0.51.0

What's Changed

• perf: preload subcache headers in FullDyldCache to avoid repeated header reads by @p-x9 in #285
• Improve dyld cache URL construction performance by @p-x9 in #286
• Add benchmark infrastructure for performance work by @p-x9 in #287
• Reuse dyld cache file handles when enumerating Mach-O files by @p-x9 in #288
• Optimize classic rebase and bind segment lookup by @p-x9 in #289
• Cache dyld cache mapping info sequences by @p-x9 in #291
• Optimize closest symbol lookup by @p-x9 in #290
• Avoid eager main cache creation for subcaches by @p-x9 in #292
• Share dyld cache rebase resolution logic by @p-x9 in #293
• Cache dyld mapping info arrays by @p-x9 in #294
• Avoid segment array allocation in address lookups by @p-x9 in #295

Performance

This release includes broad performance improvements for Mach-O and dyld shared
cache analysis. Compared with 0.50.0, representative benchmarks show:

Area	Benchmark	p50 wall-clock improvement	p50 instruction reduction
Full dyld cache mapping	FullDyldCache.mappingInfo.lookup	99.7% faster	99.7% fewer
Full dyld cache mapping	FullDyldCache.fileOffset.translate	99.5% faster	99.7% fewer
Full dyld cache lookup	FullDyldCache.cache.forURL	95.2% faster	-
Full dyld cache lookup	FullDyldCache.cache.forOffset	93.0% faster	-
Full dyld cache rebase	FullDyldCache.resolveRebase	95.0% faster	95.3% fewer
Full dyld cache rebase	FullDyldCache.resolveOptionalRebase	94.9% faster	-
Full dyld cache image enumeration	FullDyldCache.machOFiles.enumerate	93.9% faster	95.1% fewer
Dyld cache mapping	DyldCache.mappingInfo.lookup	93.6% faster	96.0% fewer
Dyld cache mapping	DyldCache.address.translate	92.9% faster	-
Dyld cache mapping	DyldCache.fileOffset.translate	92.2% faster	-
Dyld cache rebase	DyldCache.resolveRebase	87.4% faster	88.8% fewer
Dyld cache rebase	DyldCache.resolveOptionalRebase	85.2% faster	-
Dyld cache image enumeration	DyldCache.machOFiles.enumerate	92.0% faster	93.0% fewer
Mach-O symbol lookup	MachOFile.closestSymbol.repeated	91.6% faster	89.7% fewer
Mach-O address lookup	MachOFile.contains.unslidAddress	76.2% faster	79.3% fewer
Mach-O address lookup	MachOFile.fileOffset.translate	28.0% faster	30.8% fewer

Full Changelog: 0.50.0...0.51.0

MachOKit
c4139f563b56b9e6fd1bfdb9d0d85ae39b795c296ce24fbedc07536df4c90ff1

MachOKitC
8cdeb024ce530f96b40466957b0eb0839f52e0c139ae67fcfbfb75d091fceeb4