[Bug c++/13084] New: class string with find(const char* cstr, size_type idx) member have different gcc 2.95 and gcc 3.3 behaviour

[Bug c++/13084] New: class string with find(const char* cstr, size_type idx) member have different gcc 2.95 and gcc 3.3 behaviour

[ronald dot van dot uden at rabobank dot com]

problem: 
char* charPointerString = input;
while ((iPos = myStr.find(charPointerString, iPos)) != string::npos){}//an 
infinite loop
workaround:
string myString(input);
while ((iPos = myStr.find(myString, iPos)) != string::npos){}//

gcc 2.95 and gcc 3.3 implementation difference of class string with find() 
members
 
the string function
size_type string::find (const char* cstr, size_type idx) const
never finds the string it is looking for in gcc33 when cstr==""
 
the string function
size_type string::find(const string& str, size_type idx) const
works properly on gcc33 when str==""
______________________________________________________
the function find (const char* cstr, size_type idx)  seems to work fine on 
gcc2.95

Operating System: Sun Solaris9

-- 
           Summary: class string with find(const char* cstr, size_type idx)
                    member have different gcc 2.95 and gcc 3.3 behaviour
           Product: gcc
           Version: 3.3
            Status: UNCONFIRMED
          Severity: critical
          Priority: P2
         Component: c++
        AssignedTo: unassigned at gcc dot gnu dot org
        ReportedBy: ronald dot van dot uden at rabobank dot com
                CC: gcc-bugs at gcc dot gnu dot org


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13084

[Bug libstdc++/13084] class string with find(const char* cstr, size_type idx) member have different gcc 2.95 and gcc 3.3 behaviour

[pinskia at gcc dot gnu dot org]

------- Additional Comments From pinskia at gcc dot gnu dot org  2003-11-17 14:17 -------
Can you provide a full testcase?

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|UNCONFIRMED                 |WAITING
          Component|c++                         |libstdc++


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13084

[Bug libstdc++/13084] class string with find(const char* cstr, size_type idx) member have different gcc 2.95 and gcc 3.3 behaviour

[paolo at gcc dot gnu dot org]

------- Additional Comments From paolo at gcc dot gnu dot org  2003-11-18 09:04 -------
I strongly suspect submitter is thinking about this kind of testcase

#include <string>
#include <cassert>

int main()
{
  std::string aaa = "Francesca";

  char* lit = "";
  assert( aaa.find(lit) == 0 );

  std::string str("");
  assert( aaa.find(str) == 0 );
}

What happens here, is that v2's find returns 9 whereas v3 returns 0 (for
both the tests)

Now, 9 is definitely wrong since is past the end of the string aaa and
find is supposed to return aither a valid index inside aaa or string::npos.

However, I'm not deadly sure that 0 is correct: just let me think a bit more
about the issue...

Paolo.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|WAITING                     |ASSIGNED
     Ever Confirmed|                            |1
   Last reconfirmed|0000-00-00 00:00:00         |2003-11-18 09:04:49
               date|                            |


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13084

[Bug libstdc++/13084] class string with find(const char* cstr, size_type idx) member have different gcc 2.95 and gcc 3.3 behaviour

[paolo at gcc dot gnu dot org]

------- Additional Comments From paolo at gcc dot gnu dot org  2003-11-18 09:58 -------
On second tought, the true reason why 9 is not ok whereas 0 *is* ok is that,
according to the standard (21.3.6.1, p1): "determines the _lowest_ possible
xpos, if possible, such as both the following conditions..." (emphasis mine).

Now, "the following conditions" are fulfilled both by 0 and 9, and 0 is the
correct final answer. This is also consistent with the behavior for the general
str.find(lit, pos = 0), returning pos for lit == "" (when pos <= str.size(),
otherwise npos, of course).

I'm closing the report as not a bug.

-- 
           What    |Removed                     |Added
----------------------------------------------------------------------------
             Status|ASSIGNED                    |RESOLVED
         Resolution|                            |INVALID


http://gcc.gnu.org/bugzilla/show_bug.cgi?id=13084