v0.4.1

What's Changed

• Fix decoding of truncated inputs (#10)
• add decode_len and decode_len_unsafe and update dependencies by @JojiiOfficial in #5
• Fix pdep_speed return type for non x86 arch by @MarshalX in #7

New Contributors

• @JojiiOfficial made their first contribution in #5
• @MarshalX made their first contribution in #7

Full Changelog: v0.4.0...v0.4.1

v0.4.0

• Add #[no_std] support by @davnavr in #3
• Tuning for Zen 3
• New, faster scalar (SWAR) code for decoding single varints
• Faster scalar code path for encoding smaller varints

v0.3.0

• "Unsafe" interface now accepts pointers instead of slice references
• Optimizations for platforms without support for fast PDEP/PEXT
• New functions to decode two, four, and eight adjacent varints at once
• Testing and benchmarking improvements

v0.2.0

• New native-optimizations feature - enables PDEP/PEXT optimizations when compiling on some CPUs (Intel Haswell or newer, AMD Excavator and Zen 3) for nearly double throughput
• Slightly improved performance on other CPUs too
• Should now compile on x86 (still untested)