On Sun, 2009-08-23 at 09:22 -0700, Judah Jacobson wrote:
I proposal that we augment ghc-6.12.1's support for Unicode Handles by adding the following functions to System.IO:
hSetOnEncodingError :: Handle -> OnEncodingError -> IO () hGetOnEncodingError :: Handle -> IO OnEncodingError
I agree that it is important.
Note that the text package, for example, provides more sophisticated error-handling options. However, I think the above choices are useful enough without making the API too complicated.
Personally I would prefer we postpone this decision with the aim that we unify the Text encoding / decoding between the IO system and text package. Specifically I suggest we put this decision off until after ICFP where we hope that the authors of ghc's new text IO system and the authors of the text package can get together with other interested individuals to discuss some more unified system. It should be possible to make an encoder abstraction that can be used purely in the text package for conversion between ByteString <-> Text, and also used in the IO system for text mode Handles. How to handle encoding and translation errors would have to be part of the design of that encoder abstraction. My impression is that an ST version of the current text encoder abstraction in the GHC text IO system would also be usable for pure conversions in the text package. Duncan