On Tue, Jan 31, 2012 at 11:32:51AM -0700, Lee Passey wrote:
On Tue, January 24, 2012 3:08 pm, Joshua Hutchinson wrote: ...
II. from those master files, various other file formats can be [and are, currently] derived automatically.
Mister Hutchinson's vision, which I am trying to follow, is that /all/ other file formats will be derived automatically from the /one/ master version. Caching is certainly advisable, but on-demand creation would be the first-step.
Many challenges are technical, such as increased sophistication in dealing with text and HTML as master formats.
The primary technical challenge is in developing a tool chain which can produce quality instances of all derived formats, and in adopting/developing a master format with the richness necessary to support that tool chain.
I put huge backing into addressing this challenge, and so did other people on this list and elsewhere. The answer was: TEI. Then, a couple of years later, ditto. The answer was: RST. There are scarce few things that TEI or RST are not suitable for, though I would not say we have a ton of experience. Today, I count fewer than 400 files that are TEI or RST (I didn't check how many separate eBook titles those files are associated with). The WWers have procedures to process such files - bring it on. Remaining problems include: - how to convince contributors to make new submissions in these master formats - consideration of alternate master formats, as desired - providing advice on better workflows and tool sets (including at DP) for these master formats, so contributors can be comfortable with them Solved problems include: - automatically generating all derived formats, with a far higher level of integrity than other master formats - applying fixes to the master and then regenerating derived formats - having an easily editable master format Looking at weaknesses in these choices, from all possible angles, is certainly worthwhile. As is seeking improvements or alternatives. But the unfortunate fact is that the "better mousetrap" (or, at least, one that purports to have many of the improvements over HTML+text as master formats), was built and delivered years ago. But not too many folks have taken the bait. I can already hear a few voices calling (or writing), "that's because your bait sucks!" and "but your bait cannot do X" and so forth. It would be contrary to past experience for anyone to think they can, indeed, come up with a solution set that will be above criticism. -- Greg