Fetching latest commit…
Cannot retrieve the latest commit at this time.
|Failed to load latest commit information.|
HaShrink is a fun and innovative new way of distributing files using very, very small sizes. See a demo at http://e1ven.github.io/HaShrink/ WARNING- It may take a while to reproduce the file after transmission. You may want to get some tea while you wait. A lot of tea. Using this unoptimized code, an cc2.8xlarge EC2 instance can handle 25 permutations per second. This means that recreating the file will take 2^((filesize in bits))/25 seconds per file. For instance, this small README file will take over 1 centillion days. Reproducing the include 1 second sinewave will take 9.89 x 10^104544 days. Unfortunately, there may be some difficulty in retrieving files created after 10^100 years, as we may reach the heat death of the universe. The code works by creating every possible file of a given size. It checks each of them, for SHA collisions with the original file. If the SHA is a match (as filesize/hashsize of them will be) it then uses additional metadata that we have. We know it's supposed to be a valid MP3 file. This allows to further narrow down the candidates to files that both a) Have the same SHA and b) are Valid MP3s This could be further expanded by a third pass using http://echoprint.me/ That would further cull the number of candidates to those which were acoustically similar. At this point, the pool of candidates is going to be rather small. If necessary, we can review the MP3s, and find the one which sounds correct. Alternatively, before hand, we may be able to find the one which is a bit-for-bit copy, specify by filename which of these candidates is correct. People reproducing the file then know to delete copies with IDs < that file.