Jim Adcock wrote:
In MOBI land -- read Kindle and kindlegen (which is pretty much how everyone including PG is force to make MOBI files) there are two big problems:
1) Top and bottom margins are NOT merged. When a paragraph follows a paragraph then the vertical whitespace between those two paragraphs is added to each other.
2) Top and bottom margins are ROUNDED to the closest 1em.
[snip]
How then does one work around these problems? Once one recognizes that there really is a problem, then three (partial) work-around solutions come to mind:
1) Do not even specify paragraph formatting, but rather allow the built-in paragraph formatting in each HTML, EPUB, or MOBI device to do its job.
2) Specify (say) a 1.0em top margin and a 0.0em bottom margin.
3) *Almost* "Split the Baby" by specifying a 0.51 top margin and a 0.49 bottom margin.
There is a fourth way: pre-process your (X)HTML to downgrade it to an HTML 3.2 tag soup + Kindle attributes so that the kindlegen step does nothing other than wrapping up your HTML into a MOBI file. Granted, this won't help you pushing a nice MOBI through PG, but let's see how it might work for your own delivery. Let's say you've marked up your book at XHTML with CSS, and that you're pretty much using this as your delivered EPUB, as well as for browser consumption. You find that kindlegen does a half-hearted job of converting your styles into kindle-specific paragraph attributes. You've used mobiunpack to examine just how poor the situation is. You know from reading Joshua Tallent's book that using width and height attributes on paragraphs, and liberal sprinklings of NBSP will help you lay out your poetry in a way that works for the different font sizes that a user might pick. So, use xsltproc or Perl plus one of the XML modules to do what kindlegen does, except applying some of your own conversions. You can use classes in your XHTML source to target the elements that need some conversion help, which will help you apply funky conversions to just poetry or just footnotes or just the first paragraph after a heading, without needing to understand XPATH. Any non-trivial book is already going to involve some divergence between your "ideal" source and the one you chuck through kindlegen for it to up-chuck (er, I mean convert), so perhaps making kindlegen do *less* for you is the solution. Two concrete examples from the books I'm marking up at the moment: 1. The book uses unspaced em dashes, and I like them, so I'm marking them up as U+200B U+2014 U+200B (i.e. zero-width space, em dash, zero-width space), which allows lines to wrap either side of the em dash, exactly as the original book does. That's helpful because the original book is 150 years old as uses those long chapter headings that TEI calls "arguments". However, the Kindle doesn't understand ZWSP, but will do the right thing if I change them to ZWNJ, which is the wrong character but it works. 2. I have some genealogical data in the back which is a set of increasingly indented paragraphs, some of which are numbered. It looks rather like ordered lists except that the first son and first daughter both get numbered "1", and some other oddities, so I've just used paragraphs. The Kindle destroys the indentation, so again I have to pre-process the paragraphs in the fashion that Tallent demonstrates. OK, so number (2) is rather specific to the needs of a single book, but the first conversion has to be done a lot as I'm working through the book, so it's scripted, along with any other help that kindlegen needs. Kindlegen does a simple job badly, so bypass the bits that you don't need.