Closed GoogleCodeExporter closed 9 years ago
I feel not, at least not in a generic way.
After data is fed, internal states are scrambled in a way to make them look
random. Going back to previous value is almost impossible.
Now there is indeed a possibility for xxHash in particular, but it is limited.
Due to the way the intermediate buffer works, it would be possible to cancel
the last N-byte, as long it doesn't cross a "multiple of 16" value.
So, for example, if 33 bytes were fed into xxHash, it would be possible to
cancel the last one (1) byte, but not 2.
More importantly, it's also not possible to remove bytes at the beginning of
the sequence. With regards to the application you're looking for, i believe
there are much better (and faster) ressources over Internet to code a rolling
hash. Rolling hashes have mathematic properties specifically dedicated to
erasing bytes at the beginning of the sequence.
Original comment by yann.col...@gmail.com
on 25 Mar 2013 at 4:39
Thanks for the quick response.
Could you perhaps propose some keywords related to this kind of algorithm?
Searching for "running hash [algorithm]" or "windowed hash [algorithm]"
does not yield any useful results.
Original comment by fha...@gmail.com
on 25 Mar 2013 at 5:39
Original comment by yann.col...@gmail.com
on 25 Mar 2013 at 5:41
Original issue reported on code.google.com by
fha...@gmail.com
on 25 Mar 2013 at 4:22