Final tests and reporting are done. As expected there isn't much to be gained by jumping the chain. In most of my tests I did see a moderate performance gain when invoking LZ4_compress_generic() directly with normal text. This could very easily be an edge case. Either way it's interesting and worth sharing.

1 files changed, 7 insertions, 0 deletions

diff --git a/lib/lz4.c b/lib/lz4.c
index 08cf6b5..df346cf 100644
--- a/lib/lz4.c
+++ b/lib/lz4.c
@@ -1500,6 +1500,13 @@ char* LZ4_slideInputBuffer (void* LZ4_Data)
     return (char*)(ctx->bufferStart + dictSize);
 }
 
+/*
+ * Testing so we don't have to hack prototypes or the static inline nature of LZ4_compress_generic().
+ */
+int LZ4_compress_generic_wrapper(void* state, const char* source, char* dest, int inputSize, int acceleration) {
+  return LZ4_compress_generic(state, source, dest, inputSize, 0, notLimited, byU16, noDict, noDictIssue, acceleration);
+}
+
 /* Obsolete streaming decompression functions */
 
 int LZ4_decompress_safe_withPrefix64k(const char* source, char* dest, int compressedSize, int maxOutputSize)