Anyone who has dealt with this stuff knows it's a geometric relationship. Two formats is 4X as much work as one. Three is 9X. Etc.
I don't understand the overhead of on-demand. As long as you have proper dependencies and
caching set up, you only run the generation exercise once, and then everyone gets a cached
copy until one of the dependencies changes.
It's less overhead than building every format whenever a master changes, where you build formats that may never get requested. If you have a common dependency, like a common css stylesheet, then do you currently trigger the rebuild of every project that uses it, all at once?
On Tue, Jan 31, 2012 at 2:26 PM, Marcello Perathoner
<marcello@perathoner.de> wrote:
On 01/31/2012 10:36 PM, Greg Newby wrote:
Having more than one master format per book does not make sense.
Decide which format is best for that book and stick to it. Every
typo should have exactly one location that needs fixing.
In principal I agree. In practice, we often have 2 (HTML + text).
I don't think it's is very burdensome to edit 2 files rather
than 1.
If we'd have multiple masters per book, a mantainer would need to learn all master formats, and in the end we'd probably lose sync between them.
--
Marcello Perathoner
webmaster@gutenberg.org