Thierry Bouche writes:
 >  No please no!  We want to know what rules and heuristics
 >  forward-looking authors, designers and editors think should be in this
 > The notion of `facing pages' is important (also for spread, i
 > suppose): I don't care whether a float goes after or before its call,
 > but i care whether it appears on the same double as the related
 > text... This is something i currently handle manually (and painfully).

well, as you learned in Toulouse the algorithm as of today doesn't offer the
functionality to force a float onto the same page as its call-out, but as i
wrote in the article i do expect to add "fuzzy-flushing" pretty soon.