search

vincenthz / hs-cryptohash

efficient and practical cryptohashing in haskell. DEPRECATED in favor of cryptonite
Other
30 stars 24 forks source link

Consider using hashabler to support hashing arbitrary types #40

Closed jberryman closed 8 years ago

jberryman commented 8 years ago

Would you be interested in merging changes that permitted hashing of arbitrary Hashable types from my hashabler package? The resulting hash functions would probably look like:

hash :: Hashable a=> a -> ByteString

where a may also be bytestring. This would allow hashing of arbitrary haskell values in a performant way (without needing to marshal to a bytestring first), and cross-platform compatible way, etc.

I'm not sure when I'd be able to get to this, but I wanted to float the idea first.

vincenthz commented 8 years ago

Nothing is expected to change in cryptohash, as it is a maintenance only and all further development are in cryptonite.

One thing is that Hashabler seems terribly misnamed, it's a fold-over-variable-length-atoms class (which is an interesting idea). It also reimplement much part of memory which got hashes and a more efficient fold-over-memory-chunks class (ByteArrayAccess). I've also become completely allergic to dependencies in Haskell as they add a unpredictable cost to development on things

jberryman commented 8 years ago

Thanks for the pointer to the memory package. On Feb 7, 2016 4:35 AM, "Vincent Hanquez" notifications@github.com wrote:

Nothing is expected to change in cryptohash, as it is a maintenance only and all further development are in cryptonite.

One thing is that Hashabler seems terribly misnamed, it's a fold-over-variable-length-atoms class (which is an interesting idea). It also reimplement much part of memory which got hashes and a more efficient fold-over-memory-chunks class (ByteArrayAccess). I've also become completely allergic to dependencies in Haskell as they add a unpredictable cost to development on things.

— Reply to this email directly or view it on GitHub https://github.com/vincenthz/hs-cryptohash/issues/40#issuecomment-180983897 .

jberryman commented 8 years ago

...although looking at the dates it looks like it's memory that reimplemented much of hashabler :-P

vincenthz commented 8 years ago

looks can be deceiving; The repository doesn't represent the actual history, most of memory lived in cryptonite (beginning 2015), and before that in individual packages like byteable (2013) and securemem (2013), siphash (2012) and some other apps.