Just a followup on zip methods: (using zipinfo)
The main differences between the archive that Sigil creates and that WinRar creates (using default settings) is that Sigil treats most files as text (no extra field) and WinRar treats them all as binary (with an extra field). Both treat the font files as binary. Both use the Deflate method, but where Sigil uses maximum compression, WinRar uses normal compression (by default).
The WinRar archive shows a zip version of 3.1 on a "fat" system ... where the Sigil archive shows a zip version of 0.0 on an "ntf" system (using a Windows Vista platform: Sigil 0.7.2 and WinRar 4.20).
WinRar also stores the actual directories in the archive where Sigil does not.
Best Guess?? The Maximum Deflate compression on the font files in Sigil could possibly be the deal-breaker on these type of fonts (wouldn't know why at all). You could probably test that fairly easily by creating your WinRar zip file using the "maximum" deflate method and see if it fails your KDP test after the MOBI is built from it.
Haven't started picking apart the MOBI files built from the two different archives yet.
(Attaching the zip info output for each archive)
Last edited by DiapDealer; 07-26-2013 at 05:37 PM.
|