[commit: base] master: Improve documentation for mkTextEncoding (f0ca95f)

[Max Bolingbroke]

Repository : ssh://darcs.haskell.org//srv/darcs/packages/base

On branch  : master

https://github.com/ghc/packages-base/commit/f0ca95f7628b37733e343a1d9ce96ca367fe8001

>---------------------------------------------------------------

commit f0ca95f7628b37733e343a1d9ce96ca367fe8001
Author: Max Bolingbroke <batterseapower at hotmail.com>
Date:   Tue Apr 23 19:15:40 2013 +0100

    Improve documentation for mkTextEncoding

>---------------------------------------------------------------

 GHC/IO/Encoding.hs |   28 ++++++++++++++++++++++++----
 1 files changed, 24 insertions(+), 4 deletions(-)

diff --git a/GHC/IO/Encoding.hs b/GHC/IO/Encoding.hs
index bd54182..052955c 100644
--- a/GHC/IO/Encoding.hs
+++ b/GHC/IO/Encoding.hs
@@ -175,8 +175,8 @@ char8 = Latin1.latin1
 --
 --  * @UTF-32@, @UTF-32BE@, @UTF-32LE@
 --
--- On systems using GNU iconv (e.g. Linux), there is additional
--- notation for specifying how illegal characters are handled:
+-- There is additional notation (borrowed from GNU iconv) for specifying
+-- how illegal characters are handled:
 --
 --  * a suffix of @\/\/IGNORE@, e.g. @UTF-8\/\/IGNORE@, will cause 
 --    all illegal sequences on input to be ignored, and on output
@@ -186,6 +186,28 @@ char8 = Latin1.latin1
 --  * a suffix of @\/\/TRANSLIT@ will choose a replacement character
 --    for illegal sequences or code points.
 --
+--  * a suffix of @\/\/ROUNDTRIP@ will use a PEP383-style escape mechanism
+--    to represent any invalid bytes in the input as Unicode codepoints (specifically,
+--    as lone surrogates, which are normally invalid in UTF-32).
+--    Upon output, these special codepoints are detected and turned back into the
+--    corresponding original byte.
+--
+--    In theory, this mechanism allows arbitrary data to be roundtripped via
+--    a 'String' with no loss of data. In practice, there are two limitations
+--    to be aware of:
+--
+--      1. This only stands a chance of working for an encoding which is an ASCII
+--         superset, as for security reasons we refuse to escape any bytes smaller
+--         than 128. Many encodings of interest are ASCII supersets (in particular,
+--         you can assume that the locale encoding is an ASCII superset) but many
+--         (such as UTF-16) are not.
+--
+--      2. If the underlying encoding is not itself roundtrippable, this mechanism
+--         can fail. Roundtrippable encodings are those which have an injective mapping
+--         into Unicode. Almost all encodings meet this criteria, but some do not. Notably,
+--         Shift-JIS (CP932) and Big5 contain several different encodings of the same
+--         Unicode codepoint.
+--
 -- On Windows, you can access supported code pages with the prefix
 -- @CP@; for example, @\"CP1250\"@.
 --
@@ -194,8 +216,6 @@ mkTextEncoding e = case mb_coding_failure_mode of
     Nothing -> unknownEncodingErr e
     Just cfm -> mkTextEncoding' cfm enc
   where
-    -- The only problem with actually documenting //IGNORE and //TRANSLIT as
-    -- supported suffixes is that they are not necessarily supported with non-GNU iconv
     (enc, suffix) = span (/= '/') e
     mb_coding_failure_mode = case suffix of
         ""            -> Just ErrorOnCodingFailure