[Haskell-cafe] Haskell AES lazy vs. strict bytestrings

tsuraan tsuraan at gmail.com
Mon Feb 14 00:59:59 CET 2011

By my reading of the documentation of the AES package in hackage
the sizes of the strict bytestring chunks of the lazy bytestring given
to crypt must be multiples of 16 bytes or the resulting output will
act as though the strict bytestrings were padded to 16 bytes with
zeroes.  I'm having trouble seeing this behaviour, so I'm wondering if
I'm mis-reading the documentation or if I'm not understanding what's
happening.  Here's a ghci run:

Prelude> import Codec.Crypto.AES
Prelude Codec.Crypto.AES> let lconcat = Data.ByteString.Lazy.concat
Prelude Codec.Crypto.AES> let lpack=Data.ByteString.Lazy.Char8.pack
Prelude Codec.Crypto.AES> let strict = Data.ByteString.concat $
Data.ByteString.Lazy.toChunks $ lconcat [lpack "hi", lpack "there",
lpack "everybody"]
Prelude Codec.Crypto.AES> let lazy = lconcat [lpack "hi", lpack
"there", lpack "everybody"]
Prelude Codec.Crypto.AES> lazy
Chunk "hi" (Chunk "there" (Chunk "everybody" Empty))
Prelude Codec.Crypto.AES> let pw = pack "0123456789abcdef"
Prelude Codec.Crypto.AES> let sencode = crypt' CTR pw pw Encrypt strict
Prelude Codec.Crypto.AES> let lencode = crypt CTR pw pw Encrypt lazy
Prelude Codec.Crypto.AES> Data.ByteString.concat $
Data.ByteString.Lazy.toChunks lencode
Prelude Codec.Crypto.AES> sencode

So, I'm not seeing the padding of each strict bytestring of the lazy
bytestring in the output encoding.  I'd say that's a good thing, but
it's not what I expect from the documentation, so I'd just like to
know if I'm doing something wrong before I go and depend on lazy
bytestrings working like this.

More information about the Haskell-Cafe mailing list