[frabjous]

Originally Posted by ahi

A .tex source file can no more generate a typographically optimal document for an arbitrary size than an HTML file can (orphans, widows, words needing hyphenation will crop up at different places for different size screens... issues that are not machine-solvable).

LaTeX had its own hypenation rules, and you can embed custom hyphenation rules for non-standard words in the document itself. I don't see that as an issue at all. In fact, I think it does a better job than most humans would.

LaTeX also has thousands of widow and orphan control settings.

Perhaps what you're pointing out is that you'd want it to work differently on different sized screens. I noticed this when I was creating different sized PDFs for what I was working on: using the same orphan control rules for an i-phone sized page as for a A4 sized page is ridiculous. But all of that can be properly coded into a LaTeX package that scales the badness or orphans and widows accordingly.

Claiming that this isn't "machine solvable" just seems false... unless you're claiming that the rules typographers use when making this hand-corrections can't even be made into an algorithm in principle... and I can think of no evidence for such a claim.

Some people may disagree, but I don't think there need to be infinitely many sizes. 4 or 5 sizes seems like enough for eBooks. A little investigation will reveal what the best rules for each size is, and templates made for each, streamlining things for future production.

This is a very dead horse though, friend... and I suspect an evil or possibly cursed one.

Yes, you're probably right.

If you are in the habit of making your own eBooks via LaTeX/XeTeX, do drop me a private message.

Well, I've got one up here, and I'm working on more.

Originally Posted by DawnFalcon

TeX is massively more complex than ePuB,

Comparing apples and oranges. An ePub file is not more complex than a TeX file. If anything, a TeX file is simpler. If you mean the rendering engine, you need to compare TeX to the HTML-interpreting rendering engines used to display web content. It's true that TeX is huge if you include ALL of the community-made content, as you noted, devoted to particular disciplines. You wouldn't load all of that on to your reader. For specialized things you'd just embed it along with the TeX source as needed, much like you embed images etc., along with the HTML inside an ePub. The core engine is no larger than your average web browser is, and works more efficiently, given that TeX was written by one of the foremost experts in algorithm theory in the country.

The "specialized subset" already exists: it's just the core LaTeX engine -- I suppose you'd need to add in the most often used packages, like graphics and the AMS math packages, and geometry for setting the screen size properly.

I don't sure about the ambiguity in the TeX license, but it's treated as fully open source by most of the open source community, and I sure as heck don't imagine Donald Knuth is going to sue anyone over anything.

But if you think it's just as easy to advance the HTML rendering engines for ePubs, and think this can be done in a reasonable span of time, with the same quality you already find with TeX--Great! But I'm not holding my breath.