[lib] Add ZSTD_c_deterministicRefPrefix

This flag forces zstd to always load the prefix in ext-dict mode, even
if it happens to be contiguous, to force determinism. It also applies to
dictionaries that are re-processed.

A determinism test case is also added, which fails without
`ZSTD_c_deterministicRefPrefix` and passes with it set.

Question: Should this be the default behavior? It isn't in this PR.
This commit is contained in:
Nick Terrell
2021-05-05 18:49:56 -07:00
parent 2d10544b84
commit 172b4b6ac4
7 changed files with 109 additions and 9 deletions
+22 -1
View File
@@ -449,7 +449,8 @@ typedef enum {
ZSTD_c_experimentalParam11=1008,
ZSTD_c_experimentalParam12=1009,
ZSTD_c_experimentalParam13=1010,
ZSTD_c_experimentalParam14=1011
ZSTD_c_experimentalParam14=1011,
ZSTD_c_experimentalParam15=1012
} ZSTD_cParameter;
typedef struct {
@@ -1859,6 +1860,26 @@ ZSTDLIB_API size_t ZSTD_CCtx_refPrefix_advanced(ZSTD_CCtx* cctx, const void* pre
*/
#define ZSTD_c_useRowMatchFinder ZSTD_c_experimentalParam14
/* ZSTD_c_deterministicRefPrefix
* Default is 0 == disabled. Set to 1 to enable.
*
* Zstd produces different results for prefix compression when the prefix is
* directly adjacent to the data about to be compressed vs. when it isn't.
* This is because zstd detects that the two buffers are contiguous and it can
* use a more efficient match finding algorithm. However, this produces different
* results than when the two buffers are non-contiguous. This flag forces zstd
* to always load the prefix in non-contiguous mode, even if it happens to be
* adjacent to the data, to guarantee determinism.
*
* If you really care about determinism when using a dictionary or prefix,
* like when doing delta compression, you should select this option. It comes
* at a speed penalty of about ~2.5% if the dictionary and data happened to be
* contiguous, and is free if they weren't contiguous. We don't expect that
* intentionally making the dictionary and data contiguous will be worth the
* cost to memcpy() the data.
*/
#define ZSTD_c_deterministicRefPrefix ZSTD_c_experimentalParam15
/*! ZSTD_CCtx_getParameter() :
* Get the requested compression parameter value, selected by enum ZSTD_cParameter,
* and store it into int* value.