Yes book does mean TeX
November 13, 2008 4:34 PM   Subscribe

How should one present books online?

I'd say the main constraint is that the original source are latex & tex expressing mathematics and using hyperref for hyperlinks. So we've easily got the whole book available in one pdf as also as separate pdfs for each chapter, which requires modifying hyperref or editing the embedded links. But what about lower bandwidth options?

Are there any open source web app that downloads pages when needed ala Can any browsers render pdfs inside the page? Or must all such apps use images?

PDFtoHTML can produce beautiful output, but doesn't support pdf's produced from LaTeX. Any idea if this can be fixed easily?

DVI2html merely produces images files for each page. LaTeX2html seems sublimely ridiculous (for example, theorem environments are converted to images). HyperLaTeX seems workable for minor projects, but too restrictive for books. TTH seems like the best converter by far, even producing MathML, but their pages often require reconfiguration of the browser. Btw, I've noticed that wikipedia doesn't use MathML directly themselves, but typesets the equations using amstex.

Can one produce html output any more easily from ConTeXt?
posted by jeffburdges to Computers & Internet (6 answers total)
Would Scribd be useful to you?
posted by leigh1 at 6:08 PM on November 13, 2008

Response by poster: Interesting, but not really. I'm thinking vaguely between github plus latex and wikibooks. So mostly offline using git for the awesome forking & merging, but having limited wiki-like elements. Youtube seems reasonable as a service due to bandwidth, but books needn't be that large if rendered in html, or even pdf, so a service seems excessive. It'd be worth hosting books there too for publicity, once you had such a site already developed, but I'm only taking stock of technical considerations just now.
posted by jeffburdges at 8:28 PM on November 13, 2008

i have super limited experience with tex or any related programs. HOWEVER, i can say that converting theorems/formulas to images is not ridiculous at all, and is common in all the journals i work with online. it ensures that all the math stuff is being accurately presented, which is sometimes hard to do with html coding alone. so if that's you're only complaint with latex2html, i'd say go for it.
posted by misanthropicsarah at 8:21 AM on November 14, 2008

If HTML is okay, consider distributing the book in ePub format. ePub is an open standard which uses XHTML as its content format. More on ePub.

ePub would allow you to embed MathML directly with images as a graceful fallback. Allowing any form of XML (provided there's a widely-usable alternative representation as well) is one of its key strengths.

MeMail me if you have more questions. (I've also written a web application called Bookworm that allows for ePub books to be read online in a browser.)
posted by nev at 11:24 AM on November 14, 2008

Response by poster: LaTeX2HTML converts almost all environments to images, including those without math, multi page proofs, abstracts, etc., plus virtually no packages are supported.

It seems calibre is the canonical open source pdf to lrf (ePub) converter, but it fares no better than PDFtoHTML. You can always produce properly formatted pdf files from latex, so I doubt there is any point in using ePub for math or physics textbooks.
posted by jeffburdges at 4:57 PM on November 14, 2008

Response by poster: I imagine the answer must be that TTH does exactly the conversion I want, but MathML is far too limited, so TTH adds funky fonts, which breaks everybody's browser. It seems all converters either produce image files from tex & dvi files or else horrible mutilate pdf files. I found that the pdf mutilators destroy even the most basic paragraph layout, so presumably pdf supports multiple layers of layout primitives while these programs support only the high level ones.
posted by jeffburdges at 5:06 PM on November 14, 2008

« Older Vegan Chili   |   Using copyrighted music in a student film? Newer »
This thread is closed to new comments.