https://bugs.kde.org/show_bug.cgi?id=451816
--- Comment #2 from tooli...@yahoo.com --- Hi, >From faceboot zstd Q&A: https://github.com/facebook/zstd/issues/1360 Comment is: using ZSTD_e_flush will force the compressor to emit a block immediately. In general, this requirement is for communication scenarios, when some data must be sent "now". Because it increases the nb of blocks produced, this technique can be associated with some lower compression ratio. Therefore, whenever there is no need to send data immediately, it's better (for compression ratio) to let the data accumulate in the internal buffers, and let the compressor decide when it's best to output a new block. As stated above: Using flush in a device is not a good idea. It result in a flush for each pieces of data that is send to the device using << operators, which breaks compression. That is why i suggest to use ZSTD_e_continue instead. I am using it regularily for my project and without the changes, the API is just not working at all (Not compression anything). Regards -- You are receiving this mail because: You are watching all bug changes.
