05-25-2020, 10:28 PM
Yes, that seems right. It would have to decompress the whole thing first, enough to analyze the statistics and then cache them. so I assume there would be some duplicated work, just like dictionary cache building without compression.
~