PDF is dead. Or is it not?

[llasram]

Originally Posted by rlauzon

The point is that with PDF, I have to know the screen sizes of all the devices I intend to support before hand.

Yes and no.

"Yes," in that the target page size needs to be known prior to generating the PDF, but for an "ebook print-on-demand" service, that's not a problem. One could produce a service right now which dynamically generated books to users' custom size requirements using open source tools like TeX.

"No," in that page-oriented typesetting systems generally provide many useful features for tweaking the typesetting of a work. Although the core engine may generally provide good overall typesetting, generally (well, at least w/ TeX) a certain amount of manual tweaking is required to eliminate the occasional "warts," such as orphans/widows, paragraphs w/ poor greyness, and bad line-/page-breaks around figures, quotations, and poetry. Most of these tweaks will depend exactly upon the final layout, and thus the page size.

Systems based upon infinitely reflowable content definitionally lack many of these features, which in the worst-case can result in very bad typesetting the author has no recourse for repairing. The most often-occuring offender here is lack of hyphenation and the resulting horrible greyness -- especially the "rivers of whitespace" it can cause in fully-justified text. I consider this one a bit of an odd case though, especially for ebook devices. For example, the Sony Reader already needs to pre-calculate linebreaks in the reflowable formats it supports, so why not simply run the Knuth-Liang algorthim used by TeX, calculating the line-/word-breaks for future pages in the background while the user begins reading?

And so the conflict rages ever onward...

[RWood]

One of the main uses of PDFs is to send comps to a printer in exactly the format that it will be printed. As a prepress tool it is fantastic. I have also used it for many years (since version 1 from Adobe) to create exact copies of documents that were submitted to the Government -- reports, contracts, and invoices to name three. In this area PDFs are alive and well. They record the document at a specific time and are not subject to changes in the original software as future versions are released. (Case in point is the change in IE rendering around version 5.)

[Hadrien]

Originally Posted by llasram

"Yes," in that the target page size needs to be known prior to generating the PDF, but for an "ebook print-on-demand" service, that's not a problem. One could produce a service right now which dynamically generated books to users' custom size requirements using open source tools like TeX.

There is such a service: http://beta.feedbooks.com

[rlauzon]

Originally Posted by Hadrien

As for rlauzon, I guess you didn't understood a single word I said: the right choice is to make no file at all. Make the right file for the right user in an automatic way.

Yes, I understood you. I'm just not blinded by PDF.

So, what you are saying is that it's better to keep the eBook in a certain format, then, when the user wants to read it, run it through a process that will create an PDF formatted for the user's device.

Why not just keep the ebook in a reflowable format in the first place?

And what happens when you buy your eBooks for you iLiad, but then want to view them on your Palm? We're right back to the problem with PDF.

Originally Posted by Hadrien

The best choice would be a good reflowable format with advanced formatting support, but there's no such thing

Yes, there is.

Originally Posted by Hadrien

RTF is way too old and limited with very few features,

Yet it's far better an eBook format that PDF can be.

Originally Posted by Hadrien

HTML will never look as good as something formatted with LaTeX etc... The best choice would be XML based, yet support advance formatting features.

Yes, XML would be the best, but you do realize that HTML is just a specialized form of XML, right?

[Azayzel]

IMO, PDF won't be dead until there is a more user-friendly alternative.

Right now it is too easy for anyone to create free digital content via the plethora of PDF-printing solutions available. When many companies, and a lot of banks, have switched over to archival media incorportating PDF'd content, I don't see it dying any time soon (think how much it would cost to convert from PDF to the new solution for all of the companies that have been doing it for the past 5-10 years, not to mention the ones that just switched).

[Hadrien]

Well we have quite a different view on what advanced formatting is in this case (I see the same issues with reflowable formats than llasram).
When I have a RTF file on my reader and it's lacking hyphenation, zooming in and out is pretty much useless for me, the justified text is way too ugly for my taste with all these whitespaces around.
HTML is not a specialized form of XML, XHTML is, and XHTML formatting options are not suited for e-books either (the same can be said for CSS).
E-books will not be "platform agnostic" as easy as music files, if you just think about reflowable formats without the fact that it'll be rendered on all kind of screens, you'll never get any e-book as good looking as a real paper book. We'll need advance formatting that reflowable formats do not provide (at least yet), that could be software based or file based. It's not a fight for reflowable formats vs page-oriented format, it's about finding a way to keep the best of both worlds.

I see too many people screaming around that PDF is dead. Well, as long as no reflowable format provides similar formatting features, PDF won't be dead. When such a format appears and start being widely available for mobile readers, then both reflowable and page-oriented formats that we know of will be dead.

[NatCh]

Originally Posted by rlauzon

Yes, XML would be the best, but you do realize that HTML is just a specialized form of XML, right?

Heh, I thought XML was a specialized form of HTML.

If I recall correctly, BBeB is XML based. I'm not pushing it, mind you, just pointing that out to stir the discussion's pot a bit.

[Hadrien]

Originally Posted by NatCh

If I recall correctly, BBeB is XML based. I'm not pushing it, mind you, just pointing that out to stir the discussion's pot a bit.

Yes and BBeB is much better than HTML for eBooks though it lacks certain advanced features (+ it's only supported by Sony, so it's pretty unusable as a standard format I'm afraid).
If Sony could improve BBeB a bit more it would be a great start but I guess we'll need something a bit more open source friendly too...

[NatCh]

Oh, I absolutely agree that any standard format would have to be Open.

However, just because something is currently only supported by a single company doesn't mean that it has to stay that way -- PDF, for instance, used to only be supported by Adobe.

[Laurens]

Originally Posted by NatCh

If I recall correctly, BBeB is XML based. I'm not pushing it, mind you, just pointing that out to stir the discussion's pot a bit.

The BBeB LRS source format is, indeed, XML-based, but the BBeB LRF publication format is a binary representation of this XML source. Hopefully, Sony will also publish some form of official LRF specification, rather than have us rely on reverse-engineered documentation. A basic outline of the file layout and LRS to LRF mapping would be enough.

[rlauzon]

Originally Posted by Hadrien

When I have a RTF file on my reader and it's lacking hyphenation, zooming in and out is pretty much useless for me, the justified text is way too ugly for my taste with all these whitespaces around.

Hyphenation is a function of the program that created the RTF, not a problem with RTF.

Zooming is a function of the reader, not RTF.

Just because your reader poorly supports RTF doesn't mean that RTF is a poor format.

Originally Posted by Hadrien

I see too many people screaming around that PDF is dead.

PDF has no benefits and significant drawbacks when used for eBooks. For eBooks, read on small devices, PDF is dead.

Originally Posted by Hadrien

Well, as long as no reflowable format provides similar formatting features, PDF won't be dead. When such a format appears and start being widely available for mobile readers, then both reflowable and page-oriented formats that we know of will be dead.

We don't need similar formatting features. The vast majority of books sold today do not require anything more than some text highlighting (i.e. bold, underline, etc.). The old page-oriented book idea does not make sense for eBooks.

What we need is to rethink how books are displayed to the user. With eBooks the page-oriented, big margins, absolute positioning (i.e. everything that PDF is) simply doesn't make sense anymore.

[Hadrien]

Hyphenation is linked to the screen size, so it has nothing to do with the program that created the RTF file. It is a problem with all reflowable format. You could support it, but with a software on each reader and some sort of dictionnary too (won't be easy to do such a thing for multiple languages).

RTF doesn't support embedded fonts, do you believe that each and every book in your library use the same font ? They don't.

Maybe YOU just need simple features and a poor formatting system, but I don't believe that's the case for everyone. Most of the people screaming around that PDF is dead don't know anything about LaTeX and just had a poor experience with A4 PDFs once. The portability of PDF files may be an issue, but the poor formatting of RTF is an issue too. If you really believe that RTF is perfect and that PDF is the devil, well then good for you... I'm looking forward for something both portable and powerful, but I know that it'll be hard to make such a format. In the meantime, I believe the best solution is to store books in a database or using XML, and create files out of it (some editors use LaTeX already). Keeping all these books in a single text file (TXT or RTF) is the wrong choice, you need to know where the chapters start and end for example if you want something ready for any advanced format. With such a system you can create what you call "absolute positioning" in a relative way: each device and each person can have his/her own formatting. You can also create pure text files out of it, but that would be like downsampling an audio file. When and if a powerful XML format become available, the same system will still work, creating just another kind of files as an output.

[Laurens]

The world has changed significantly. Magazines used to be the cultural centre of every possible topic—and they used to sell well. Today, the internet is slowly but surely taking the place of magazines—and taking more and more advertising money away from paper publishing.

It's important to note that the original quote comes from a technology magazine, because the assertion above doesn't apply to every magazine category.

Lifestyle/fashion (Cosmopolitan, Vogue, etc.) magazines, for instance, still work best within paper format and would make a significantly worse impression if they were offered only as a web site. I couldn't imagine an electronic National Geographic having the same impact either. Aside from that, electronic publications are less prestigious, simply because everyone can do it.

Relatively dry, technical publications like FSM are well-suited to the web site format, however.

[Steven Lyle Jordan]

Most magazines can be optimized for web and for print presentation... go online, and you'll find quite a few--some do it better than others.

Okay, maybe PDF isn't well suited for e-books (although I've clearly not had the problems with tagging and reflowing that others, like rlauzon, have had). I've argued for PDF e-books in the past, but I agree that, for a document provider, offering multiple PDF formats for multiple readers isn't worth the effort... better to offer something that the user can customize into PDF themselves, like RTF or HTML. Or standardize everyone into one format and size in the first place.

But PDF is still perfectly suited for documents that are going to end up on paper. And since we won't be losing paper documents anytime soon, PDF is far from dead.

[Hadrien]

Originally Posted by Steve Jordan

better to offer something that the user can customize into PDF themselves,

Users shouldn't be the one customizing anything. There's a whole lot of people that are used to tech around here, but that's not the case of a more mainstream public. A service should customize automatically e-books for the users, or even better there shouldn't be any customization at all (but we don't have the format and software for this yet).