[compression] NewZstdCompressor #9504
Conversation
| @@ -116,6 +116,35 @@ func (d *zstdDecompressor) Close() error { | |||
| return lastErr | |||
| } | |||
|
|
|||
| func NewZstdCompressor(w io.Writer, compressBuf []byte) io.Writer { | |||
There was a problem hiding this comment.
Is it necessary for the caller to pass the buffer slice? Can they pass a length instead?
If the answer is that it's needed then please add a doc string for the function and document how compressBuf should be used/sized by the caller.
There was a problem hiding this comment.
Not necessary (though other public functions accept a compressBuf).
Safe to create a variabled-sized buffer pool for the package?
There was a problem hiding this comment.
I think you're better off leaving the buffer allocation to the caller, as you're doing now. That way the user can decide if they want to use a pool, or allocate a new buffer, or reuse a buffer they already have.
But do leave a comment on the exported function.
| compressBuf []byte | ||
| } | ||
|
|
||
| func (z *zstdCompressor) Write(decompressedBytes []byte) (int, error) { |
There was a problem hiding this comment.
nit: uncompressedBytes is slightly nicer than decompressedBytes. The latter implies that they were originally compressed.
| @@ -116,6 +116,35 @@ func (d *zstdDecompressor) Close() error { | |||
| return lastErr | |||
| } | |||
|
|
|||
| func NewZstdCompressor(w io.Writer, compressBuf []byte) io.Writer { | |||
There was a problem hiding this comment.
I think you're better off leaving the buffer allocation to the caller, as you're doing now. That way the user can decide if they want to use a pool, or allocate a new buffer, or reuse a buffer they already have.
But do leave a comment on the exported function.
|
Closing in favor of #9521 |
This change introduce
NewZstdCompressor(...)(copied and adapted frompebble_cache.NewZstdCompressor(...).A follow-on change will wrap a
cachetools.UploadWriterwith this compressor when zstd compression is requested.