@Cyan4973: should we deprecate ZSTD_initDStream as well?

ZSTD_initDStream() is part of the stable API,
I believe it's good to stay.

Converting back to draft status to fix lz4-threadpool-libs test

Ready for review. I decided to #define ZSTD_DISABLE_DEPRECATE_WARNINGS in zbuff.h to suppress the warning for ZBUFF_decompressInitDictionary().

I originally tried this:

 size_t ZBUFF_decompressInitDictionary(ZBUFF_DCtx* zbd, const void* dict, size_t dictSize)
 {
+    {   const size_t resetRet = ZSTD_DCtx_reset(zbd, ZSTD_reset_session_only);
+        if (ZSTD_isError(resetRet)) return resetRet;   }
+    {   const size_t loadRet = ZSTD_DCtx_loadDictionary(zbd, dict, dictSize);
+        if (ZSTD_isError(loadRet)) return loadRet;   }
+    return ZSTD_startingInputLength(zbd->format);
-    return ZSTD_initDStream_usingDict(zbd, dict, dictSize);
 }

The problem is that ZBUFF_DCtx is an opaque struct, so we can't access zbd->format. Rather than including non-public zstd headers to get the definition of the struct, I felt it's cleaner to just suppress the deprecation warning. I'm open to other solutions, though.

Ready for review. I decided to #define ZSTD_DISABLE_DEPRECATE_WARNINGS in zbuff.h to suppress the warning for ZBUFF_decompressInitDictionary().

zbuff itself is deprecated.
Do you need to #define ZSTD_DISABLE_DEPRECATE_WARNINGS in zbuff.h , or could it be defined in zbuff.c instead ?
This would avoid propagation of this constant.

Do you need to #define ZSTD_DISABLE_DEPRECATE_WARNINGS in zbuff.h , or could it be defined in zbuff.c instead ? This would avoid propagation of this constant.

@Cyan4973 Good point, moved it to zbuff_decompress.c.

Rebased. There seems to be an unrelated CI issue with the mingw tests, will hold off landing this until I figure that out.

Rebased. There seems to be an unrelated CI issue with the mingw tests, will hold off landing this until I figure that out.

The last PR to be merged was #3410. Could it be related ?

@cyan I confirmed that the CI failure is not related to the changes from this PR. (See #3420 for details). Since all other tests are passing, can I land?