[llvm-bugs] [Bug 36914] New: libc++ incorrectly treats operator>>(istream& unsigned int&) as an error when converting negative values

[via llvm-bugs]

https://bugs.llvm.org/show_bug.cgi?id=36914

            Bug ID: 36914
           Summary: libc++ incorrectly treats operator>>(istream& unsigned
                    int&) as an error when converting negative values
           Product: libc++
           Version: 6.0
          Hardware: PC
                OS: Windows NT
            Status: NEW
          Severity: enhancement
          Priority: P
         Component: All Bugs
          Assignee: unassignedclangbugs at nondot.org
          Reporter: billy.oneal at gmail.com
                CC: llvm-bugs at lists.llvm.org, mclow.lists at gmail.com

Consider the following program:

#include <iostream>
#include <sstream>

int main()
{
    std::istringstream a("-1");
    unsigned i;

    a >> i;
    if (a.fail())
        std::cout << "Bug\n";
    else
        std::cout << "Ok\n";

    std::istringstream b("4294967296");

    b >> i;
    if (b.fail())
        std::cout << "Ok\n";
    else
        std::cout << "Bug\n";
    return 0;
}

libc++ thinks that -1 isn't representable in the type unsigned int, and so
treats this as an error. However, the standard disagrees with this. N4727
[istream.formatted.arithmetic] says that the operator>> calls locale's num_get.
[facet.num.get.virtuals]/3 says it reads as if by scanf %u. C11 7.21.6.2/12
says that %u behaves as if by strtoul. C11 7.22.1.4/5 says:

If the subject sequence begins with a minus sign, the value resulting from
the conversion is negated (in the return type).

That means that the parsing should remember that there was a minus sign, parse
the number as an unsigned int (with usual ERANGE checks), then, because the
input started with a minus sign, negate the resulting value.

(I ran into this bug when a user reported that they expected MSVC++ to treat -1
input as an error here; but libstdc++ agrees with us in not setting failbit
here, and C's rules, however bizarre, do not support setting failbit here)

-- 
You are receiving this mail because:
You are on the CC list for the bug.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.llvm.org/pipermail/llvm-bugs/attachments/20180327/21accaca/attachment.html>