4 ms·
we need to publish documents in a new format. pdf is ridiculously restrictive and makes training the language models a nightmare.
by behnamoh 4y ago
we need to publish documents in a new format. pdf is ridiculously restrictive and makes training the language models a nightmare.
- SllX 4y agoWell there’s this thing called HTML that is pretty popular. You can see an example of it here: https://motherfuckingwebsite.com/ https://motherfuckingwebsite.com/ You even have formatting options as discussed here: http://bettermotherfuckingwebsite.com/ http://bettermotherfuckingwebsite.com/
- jaywalk 4y agoIf I send someone a PDF document, I know exactly how it will look for them on screen and on paper. HTML, not so much.
- SllX 4y agoSure. Part of that is PDF’s restrictiveness. We either value the consistency, value the flexibility or pick the appropriate format for the application. Basic HTML gets you something readable with any HTML renderer, and you can pretty much count on it, but not to the same degree or certainty as you can with a PDF (and bad PDF renderers exist too). You can still read a book in it. HTML+other stuff is where it gets dicey, and temptations like on-demand loading and endless scrolling, or “rich” features like dictionary lookups (that my OS already natively supports in any application from any text selection including the web browser) start to make reading a book this way sound like a bad idea and PDF starts to sound a lot more appealing. All I’m saying is when PDF starts to sound like the wrong choice, consider HTML. Not HTML plus stuff, just HTML. Maybe a very very light touch of CSS if you’re feeling fancy, but it’s really not necessary.
- notRobot 4y agoHTML should be used for books and stuff. (Accessible) PDF should be used when you want to control exactly how it looks on the receiving end, but this should be a rare use case.
- Wowfunhappy 4y agoOutside of actual raster image formats, PDF is the only file type I can export to and be confident it will render correctly on the recipient's machine. SVGs are a mess, MS Office documents are a crapshoot unless everyone is using official Microsoft Office, and I probably don't need to even discuss HTML. There are PDFs out there which will only render correctly in Adobe Acrobat, but unless I'm using some niche Adobe-only feature, I basically never encounter problems. I've also heard that PDFs are technically horrible under-the-hood, but for some reason they just work like nothing else does!
- jaywalk 4y agoAs sad as it is, the reason PDFs just work like nothing else does is probably because they're technically horrible under the hood.
- merb 4y agoThey are so horrible that you still need to have some quirks in your reader, because some important tool still generates it incorrectly , but it is still correct. Then there are loops and some software Try to parse them. The encryption types, the security, which is bonkers… why should it be possible to disallow printing or analyzing it with another software, there are attachments, forms, JavaScript… And then there is xfa (Especially the dynamic one), a format in a format which basically totally different than a pdf. But it is inside the pdf container and people expect that you understand it. Thank god that pdf 2.0 probably kills it.
- anthk 4y agoPDF is for printing. Use text formats for everything else. Or, as a PDF alternative, DJVU.