Data.ByteString candidate 3

Bulat Ziganshin bulat.ziganshin at
Wed Apr 26 05:40:13 EDT 2006

Hello Donald,

Wednesday, April 26, 2006, 12:35:16 PM, you wrote:

> I'm not sure we need 5 kinds of Foo-encoding layers, and I don't plan to
> write them.

let's count:

Latin1 - already written by you
UTF8 - requested by many people here, required to work with compiler's
input in ghc/jhc, and is the most compact representation for general
UCS4 - already implemented in Data.PackedString, fastest way to work
with general strings (i mean faster indexing and other direct-index
UTF16 - used in Windows API, so it's implementation will be really
useful to simplify this API implementation and to allow application
programs to work directly with such strings instead of converting them
from/to some other format

btw, what will be really useful now, imho, is the interface to
Text.Regex. how about working on it as next stage?

and one more suggestion - you can significantly speedup your code by
importing the 6.5's ForeignPtr implementation inside your library.
This type almost don't appears in ByteString external interface, so
this should be not so huge work

Best regards,
 Bulat                            mailto:Bulat.Ziganshin at

More information about the Libraries mailing list