you can have very simple PDFs that are just a few images.
- andrepd 6y ago
  If pretty sure he's opening it in the browser.
rayiner 6y ago
> I'd like to see you try to have a conversation on a tech & startup news aggregator built in PDF, see how quickly your reader loads it then. Sure. I agree we shouldn’t replace interactive web apps with PDF.
LoSboccacc 6y ago
weird, my Alfa Romeo user manual is 270 pages filled with graphics (literally, they are jpeg scanned to a pdf) and loads instantly even on my mobile phone
- corty 6y ago
  The first page is rendered instantly you mean. PDF, at least when generated by a sane generator, can be parsed pagewise. HTML cannot, you always have to parse everything in a page to do layout, because later objects can change or overlay earlier ones.
  - Santosh83 6y ago
    This is being partly addressed in the latest draft CSS specifications... https://news.ycombinator.com/item?id=24093273 https://news.ycombinator.com/item?id=24093273
  - JKCalhoun 6y ago
    Is PDF still unstreamable? AFAIK, the TOC (catalog?) in a PDF was located at the end of the file, meaning the whole PDF had to come down in order to parse the PDF. (With the exception of the first page, as you say — some aspect of the PDF spec allowed for a self-contained page 1.)
    - nogridbag 6y ago
      There are libraries to linearize PDFs but not all PDFs can be converted. Some of the more popular open source PDF libraries do not support it though.
    - hnick 6y ago
      Linearized PDF has existed as a concept since v1.2 which I think was released in 1996. You can see it mentioned in the v1.4 spec at https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/pdf_reference_archives/PDFReference.pdf https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/pd... [follow the link from the contents].
  - 6y ago

gspr 6y ago

> I don't know what "multi-hundred page PDF loads in the blink of an eye" for you but even a 100 blank page PDF takes nearly a second to fully load on my beefy rig (I did the test a few months back to prove a point). The manual for PGF/TikZ [1] is a huge PDF I frequently open. It's more than 1300 pages and has lots of graphics. It opens and navigates in the blink of an eye on my 3 year old laptop (with the Okular reader). PDFs aren't perfect, but they sure feel spiffy compared to modern webpages. I do agree with some of the article's complaints, but not this one. [1] http://mirrors.ctan.org/graphics/pgf/base/doc/pgfmanual.pdf http://mirrors.ctan.org/graphics/pgf/base/doc/pgfmanual.pdf

athriren 6y ago
What reader do you use?
- gspr 6y ago
  Okular [1]. It's strange; I'm a KDE user and big fan of the core DE, but I find almost all the KDE software outside of that core DE nearly unusable. Except Okular – it's by far the best PDF reader I know. I guess credit goes to Poppler for the heavy lifting [2]. [1] https://okular.kde.org/ https://okular.kde.org/ [2] https://poppler.freedesktop.org/ https://poppler.freedesktop.org/
  - leephillips 6y ago
    Try zathura, if you haven’t. Super fast, and keyboard oriented.
    - savingsPossible 6y ago
      thanks! really cool
  - andrepd 6y ago
    I use qpdfview and it works very well.
tehabe 6y ago
That is created using LuaTeX and I'm sure the sources behind that PDF document are carefully crafted and LuaTeX works really well. But if you would do the same document with the same amount of images in Microsoft Word and create a PDF document is would be much much bigger and it won't load that quickly. I will take the last part back, if someone can prove that I'm wrong about Word and PDF documents.
- gspr 6y ago
  In that case it sounds like a problem with Word and not with PDF. I wouldn't know – most PDFs I consume are generated by some variant of TeX. I gave a random 300-page datasheet I have lying around a go. It says it was made with Acrobat Distiller and "C2 Rendition". Feels just as spiffy as the PGF/TikZ manual.
  - tehabe 6y ago
    All that I wanted to say is: not all PDF documents are created equal, some are really well and some are just awful.
7952 6y ago
The speed depends a lot on how the PDF is structured. If you export a complex CAD drawing you may have a ridiculous amount of detail that has to be fully rendered before the page can be viewed. Or you can have very simple PDFs that are just a few images.
andrepd 6y ago
If pretty sure he's opening it in the browser.

rayiner 6y ago

> I'd like to see you try to have a conversation on a tech & startup news aggregator built in PDF, see how quickly your reader loads it then. Sure. I agree we shouldn’t replace interactive web apps with PDF.

LoSboccacc 6y ago

weird, my Alfa Romeo user manual is 270 pages filled with graphics (literally, they are jpeg scanned to a pdf) and loads instantly even on my mobile phone

corty 6y ago
The first page is rendered instantly you mean. PDF, at least when generated by a sane generator, can be parsed pagewise. HTML cannot, you always have to parse everything in a page to do layout, because later objects can change or overlay earlier ones.
- Santosh83 6y ago
  This is being partly addressed in the latest draft CSS specifications... https://news.ycombinator.com/item?id=24093273 https://news.ycombinator.com/item?id=24093273
- JKCalhoun 6y ago
  Is PDF still unstreamable? AFAIK, the TOC (catalog?) in a PDF was located at the end of the file, meaning the whole PDF had to come down in order to parse the PDF. (With the exception of the first page, as you say — some aspect of the PDF spec allowed for a self-contained page 1.)
  - nogridbag 6y ago
    There are libraries to linearize PDFs but not all PDFs can be converted. Some of the more popular open source PDF libraries do not support it though.
  - hnick 6y ago
    Linearized PDF has existed as a concept since v1.2 which I think was released in 1996. You can see it mentioned in the v1.4 spec at https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/pdf_reference_archives/PDFReference.pdf https://www.adobe.com/content/dam/acom/en/devnet/pdf/pdfs/pd... [follow the link from the contents].
- 6y ago