Re: libstdc++/3017

[bkoz@gcc.gnu.org]

The following reply was made to PR libstdc++/3017; it has been noted by GNATS.

From: bkoz@gcc.gnu.org
To: bkoz@gcc.gnu.org, brendan@zen.org, gcc-gnats@gcc.gnu.org,
  nobody@gcc.gnu.org
Cc:  
Subject: Re: libstdc++/3017
Date: 4 Jun 2001 17:44:42 -0000

 Synopsis: locale_facets.h is missing some methods
 
 Responsible-Changed-From-To: unassigned->bkoz
 Responsible-Changed-By: bkoz
 Responsible-Changed-When: Mon Jun  4 10:44:41 2001
 Responsible-Changed-Why:
     Mine.
 State-Changed-From-To: open->feedback
 State-Changed-By: bkoz
 State-Changed-When: Mon Jun  4 10:44:41 2001
 State-Changed-Why:
     Thanks Brendan. Although your attachment came through as just a ChangeLog entry, I can see what you are getting at.
     
     I suspect that the declarations should go in, so that code like the following can compile:
     
     #include <locale>
     
     class gnu_obj {};
     
     class gnu_ctype: public std::ctype<gnu_obj> 
     { };
     
     void test01()
     {
       gnu_ctype obj;
     }
     
     
     Anyway, here's the deal:
     
     1) ctype<char> ctype<wchar_t> are the only required facets for the ctype class
     
     2) I postulate that a generic ctype cannot be done, and that link errors to indicate this are better than linking a bunk program. I realize that this is a quality of implementation issue and opinions may differ, so I'm interested in hearing yours.
     
     The other thing is that ctype, in general, needs to be re-worked in favor of a true locale-independent model. This is first up on the post gcc-3 TODO list.
     
     best,
     benjamin
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=3017&database=gcc