LYNX-DEV fixed bug on bsdi

[Laura Eaves]

Well I didn't intend to spend Christmas debugging lynx, but
I have fixed the problem on bsdi.

The problem was that the value of a local array (replace_buf[] in
LYFullyTranslateString_1()) was being used before being initialized.
The code in this function is quite harry, and I'm not really sure what
it's supposed to do, but looking at the trace output I'm wondering of it's
really appropriate to call this function here in the first place.

I consider my fix (see below) to be a stopgap measure.
This code needs to be scanned with a fine toothed comb, if not rewritten.

please apply this patch to the latest ac.
(Fote may also want to pick it up, if he has this function i his code.)

Thanks and Merry Christmas.
--le

*** old/LYCharUtils.c   Tue Dec 23 15:02:42 1997
--- src/LYCharUtils.c   Thu Dec 25 14:52:42 1997
***************
*** 1062,1067 ****
--- 1062,1069 ----
      */
      if (!str || *str == NULL || **str == '\0')
          return str;
+     for ( i = 0; i < 64; ++i )
+       replace_buf[i] = '\0';
      /*
      **  Don't do byte translation
      **  if original AND target character sets
***************
*** 1143,1149 ****
                HTChunkPuts(CHUNK, s); \
                qs = q = *str
  
! #define REPLACE_CHAR(c) if (q > p) { \
                HTChunkPutb(CHUNK, qs, q-qs); \
                qs = q = *str; \
                *q++ = c; \
--- 1145,1152 ----
                HTChunkPuts(CHUNK, s); \
                qs = q = *str
  
! #define REPLACE_CHAR(c) \
!           if (q > p) { \
                HTChunkPutb(CHUNK, qs, q-qs); \
                qs = q = *str; \
                *q++ = c; \