This repository has been archived by the owner on Jul 3, 2019. It is now read-only.
This repository has been archived by the owner on Jul 3, 2019. It is now read-only.
Open
Description
I think that there is an issue with readFloat16
, it uses big endian representation for the underlying bytes instead of little endian.
This is only used for font alignment zone data parsing (which Shumway does not seem to support).
The endianness is not specified, but:
- Float32 and Float64 are little-endian
- When testing font alignment parsing zone data parsing on multiple files, little-endian seems to produce better values (between 0.1 and 3) while big-endian returns odd values or even NaN.
Line 132 in 16451d8
(the
littleEndian
should be true
instead of false
)Metadata
Assignees
Labels
No labels