* Michael Gilbert: > Along with Raphael's suggestion, perhaps during updates we could > load the new dictionaries into memory concurrently with the old > ones. Then we could compare the two and only act on items that > actually have differences before pushing the new updates.
As long as things fit into memory, I think it's cheaper to recompute everything. The additional bookkeeping for dependency tracking will kill any performance advantage, and a manual incremental algorithm is just too error-prone, IMHO. -- To UNSUBSCRIBE, email to [email protected] with a subject of "unsubscribe". Trouble? Contact [email protected] Archive: http://lists.debian.org/[email protected]
