3 ms·

Predeclared aliased pointers, STATEM/HSTATEM constants, len/Len, that kind of stuff.

fantastisch 6y ago
Thank you, Austin! BTW, not sure exactly what you mean by cross-half mixing, but I'm assuming it means mixing both halves of the 128 bit state with each other. There is that, on 3 lines 91[0], 100 and 108. The 128 bit mix, stirs two adjacent blocks of 64 bits. Those 3 lines can mix blocks 1 & 2. Subsequent mixes on either half then propagate that cross mix. So it does use the full 256 bit state. Is that what you meant? [0]: https://github.com/cris691/discohash/blob/master/src/discohash.cpp#L91 https://github.com/cris691/discohash/blob/master/src/discoha...
- antirez 6y ago
  Looks like a solid hash function indeed, with a very simple round. Q: Is there any reason you don't use the same mixing function for the residue of the string (the final tail that is not a full 8 bytes block)? You could copy the bytes you have into a 8 byte array that is pre-padded with some pattern, and call the same mixing function again. Is this in order to save the distribution / avalanche properties in case the string is very short? In general would be nice to read some design note, if P and Q were obtained experimentally by checking for distribution or alike, if a different rotation length changes significantly the distribution properties and so forth.
  - fantastisch 6y ago
    Great question, and thank you, Salvatore! Some reasons for that are: different hashes for s and s\0 (without explicitly using length) to make it harder to create collisions, related to as you say better distribution/avalanche for short strings, also design-wise I like the asymmetry that strings not perfectly divisible by 8 will be treated a little differently. I just think that makes it better overall.
    - antirez 6y ago
      Thanks for the replies!
      - fantastisch 6y ago
        No probs. I think your idea of unifying the two loops is a good one. A new version that uses the idea and overcomes the challenges would be great, and would simplify the code even more. If you had a way to make a start on it, I welcome your PR.

Thank you, Austin! BTW, not sure exactly what you mean by cross-half mixing, but I'm assuming it means mixing both halves of the 128 bit state with each other. There is that, on 3 lines 91[0], 100 and 108. The 128 bit mix, stirs two adjacent blocks of 64 bits. Those 3 lines can mix blocks 1 & 2. Subsequent mixes on either half then propagate that cross mix. So it does use the full 256 bit state. Is that what you meant? [0]: https://github.com/cris691/discohash/blob/master/src/discohash.cpp#L91 https://github.com/cris691/discohash/blob/master/src/discoha...

antirez 6y ago
Looks like a solid hash function indeed, with a very simple round. Q: Is there any reason you don't use the same mixing function for the residue of the string (the final tail that is not a full 8 bytes block)? You could copy the bytes you have into a 8 byte array that is pre-padded with some pattern, and call the same mixing function again. Is this in order to save the distribution / avalanche properties in case the string is very short? In general would be nice to read some design note, if P and Q were obtained experimentally by checking for distribution or alike, if a different rotation length changes significantly the distribution properties and so forth.
- fantastisch 6y ago
  Great question, and thank you, Salvatore! Some reasons for that are: different hashes for s and s\0 (without explicitly using length) to make it harder to create collisions, related to as you say better distribution/avalanche for short strings, also design-wise I like the asymmetry that strings not perfectly divisible by 8 will be treated a little differently. I just think that makes it better overall.
  - antirez 6y ago
    Thanks for the replies!
    - fantastisch 6y ago
      No probs. I think your idea of unifying the two loops is a good one. A new version that uses the idea and overcomes the challenges would be great, and would simplify the code even more. If you had a way to make a start on it, I welcome your PR.