Updated ISO-C++ patchset for Octave CVS

[John W. Eaton]

On  3-Feb-2001, Mumit Khan <address@hidden> wrote:

| [ I'm not on the Octave maintainers list, so please Cc me if appropriate ]
| 
| Updated patch to Octave CVS (2001-02-02) for ISO C++ conformance, as 
| well as a few other minor changes. Broken up into 6 easy pieces.
| 
| I've fixed all the regressions with complex numbers, and now only the
| failure using a conformant library is the expected one -- lack of 
| formatting that was available on libstdc++-v2/libio (scan and form
| and variants). I'll hopefully contribute an alternative implementation.
| 
| Checked on following systems:
|   
|   SPARC Solaris 2.8 -- GCC 2.95.2 and Sun Workshop 5.0 (fully patched)
|   Linux/GNU (RH 6.2) -- GCC 2.97, GCC 2.95.2 and EGCS 1.1.2.
| 
| Using Sun's licensed performance library and BLAS implementation has
| some interesting implications for the testsuite, and I'll send a 
| separate message on that later. Good old precision issue.
| 
| Generated files not included as usual.
| 
| I just realized that the patch Index for liboctave directory don't have
| liboctave/ in the pathnames, so apply with -p0 in that directory.

OK.  I included these patches except

  * I skipped these changes:

      2001-02-03  Mumit Khan  <address@hidden>

              * CColVector.cc, CMatrix.cc, CRowVector.cc, CmplxSCHUR.cc,
              lo-mappers.cc, lo-specfun.cc: Add std:: namespace qualifier as 
              needed.

    because I added a set of forwarding functions to
    liboctave/oct-cmplx.h.  For now, I think that's the better way to
    go, since it allows us to easily handle the case where we might
    want to define our own complex data type instead of using
    complex<double>.

  * I also skipped this change:

      2001-02-03  Mumit Khan  <address@hidden>

              * Makefile.in (DLD_XSRC): Remove debug.cc.

    That file should now be checked in to the CVS archive.  It is part
    of a patch by Ben Sapp to implement some basic interactive
    debugger features for the interpreter.

  * I used

      #if defined (CXX_ISO_COMPLIANT_LIBRARY)
        ...

    instead of

      #if CXX_ISO_COMPLIANT_LIBRARY
        ...

    because I believe it is possible for CXX_ISO_COMPLIANT_LIBRARY to
    be undefined, not just a true/false value.

These changes should be checked in the the CVS tree as soon as I
finish recompiling to make sure I haven't misapplied something.

Thanks,

jwe