Re: libstdc++/9234: [3.3/3.4 regression] unary minus for valarrays broken

Re: libstdc++/9234: [3.3/3.4 regression] unary minus for valarrays broken

[Volker Reichelt]

The following reply was made to PR libstdc++/9234; it has been noted by GNATS.

From: Volker Reichelt <reichelt@igpm.rwth-aachen.de>
To: paolo@gcc.gnu.org
Cc: gcc-bugs@gcc.gnu.org, gdr@integrable-solutions.net, nobody@gcc.gnu.org,
        gcc-gnats@gcc.gnu.org
Subject: Re: libstdc++/9234: [3.3/3.4 regression] unary minus for valarrays broken
Date: Sun, 2 Feb 2003 22:08:44 +0100

 Hi Paolo,
 
 On  2 Feb, paolo@gcc.gnu.org wrote:
 
 > Volker, Gaby approved your patch (see PR 9235). Could you
 > please commit it to both mainline and 3_3?
 
 I'm going to this tomorrrow morning. I wanted to do this earlier,
 but had trouble with the testsuite (I got a new computer and had
 some configury to do).
 
 In addition, I found another failure in valarray: operator!=() is
 broken, too, because of the following cut-n-paste error:
 
   struct __not_equal_to
   {
     template<typename _Tp>
       bool operator()(const _Tp& __x, const _Tp& __y) const
       { return __x == __y; }
                   ^^^^
   };
 
 I'll commit that (as obvious) and a testcase, too.
 
 > FYI, now in mainline the relevant code belongs to valarray_before.h,
 > ~ line 462.
 
 I already noticed, but thanks for the hint, anyway.
 
 Regards,
 Volker
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9234
 
 

Re: libstdc++/9234: [3.3/3.4 regression] unary minus for valarrays broken

[reichelt]

Synopsis: [3.3/3.4 regression] unary minus for valarrays broken

State-Changed-From-To: analyzed->closed
State-Changed-By: reichelt
State-Changed-When: Mon Feb  3 10:56:42 2003
State-Changed-Why:
    Fixed.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9234

Re: libstdc++/9234: [3.3/3.4 regression] unary minus for valarrays broken

[paolo]

Synopsis: [3.3/3.4 regression] unary minus for valarrays broken

Responsible-Changed-From-To: unassigned->reichelt
Responsible-Changed-By: paolo
Responsible-Changed-When: Sun Feb  2 09:35:17 2003
Responsible-Changed-Why:
    Patch ready.
State-Changed-From-To: open->analyzed
State-Changed-By: paolo
State-Changed-When: Sun Feb  2 09:35:17 2003
State-Changed-Why:
    Volker, Gaby approved your patch (see PR 9235). Could you
    please commit it to both mainline and 3_3? FYI, now in
    mainline the relevant code belongs to valarray_before.h,
    ~ line 462.
    Thanks, Paolo.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=9234

libstdc++/9234: [3.3/3.4 regression] unary minus for valarrays broken

[reichelt]

>Number:         9234
>Category:       libstdc++
>Synopsis:       [3.3/3.4 regression] unary minus for valarrays broken
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          wrong-code
>Submitter-Id:   net
>Arrival-Date:   Wed Jan 08 14:56:02 PST 2003
>Closed-Date:
>Last-Modified:
>Originator:     Volker Reichelt
>Release:        3.3-20021230, 3.4-20021230
>Organization:
>Environment:
i686-pc-linux-gnu
>Description:
The following program should return -1, but returns 1 instead
(just compile with "g++ -c"):

-----------------------snip here------------------
#include<valarray>

int main ()
{
    std::valarray<int> u(1);
    u[0]=1;
    return (-u)[0];
}
-----------------------snip here------------------

I think that Gaby's patch

http://gcc.gnu.org/ml/gcc-patches/2002-08/msg00197.html

caused the regression, since the patch below fixes the
problem for me. (I didn't do a bootstrap or regression
tests, though.)
>How-To-Repeat:
g++ -c
>Fix:
===========================================================
--- gcc-old/libstdc++-v3/include/bits/valarray_meta.h	Sat Jan  4 01:37:31 2003
+++ gcc-new/libstdc++-v3/include/bits/valarray_meta.h	Wed Jan  8 23:26:44 2003
@@ -457,7 +457,7 @@ namespace std
       _UnBase(const _Arg& __e) : _M_expr(__e) {}
 
       value_type operator[](size_t __i) const
-      { return _M_expr[__i]; }
+      { return _Oper() (_M_expr[__i]); }
 
       size_t size() const { return _M_expr.size(); }
 
============================================================
>Release-Note:
>Audit-Trail:
>Unformatted: