v1.1

cain v1.1

Performance update

After profiling extensively¹ and using cain at a larger scale for a while (see translatepy@next for more information), this is an update bringing better performance in decoding using the cain data format. (this was enough to be able to see a 1.5~2x performance improvements during decoding)

New

• Added an Enum datatype

Updates

• Adding examples in datatypes docstring
• Using Datatype._cain_value instead of Datatype. to avoid conflicting attributes
• Using Datatype.__root__ instead of Datatype.origin to avoid conflicting attributes
• Speeding up annotations parsing (introducing DatatypeMeta.__type_hints__)
• Using the current Object instead of the built-in dict for the decoding result
• Other minor performance improvements

Everything is detailed in the README.md file

A comparison from the previous release can be found here v1...v1.1

A comparison from the current branch can be found here v1.1...main

🍡 Animenosekai

1. Profiled using tools: cProfile, pyinstrument and line_profiler ↩