This is the mail archive of the gcc-prs@gcc.gnu.org mailing list for the GCC project.


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

libstdc++/6360: Wrong workaround in char_traits.h - ignore stops on \0xff



>Number:         6360
>Category:       libstdc++
>Synopsis:       Wrong workaround in char_traits.h - ignore stops on \0xff
>Confidential:   no
>Severity:       serious
>Priority:       medium
>Responsible:    unassigned
>State:          open
>Class:          sw-bug
>Submitter-Id:   net
>Arrival-Date:   Thu Apr 18 08:06:00 PDT 2002
>Closed-Date:
>Last-Modified:
>Originator:     Mads Kiilerich
>Release:        3.0.4
>Organization:
>Environment:
Seen on RedHat linux 6.2 - and on Solaris 2.7
>Description:
The default delimiter for Istream::ignore is EOF (-1).

Several places in include/g++-v3/bits/istream.tcc the following is found:

    const int_type __idelim = traits_type::to_int_type(__delim);

calling the method in gcc/include/g++-v3/bits/char_traits.h:184 :

        // To keep both the byte 0xff and the eof symbol 0xffffffff
        // from ending up as 0xffffffff.
        static int_type
        to_int_type(const char_type& __c)
        { return static_cast<int_type>(static_cast<unsigned char>(__c)); }

- but then __idelim is set to 255, causing Istream::ignore to stop ignoring first time it meets a char value of 255.

In this case the delimiter should be -1, so that ignorance only stops when a real negative unsigned char is found - and that hardly happens...
>How-To-Repeat:
The following testcase is expected to do as it does when compiled with KAI:

[mk@linux-tools IstreamIgnore]$ ./IstreamIgnore.KCC
1st char is: a
Skipping 8 chars (i.e. until after the digits)...
The rest of the stream contains: 'abcd'

But using gcc 3.0.4 we get:

[mk@linux-tools IstreamIgnore]$ ./IstreamIgnore.gcc
1st char is: a
Skipping 8 chars (i.e. until after the digits)...
The rest of the stream contains: '1234abcd'
>Fix:
Idunno. But the current workaround does not seem good...
>Release-Note:
>Audit-Trail:
>Unformatted:


Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]