Hello Ric,
I've finally tracked down all the problems with epub test suite embedded font tests.
The obfuscated font tests were indeed failing due to a bug in the tests themselves. After tearing some hair out during exhaustive JavaScript debugging sessions (Chrome doesn't report in any way that there's a problem with parsing a font's binary format) where everything seemed to work OK, except for the actual font face selection and rendering, I've narrowed this down to the fact that fonts are actually being deobfuscated by ReadiumJS, but their data is garbled both before and after deobfuscation. This had led me to suspect that they were incorrectly obfuscated.
I've modified my font mangling Ruby script (attached) to perform obfuscation/deobfuscation manually and it turned out that fonts in the test EPUB are indeed incorrectly obfuscated - most likely a different publication ID has been used for obfuscation key than is currently specified in package.opf.
I'm attaching a corrected EPUB file where fonts are obfuscated with a correct key based on current publication ID.
The last test failure, fonts-070 with unicode ranges, was indeed a readium-js bug - I've fixed it in this pull request:
BTW, about the nearest Readium JS engineering meeting on Tuesday, 18th of February.
Due to family reasons I'll be absent from the this time. The status from my side is that I've finished EPUB test suite font bugs and am starting work on epub-fetch refactoring as planned. If you have any questions or issues to discuss, feel free to e-mail me.
Best Regards,