[OpenJade-devel] Build problems with OpenJade 1.3.2 on AIX 4, 5

[Albert C. <ope...@ml...>]

openjade 1.3.2 fails on AIX 4.3.2 and 5.1 with:
  ...
  In file included from GroveApp.cxx:9:
  GroveApp.h: In function `OpenSP::OutputCharStream& 
     OpenSP::operator<<(OpenSP::OutputCharStream&, OpenJade_Grove::GroveString&)
     ':
  GroveApp.h:52: error: no matching function for call to `
     OpenSP::OutputCharStream::write(const GroveChar*, size_t)'
  /opt/TWWfsw/opensp15/include/OutputCharStream.h:26: error: candidates are: 
     OpenSP::OutputCharStream& OpenSP::OutputCharStream::write(const Char*,
     long unsigned int)

On AIX 4.3 and 5.1, <sys/types.h> has:
  typedef unsigned short wchar_t;

So, from OpenSP's <config.h>, we have:
  #if (SIZEOF_WCHAR_T == SIZEOF_UNSIGNED_SHORT)
  #define SP_WCHAR_T_USHORT
  #endif

class GroveString has:
  ...
  public:
    typedef const GroveChar *const_iterator;
    GroveString() : data_(0), size_(0) { }
    GroveString(const GroveChar *data, size_t size)
      : data_(data), size_(size) { }
    size_t size() const { return size_; }
    const GroveChar *data() const { return data_; }
    ...

GroveChar, because of SP_WCHAR_T_USHORT, is defined to wchar_t because
of the following code in grove/Node.h:
  #ifdef SP_MULTI_BYTE
  #ifdef SP_WCHAR_T_USHORT
  typedef wchar_t GroveChar;
  #else
  typedef unsigned int GroveChar;
  #endif
  #else /* not SP_MULTI_BYTE */
  typedef unsigned char GroveChar;
  #endif /* not SP_MULTI_BYTE */

spgrove/GroveApp.h has:
  inline
  OutputCharStream &operator<<(OutputCharStream &os, 
  #ifdef GROVE_NAMESPACE
                               GROVE_NAMESPACE::
  #endif
                               GroveString &str)
  {
    return os.write(str.data(), str.size());
  }

OpenSP::OutputCharStream::write has only one prototype:
  OutputCharStream &write(const Char *, size_t);

If GroveChar was defined as:
  typedef unsigned int GroveChar;
then everything would be OK. This is because:
  ...
  typedef unsigned int Unsigned32;
  ...
  typedef Unsigned32 Char;

Any ideas on a fix? Should I add a --disable-multi-byte option to
disable SP_MULTI_BYTE?

-- 
albert chin (ch...@th...)