icu4c api proposal: Unicode IDs in utrans_ C API

The Transliterator C++ API uses UnicodeString IDs, while the C API uses char * IDs. This works for 
true and pure identifiers that these APIs were designed for, for example "Cyrillic-Latin". It does 
not work when the ID contains filters ("[:Script=Cyrl:]") or even a complete set of rules because 
then the ID string contains more than just "invariant" characters (see utypes.h).

I propose to deprecate (but keep) the old APIs, document that they do not work for IDs with filters 
or rules, and add new functions with the necessary behavior:

U_CAPI UTransliterator* U_EXPORT2
utrans_openU(const UChar *id, // must not be NULL
              int32_t idLength, // -1 for NUL-terminated ID
              UTransDirection dir,
              const UChar *rules, // can be NULL
              int32_t rulesLength, // -1 for NUL-terminated rules
              UParseError *parseError, // can be NULL
              UErrorCode *pErrorCode);

(const char *id -> const UChar *id + idLength)

U_CAPI const UChar *U_EXPORT2 // will be NUL-terminated
utrans_getUnicodeID(const UTransliterator *trans,
                     int32_t *resultLength); // can be NULL

U_CAPI void U_EXPORT2
utrans_unregisterID(const UChar* id, int32_t idLength);


U_CAPI UEnumeration * U_EXPORT2
utrans_openIDs(UErrorCode *pErrorCode);

(To replace utrans_getAvailableID().)

Expiration: 2003-jul-07

Sincerely,
markus