ICU4C proposal: Change some functions in icuio

I'd like to make the following changes to icuio API.  These changes are 
being proposed to make it easier for current stdio C users to transition 
to the newer implementation of the icuio library, and to better support 
the features of Unicode.

1) Reverse the order of arguments for u_fgets.  Right now, the opposite 
ordering of the arguments makes it annoying when porting an old 
application to using fgets() to start using u_fgets().  You have to make 
some unnecessary changes to use the ICU API.  Here is the new ordering:

u_fgets(UChar *s, int32_t n, UFILE *f);

For reference, here is the original ordering.
u_fgets(UFILE *f, int32_t n, UChar *s);


2) Make u_fungetc take a UChar32 parameter and return a UChar32 value. 
This makes it easier to put supplementary characters back into the file. A 
UChar can still be used with this API without a cast.  For reference, the 
ungetc() function from stdio takes an int parameter and not a char 
parameter.

u_fungetc(UChar32 c, UFILE *f);


3) Stop unescaping characters in u_fgetcx.  A Transliterator can be used 
to achieve the same results, and a Transliterator is better suited to get 
the right unescaping to be performed.  This function will remain as API to 
allow getting UChar32 characters, but it won't escape the characters 
anymore.


Background information:
The icuio library was scheduled to be supported in ICU 2.8.  Due to 
scheduling conflicts with more important bugs and requests that recently 
came up, the icuio library won't become supported in 2.8 as originally 
planned.  The icuio library will remain unsupported in 2.8.  These 
proposed changes are just a step towards supporting this library.  Some of 
the changes for 2.8 will improve the features, performance and stability 
of the icuio library.  The changes just aren't as grand as we had hoped 
for.

Deadline for comments: October 15, 2003

George Rhoten
IBM Globalization Center of Competency/ICU  San José, CA, USA