James Adcock wrote:
But we do host them...it's just that we're generating them on demand, which has numerous practical benefits for the producers, and is close to transparent for the human readers.
Three problems with the "on demand" approach [which in general I like]
1) It doesn't do the "spine" stuff right like TOC and covers.
It builds the TOC from your HTML headers. If the TOC doesn't come out right its because your HTML header structure is a mess.
2) If the "on demand" has problems with the generated results you ought to let people submit versions of those file formats that do work, and those submissions ought to override the "on demand" stuff until if and when the "on demand" version gets fixed.
That's another (binary) file for the WWers to maintain ... A better solution is to repost a HTML file that works. Convert to XHTML 1.1 and leave fancy formatting out.
3) It would be nice if you could provide a version of your conversion tools to PG "authors" so that they could work to make the submitted HTML work correctly with the conversion tools. Right now we "authors" are guestimating to the results of the "on demand" outputs by cross-compiling the HTML to EPUB and MOBI using one of a variety of non-PG conversion tools -- and then submitting the HTML to PG when that works -- which seems kind of silly given that we are actually targeting the unseen PG tools.
http://www.gutenberg.org/tools/epubmaker-0.02-preview-2009-11-26.tgz -- Marcello Perathoner webmaster@gutenberg.org