Re: libstdc++/6090: Problem with copy of std_valarray

[Gabriel Dos Reis <gdr@codesourcery.com>]

The following reply was made to PR libstdc++/6090; it has been noted by GNATS.

From: Gabriel Dos Reis <gdr@codesourcery.com>
To: Benjamin Kosnik <bkoz@redhat.com>
Cc: gdr@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Subject: Re: libstdc++/6090: Problem with copy of std_valarray
Date: 12 Apr 2002 01:53:10 +0200

 Benjamin Kosnik <bkoz@redhat.com> writes:
 
 | The following reply was made to PR libstdc++/6090; it has been noted by GNATS.
 | 
 | From: Benjamin Kosnik <bkoz@redhat.com>
 | To: Gabriel Dos Reis <gdr@codesourcery.com>
 | Cc: erwan.goasdoue@ipsis.fr, gcc-bugs@gcc.gnu.org, gdr@gcc.gnu.org,
 |         gcc-gnats@gcc.gnu.org
 | Subject: Re: libstdc++/6090: Problem with copy of std_valarray
 | Date: Wed, 10 Apr 2002 09:49:17 -0700 (PDT)
 | 
 |  > That is the mandated behaviour, in particular if the source and the
 |  > target don't have the same length then the behaviour is undefined.
 |  
 |  I think the problem is that the assignment and copy ctor do different 
 |  things. Why is that?
 
 (I'm not sure that is really a problem).
 
 The copy constructor is supposed to construct a new object out of an
 existing one, whereas the assignment operator is supposed to assign
 values to components of an existing object.  Those two operations are
 fundamentally semantically different.  Valarray<>'s design is more
 driven by speed than by security; the assignement operator is more
 memcpy()-like than std::vector<>::operator=()-like.  That was a design
 choice explained by the fact that valarray<>s are meant to be basic
 blocks with "basic" operations. 
 
 -- Gaby