Skip to main content

Timeline for answer to Huffman Compression by user555045

Current License: CC BY-SA 4.0

Post Revisions

6 events
when toggle format what by license comment
4 hours ago history edited user555045 CC BY-SA 4.0
added 2228 characters in body
5 hours ago comment added user555045 @LokiAstari the serialized tree has at least 2 bytes per symbol (counting only the leaves for convenience), you can reduce that to less than a byte per symbol (plus some overhead which I'm ignoring for convenience but it doesn't ruin the comparison). DEFLATE has both run-length encoding for its table of code lengths (which already are only numbers in 0..15 so they would fit 2-per-byte without serious compression) and another layer of Huffman coding on top
7 hours ago comment added Loki Astari Endianess is a problem I will noodle on that.
7 hours ago comment added Loki Astari Changed std::size_t to std::uint64_t so we know the size of the object. Fixed the 1UL Changed to std::uint64_t{1}
8 hours ago comment added Loki Astari Serializing the tree seems to provide a very compact representation. Also interested in decode efficiency.
9 hours ago history answered user555045 CC BY-SA 4.0