Master Format for multi-format eBook Generation?

[cerement]

Mainly because I realize we are still in the "eBabel" stage of eBook formats, I'm wondering which format, so far, has proven to be most reliable and flexible when being used as a master or archival format for generating or being transformed into the many output formats (PDF, ePub, mobi, lit, lrf, etc., etc., etc.)? (And yes, I understand, your choice of output formats is still determined by what tools are actually available.)

XML Based:

ePub seems to be the dominant choice for a output format, but it doesn't feel quite rigorous enough yet for the master format (even in something as simple as confusion over using <q> vs " for quotes (related DTBook solves this the wrong way, by using both) -- but then, this is simply a matter of giving it time to mature).

TEI looks interesting and has been around for awhile, has the maturity, and has a couple useful implementations. TEI-Lite being the most well known. At one point, Project Gutenberg was working on PGTEI, but the status of that seems to be unknown at this point although documentation is clear and easy to understand.

Then there's the father (DocBook) and the grandfather (SGML) formats ...

TeX Based:

Mainly the LaTeX macros (and I love Donald Knuth's writing style and rigorous attention to detail). And while TeX can be output to almost anything, the conversion process between TeX and XML seems to still be pretty vague (although its ability to generate table-of-contents from full files would be very useful for the multi-file format of ePub).

[DDHarriman]

I would say plain html and associated images if they exist.
[I personally also keep a copy of it in word (1997-2003), txt and pdf, just in case - but that is just me…]

Best

[JSWolf]

Given all the eBook formats out there, the best format right now is ePub. You can take this ePub and use it to convert to LIT, Mobipocket, and LRF using Calibre. Plus, you'd have it in ePub already.

[Andybaby]

if you have a perfect copy. Epub is a good storage format.

If you have a non perfect copy, HTML or RTF, are good storage and editable formats

[ilovejedd]

For fiction, I'm leaning towards something XML-based for a master/source format. The structure lends itself quite well to digital publishing/ebook transformation. epub's a good example of that. Even Microsoft's .LIT is somewhat XML based, albeit, encapsulated in a prioprietary DRM format. I didn't check the numerous implementations but I think for a master file, an XML with a simple layout would be ideal.

<xml>
  <title>some title</title>
  <author>some author</author>
  <cover>some cover.jpg</cover>
  <published>some date</published>
  <description>summary</description>
  <toc>
    <entry>
      <title>Chapter 1</title>
      <description>short chapter summary?</description>
    </entry>
    <entry>
      <title>Chapter 2</title>
      <description>summary/whatever</description>
    </entry>
    <entry>
      <title>Chapter 3</title>
      <description>summary/whatever</description>
    </entry>
  </toc>
  <entry>
    <title>Chapter 1</title>
    <content>Chapter 1 text here (possibly in HTML/XHTML format)</content>
  </entry>
  <entry>
    <title>Chapter 2</title>
    <content>Chapter 2 text here</content>
  </entry>
  <entry>
    <title>Chapter 3</title>
    <content>Chapter 3 text here</content>
  </entry>
</xml>

Another option, instead of embedding chapter content into the file (might make for gigantic XML files for long books)

<content>Chapter 1 text here (possibly in HTML/XML format)</content>
can be changed to
<link rel="alternate" href="chapter1.html" />

Mind you, I'm not very familiar with XML as evidenced above, but the sample layout should give you a general idea. From the XML, it would be fairly trivial to convert to other formats. I think the least formatting available on the master XML, the better since that gives ebook creation/conversion utilities final say in how to handle style, flow, etc.

For reference and technical material, I think TeX is the obvious choice.

Edit:
Oh, whoops. Misunderstood the question. Just goes to show if you can barely keep your eyes open, you have no business replying on forums. To answer your question, given currently available software and utilities, I'd keep ebooks in both ePub and single HTML format. ePub for the metadata, cover, etc. Single HTML files for ease of editing and reading on the computer. As much as I love Calibre for general ebook management, I don't really like its built-in viewer.

[HarryT]

For me it's a folder for each book, containing HTML source and whatever images are used in the book.

[mtravellerh]

Epub (or zipped xhtml) which is basically the same

[Sweetpea]

I am currently in the process of creating all my HTML files into an epub format directory structure. Which will give me a directory per book (which can contain other directories)

[Kirtai]

I currently use html though I also find it's not rigorous enough.
I hate that things like chapter headers are marked up merely by convention instead of properly and would really like to find a good format that does allow proper markup without being excessively complicated or rigid.

[cerement]

Originally Posted by Kirtai

I currently use html though I also find it's not rigorous enough.
I hate that things like chapter headers are marked up merely by convention instead of properly and would really like to find a good format that does allow proper markup without being excessively complicated or rigid.

So far, for accuracy, TEI-Lite (a subset of TEI) seems to be the best bet. Stable, been around for awhile, regularly used in humanities. Does suffer from being a little complicated. TEI does have the advantage of allowing snippets of TeX for formulas. ePUB Books started using a "Super-Lite TEI" for his master format and then (learned and) used XSLT to handle the transform to ePub. Project Gutenberg was playing around with it for a little while, but the last revision to PGTEI was back in 2005 and otherwise there's no new news. On the main Project Gutenberg website, the File Formats FAQ has the following note under XML:

Originally Posted by Project Gutenberg

Some PG volunteers have looked at, and are looking at, ways to convert the entire archive using a satisfactory DTD; however, meantime we aren't actually producing much XML, since most volunteers aren't working with it, and nobody wants to start producing many XML texts until we have agreed standards.

Only real major difference (aside from terms and keywords) is TEI prefers ODD (one-document-does-it-all) approach whereas ePub has more similarities to DITA in generally preferring to break books into chapters to make things easier to handle and manage.

[Sweetpea]

Originally Posted by Kirtai

I currently use html though I also find it's not rigorous enough.
I hate that things like chapter headers are marked up merely by convention instead of properly and would really like to find a good format that does allow proper markup without being excessively complicated or rigid.

I wouldn't call stylesheets rigid... I make up my headers as I want them to and usually completely overrule the default mark-up.

[Kirtai]

Originally Posted by Sweetpea

I wouldn't call stylesheets rigid... I make up my headers as I want them to and usually completely overrule the default mark-up.

By rigid I was more referring to something like DocBook which has its own ideas about exactly how a document is constructed.

[tompe]

Originally Posted by cerement

Only real major difference (aside from terms and keywords) is TEI prefers ODD (one-document-does-it-all) approach whereas ePub has more similarities to DITA in generally preferring to break books into chapters to make things easier to handle and manage.

If I remember correctly TEI uses SGML so you can just define external entities for each chapter if you want.

[cerement]

Originally Posted by tompe

If I remember correctly TEI uses SGML so you can just define external entities for each chapter if you want.

TEI, like DITA, DocBook, and ePub uses XML (which is technically SGML) ...

Found this comparison yesterday, XML Models for Books (PDF) on O'Reilly.

The simplistic view looks like science and technology are sticking with DocBook or LaTeX. Humanities likes TEI. Publishers are starting to lean towards ePub.

[frabjous]

Originally Posted by cerement

TeX Based:

Mainly the LaTeX macros (and I love Donald Knuth's writing style and rigorous attention to detail). And while TeX can be output to almost anything, the conversion process between TeX and XML seems to still be pretty vague (although its ability to generate table-of-contents from full files would be very useful for the multi-file format of ePub).

No one has recommended this, and that's understandable. I don't disagree. However, since I use LaTeX all the time for my own projects unrelated to portable eBook reading (though I'm not an expert user by any means), I would be curious if anyone has much experience with working with TeX > eBook conversions.

TeX > PDF conversions are of course standard and easy, and with the right packages, I imagine it's pretty easy to make properly eBook-sized and formatted .PDFs.

But what about TeX > other eBook formats, like Mobi, ePub and LRF, etc.? Anyone have any experience with this, or know of any good tools/packages, etc., for it? How well do these support the features a TeX user typically expects, like decent looking mathematical formulae, footnotes, etc.?