LYNX-DEV EBCDIC and OS/390

[pg]

As I make slow, intermittent progress with porting Lynx to OS/390,
a number of questions occur:

I'm digging out all code sensitive constructs and changing them,
#ifdef'ing where it seems appropriate.  For example, I'm
substituting CH_ESC wherever I find '\033' (or maybe just 27).

I've created a header file containing:

#ifndef   CH_ESC
#ifdef    EBCDIC
#define CH_DEL     '\x07'
#define CH_ESC     '\x27'
#define CH_ESC_PAR '\x27'
#define CH_HICTL   '\x3f'
#define CH_NBSP    '\x41'
#define CH_SHY     '\xca'
#else  /* EBCDIC */
#define CH_ESC     '\033'
#define CH_DEL     '\177'
#define CH_ESC_PAR '\233'
#define CH_HICTL   '\237'
#define CH_NBSP    '\240'
#define CH_SHY     '\255'
#endif /* EBCDIC */
#endif /* CH_ESC */

Questions:

o Are the identifiers I chose suitable?  I chose the "CH_"
  prefix because it's unique in a "grep" through the source code.
  
o Where should this header file go/what should it be called?  For
  now, I'm using WWW/Library/Implementation/LYCharVals.h.  Is this
  consistent with existing Lynx naming conventions?
  
o Would it be better to incorporate the definitions in an existing
  header file, perhaps WWW/Library/Implementation/tcp.h?
  
o Will the Lynx developer community accept such pervasive changes
  to the Lynx source code?
  
I've had to shuffle two character attribute tables, keymap and
EditBinding, from ASCII to EBCDIC order.

o I do this in executable code -- I believe this is more
  maintainable than static initializations.  Is there a
  better way?
  
o The shuffle code is in the initialization part of LYMain.c
  (with #ifdef's for EBCDIC only).  Is there a better place?
  
o Will I encounter other such tables that need to be shuffled?

o Are the recent character translation tables any use to me?
  I suspect not, since the underlying assumption is that
  whatever the external character set, Lynx itself runs
  in ASCII.

Thanks for your attention,
gil