classic_table

[Krister Walfridsson <cato at df dot lth dot se>]

The test 22_locale/ctype/cons/char/1.cc fails on NetBSD due to differences
between ctype<char>::classic_table and the _M_table stuff in
ctype<char>::ctype.

While this is easily fixed (by the patch below), I'm not sure what is the
correct thing to do here.  The current NetBSD implementation returns
the table for the current locale, but looking at the FreeBSD code shows
that it just returns 0.  And the Linux code does a setlocale song-and-dance
to get a pointer to the "C" locale.

Am I correct to believe that it is the Linux code that does the correct
thing, so that I should modify the NetBSD code to return the table for
the C locale?

   /Krister



Index: netbsd/ctype_noninline.h
===================================================================
RCS file: /cvs/gcc/gcc/libstdc++-v3/config/os/bsd/netbsd/ctype_noninline.h,v
retrieving revision 1.1
diff -c -3 -p -r1.1 ctype_noninline.h
*** netbsd/ctype_noninline.h	24 Jun 2002 05:48:58 -0000	1.1
--- netbsd/ctype_noninline.h	11 May 2003 23:11:53 -0000
***************
*** 1,6 ****
  // Locale support -*- C++ -*-

! // Copyright (C) 2000, 2001, 2002 Free Software Foundation, Inc.
  //
  // This file is part of the GNU ISO C++ Library.  This library is free
  // software; you can redistribute it and/or modify it under the
--- 1,6 ----
  // Locale support -*- C++ -*-

! // Copyright (C) 2000, 2001, 2002, 2003 Free Software Foundation, Inc.
  //
  // This file is part of the GNU ISO C++ Library.  This library is free
  // software; you can redistribute it and/or modify it under the
***************
*** 35,41 ****

    const ctype_base::mask*
    ctype<char>::classic_table() throw()
!   { return 0; }

    ctype<char>::ctype(__c_locale, const mask* __table, bool __del,
  		     size_t __refs)
--- 35,41 ----

    const ctype_base::mask*
    ctype<char>::classic_table() throw()
!   { return _ctype_ + 1; }

    ctype<char>::ctype(__c_locale, const mask* __table, bool __del,
  		     size_t __refs)