Re: libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure

Re: libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure

[rodrigc]

The following reply was made to PR libstdc++/3141; it has been noted by GNATS.

From: rodrigc@gcc.gnu.org
To: bkoz@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-gnats@gcc.gnu.org,
  gcc-prs@gcc.gnu.org, kenny.simpson@gs.com, theonetruekenny@yahoo.com
Cc:  
Subject: Re: libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure
Date: 21 Nov 2001 01:57:34 -0000

 Synopsis: DR 243: basic_istream::get and getline when sentry reports failure
 
 State-Changed-From-To: feedback->closed
 State-Changed-By: rodrigc
 State-Changed-When: Tue Nov 20 17:57:34 2001
 State-Changed-Why:
     bkoz fixed this and provided a description why
     additional changes suggested by submitter should not be
     made.
 
 http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3141&database=gcc

Re: libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure

[Benjamin Kosnik]

The following reply was made to PR libstdc++/3141; it has been noted by GNATS.

From: Benjamin Kosnik <bkoz@redhat.com>
To: rodrigc@gcc.gnu.org
Cc: gcc-prs@gcc.gnu.org
Subject: Re: libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure
Date: Tue, 20 Nov 2001 19:34:29 -0800 (PST)

 Actually.... this should probably be kept open. There was some commentary 
 that I would like to respond to, when time permits.

Re: libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure

[rodrigc]

Synopsis: DR 243: basic_istream::get and getline when sentry reports failure

State-Changed-From-To: feedback->closed
State-Changed-By: rodrigc
State-Changed-When: Tue Nov 20 17:57:34 2001
State-Changed-Why:
    bkoz fixed this and provided a description why
    additional changes suggested by submitter should not be
    made.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&pr=3141&database=gcc

Re: libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure

[bkoz]

Synopsis: DR 243: basic_istream::get and getline when sentry reports failure

Responsible-Changed-From-To: unassigned->bkoz
Responsible-Changed-By: bkoz
Responsible-Changed-When: Tue Jun 12 14:36:31 2001
Responsible-Changed-Why:
    Mine.
State-Changed-From-To: open->feedback
State-Changed-By: bkoz
State-Changed-When: Tue Jun 12 14:36:31 2001
State-Changed-Why:
    I've made the char_type() changes.
    
    For the rest, I think your analysis is wrong. The point is that the terminating NULL *is* added unconditionally, and this is clarifying language to support this. Thus, the moving of the adding char_type() to terminate the string inside the if(sentry) loop would cause a defect, not fix one.
    
    There's a reason for the not checking n > 0 for the second bit, I can't remember it offhand but it has to do with changing required error flags. It's checked for in the testsuite, so changing this one small bit will probably show the failure. Anyway.
    
    -benjamin

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view&pr=3141&database=gcc

libstdc++/3141: DR 243: basic_istream::get and getline when sentry reports failure

[theonetruekenny]

>Number:         3141
>Category:       libstdc++
>Synopsis:       DR 243: basic_istream::get and getline when sentry reports failure
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Tue Jun 12 10:26:01 PDT 2001
>Closed-Date:
>Last-Modified:
>Originator:     theonetruekenny@yahoo.com
>Release:        20010612 cvs
>Organization:
>Environment:
source
>Description:
see
http://anubis.dkuug.dk/jtc1/sc22/wg21/docs/lwg-defects.html#243
>How-To-Repeat:
observe istream::get(char_type* __s, streamsize __n, char_type __delim)
in line 536 of include/bits/istream.tcc

note that on line 573: *__s = char_type(NULL);

observe istream::getline(char_type* __s, streamsize __n, char_type __delim)
in line 627 of include/bits/istream.tcc
note on line 673: *__s = char_type(NULL);
>Fix:
as per DR 243,
i.e.:
before the sentry construction in both the functions, add:
if( __n ) *s == char_type(); // same as charT() as specified in the DR 243

change the existing *__s = char_type(NULL)'s in 573/673 to
*s == char_type(); and move then inside the body of the if's on lines
540 and 631.

The condition in the if() on line 631 should be changed to match that
of the if() on line 540.
>Release-Note:
>Audit-Trail:
>Unformatted: