Quote:
Originally Posted by Coops
Would the mobi format still allow hyphenation on the Oasis?
|
No. Or, rather, maybe
There are three readers involved here, using completely separate code and with totally divergent features. The appropriate reader is invoked automatically. In brief: the newer two support hyphenation. The older one doesn't.
1) The old MOBI reader, reading the Mobipocket format, some but by no means all *.mobis. This is still used for periodicals (since the periodical reader was never reimplemented for other formats) and for any ebooks you might have downloaded way back in the mists of time before AZW3 started becoming common: these days most books are dual-format, so any remotely newer reader will receive an AZW3-format file instead. It uses a different set of line spacings to everything else, though they are derived from the same configuration settings, which is deeply irritating because everything leaps closer together or further apart from what you're used to if you read one of these. Voice View doesn't work with this because apparently blind people don't need to read newspapers in Amazon's world. No hyphenation, and really bad line spacing as a result. Discretionary hyphens are not supported: the reader never breaks at hyphens at all, not even at em dashes unless they are surrounded by spaces (which is frankly rare).
2) The AZW3 reader, used for *.azw3 books and "dual-format" mobis such as those created by default by Calibre for non-periodicals; this format is essentially epub in wolf's clothing. Handles discretionary hyphens, so the Calibre Hyphenate This! plugin works with this reader. On older firmware (5.6.1.x and older, definitely: maybe some later firmware, but definitely not by 5.8.x) this too does not split on non-discretionary hyphens, but by the time of the Oasis firmware it was definitely splitting on those too. Because it supports discretionary hyphens you can hyphenate however you wish: I often run things through TeX::Hyphen with a bit of fearfully ugly Perl
3) The KFX reader, only generatable by Amazon from uploaded books according to a schedule laid down on sheets of carefully prepared human skin in the sixth circle of Hell, so almost impossible to test your newly-written book on. A much simpler format than azw3, it can't handle a lot of things AZW3 can (like complex tables), but what it can handle imposes much less CPU load on the Kindle, so it has CPU time to render automatically-generated hyphenation info sent down as part of the file. This is allegedly better than the hyphenation Hyphenate This! does, but having seen it hyphenate 'hyperspace' as 'hypers-pace', I don't believe it for a moment: it is fairly terrible and I would guess that about 2% of its hyphenation choices are hilariously wrong. Amazon need to fire whoever wrote this hyphenator and just use the Knuth/Liang TeX algorithm like everyone else. (As a side benefit, that algorithm is so cheap to run that it can easily hyphenate in realtime on the Kindle with zero battery life impact; it also works for lots of languages with minimal per-language state. But nooo, Amazon have to do it their own way, worse than everything else out there including things people designed when Jeff Bezos was still a teenager.)
You can tell if you're using the KFX reader at runtime because the nearly-useless multi-page page-flip stuff works, and because if there are hyphens you'll sometimes be ripped out of the reading flow by really bad ones like "teat-ime" (yes, I really saw that one; no lactating mother was involved). You can tell if you're using the old .mobi reader because of the jagged line endings due to the total absence of hyphenation and because your line spacing has suddenly changed. Otherwise, you're using azw3.