Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.

Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.

[bkoz]

Synopsis: basic_string seems to fail to use template-defined allocator.

State-Changed-From-To: analyzed->feedback
State-Changed-By: bkoz
State-Changed-When: Sat Apr 20 16:55:25 2002
State-Changed-Why:
    I don't think this is a bug.
    
    Here's a testcase:
    
    #include "post_stl.h"
    #include <iostream>
    #include <sstream>
    #include <vector>
    #include <exception>
    #include <stdio.h>
    
    using namespace std;
    
    storage* sto;
    
    struct A
    {
      pstring 	msg;
      double 	num;
      short		pad[5];
    
      A(const char* s, streamsize len, double d) : msg(s, len), num(d) { }
    
      virtual 
      ~A() { }
    
      virtual string 
      getInfo()
      {
        ostringstream out;
        out << "class type 'A'" << endl;
        out << "msg object address = " << (void*)(&msg) << endl;
        out << "msg data address = " << (void*)(msg.data()) << endl;
        out << "msg = '" << msg.c_str() << "'" << endl; 
        out << "num = '" << num << "'" << endl;
        return out.str();
      }
    };
    
    typedef vector<A*, post_alloc<A*> > A_vector;
    
    int main()
    {
      sto = new storage("stringtest.odb");
      if (sto->open(storage::fixed)) 
        {
          A_vector *root = (A_vector*)sto->get_root_object();
          if (root != NULL) 
    	{
    	  cout << "Existing database found at " << root << endl;
    	  for (A_vector::iterator i = root->begin(); i != root->end(); i++) 
    	    {
    	      cout << "object found at " << *i << ":" << endl;
    	      cout << (*i)->getInfo();
    	    }
    
    	} 
          else 
    	{
    	  root = new (*sto) A_vector;
    	  cout << "Creating database at " << root << endl;
    	  sto->set_root_object((object*)root);
    	}
    
          cout << endl;
          cout << "Add some lines.  Terminate input with empty line." << endl;
    
          while (cin.good()) 
    	{
    	  char buf[256];
    	  cin.getline(buf, 256);
    	  streamsize len = cin.gcount() - 1;
    	  if (!len) 
    	    {
    	      cout << "done." << endl;
    	      sto->flush();
    	      sto->close();
    	      return EXIT_SUCCESS;
    	    } 
    	  else 
    	    {
    	      A* a = new (*sto) A(buf, len, 44.5);
    	      cout << "created new A object at " << a << endl;
    	      cout << a->getInfo() << endl;
    	      root->push_back(a);
    	    }
    	}
        } 
      else 
        {
          cerr << "Failed to open storage" << endl;
          return EXIT_FAILURE;
        }
    }
    
    Gives:
    %a.out 
    Creating database at 0x400005ac
    
    Add some lines.  Terminate input with empty line.
    this is a test
    
    created new A object at 0x4000060c
    class type 'A'
    msg object address = 0x40000610
    msg data address = 0x81344dc
    msg = 'this is a test'
    num = '44.5'
    
    And then re-running the binary:
    
    %a.out 
    Existing database found at 0x400005ac
    object found at 0x4000060c:
    class type 'A'
    msg object address = 0x40000610
    msg data address = 0x81344dc
    msg = ''
    num = '44.5'
    
    
    It looks like 
    
    1) find_storage in storage.h, line 248
    has problems finding a block for std::string because sizeof(string) != sizeof(string) + sizeof(string::_Rep)
    
    2) allocate, post_stl.h line 92:
    
    ends up using malloc to get storage.
    
    The results are obvious from the above output.
    
    I think this is an error with the POST++ allocation strategy, or else I'm not configuring it to work correctly.
    
    Thoughts?
    
    -benjamin

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5492

Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.

[pme]

Synopsis: basic_string seems to fail to use template-defined allocator.

State-Changed-From-To: feedback->closed
State-Changed-By: pme
State-Changed-When: Fri May 31 18:09:02 2002
State-Changed-Why:
    Not a bug.  See audit trail for analysis.  (POST++ is
    calling malloc directly instead of using the allocator.)

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5492

Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.

[Benjamin Kosnik]

The following reply was made to PR libstdc++/5492; it has been noted by GNATS.

From: Benjamin Kosnik <bkoz@redhat.com>
To: Phil Edwards <phil@jaj.com>
Cc: gcc-gnats@gcc.gnu.org, carlo@alinoe.com, kk71878@hotmail.com,
   gcc-bugs@gcc.gnu.org
Subject: Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.
Date: Fri, 19 Apr 2002 13:31:27 -0700

 Oh! I did some work on this and then never put it in the PR. It would be
 great if you could take over this phil. I'll add in the notes later
 today, but POST++ is allocating the string contents on the heap, so they
 are not being saved in the persistent object.
 
 I have a smaller example. I don't think this is a library bug anymore,
 but a post++ bug. 
 
 More details later

Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.

[Phil Edwards]

The following reply was made to PR libstdc++/5492; it has been noted by GNATS.

From: Phil Edwards <phil@jaj.com>
To: gcc-gnats@gcc.gnu.org
Cc: bkoz@redhat.com, carlo@alinoe.com, kk71878@hotmail.com,
   gcc-bugs@gcc.gnu.org
Subject: Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.
Date: Fri, 19 Apr 2002 16:01:03 -0400

 The two files attached to the audit trail aren't self-contained.  So I
 downloaded post-1.35, and the file stltest.cxx compiles fine for me using
 3.1 prerelease.  It runs without printing any error messages.  There aren't
 any instructions on what exactly I'm supposed to look for to tell whether
 it's working, so I don't know whether it works or not.
 
 It creates a file called "teststl.odb" but then doesn't do anything with it.
 None of the other binaries built during installation do anything with
 it either.
 
 
 Phil
 
 -- 
 If ye love wealth greater than liberty, the tranquility of servitude greater
 than the animating contest for freedom, go home and leave us in peace.  We seek
 not your counsel, nor your arms.  Crouch down and lick the hand that feeds you;
 and may posterity forget that ye were our countrymen.            - Samuel Adams

Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator

[Kevin Killingsworth]

The following reply was made to PR libstdc++/5492; it has been noted by GNATS.

From: "Kevin Killingsworth" <kk71878@hotmail.com>
To: bkoz@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, kk71878@hotmail.com, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
Cc:  
Subject: Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator
Date: Thu, 04 Apr 2002 21:45:49 -0600

 Benjamin,
 
 I checked out what Carlo suggested and couldn't find anything wrong with the 
 allocator the first time around.  I can do some further analysis possibly 
 this weekend.  How much time before the release?
 
 --Kevin
 
 
 >From: bkoz@gcc.gnu.org
 >Reply-To: bkoz@gcc.gnu.org, gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org,  
 >kk71878@hotmail.com, nobody@gcc.gnu.org, gcc-gnats@gcc.gnu.org
 >To: gcc-bugs@gcc.gnu.org, gcc-prs@gcc.gnu.org, kk71878@hotmail.com,  
 >nobody@gcc.gnu.org
 >Subject: Re: libstdc++/5492: basic_string seems to fail to use 
 >template-defined allocator.
 >Date: 4 Apr 2002 19:51:05 -0000
 >
 >Synopsis: basic_string seems to fail to use template-defined allocator.
 >
 >State-Changed-From-To: feedback->analyzed
 >State-Changed-By: bkoz
 >State-Changed-When: Thu Apr  4 11:51:05 2002
 >State-Changed-Why:
 >     I would like to get resolution on this before gcc-3.1 ships.
 >
 >     To do this, it looks like somebody will have to download post-1.35, 
 >and the file stltest.cxx and run it, try to find out where it breaks.
 >
 >     -benjamin
 >
 >http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5492
 
 
 
 
 _________________________________________________________________
 MSN Photos is the easiest way to share and print your photos: 
 http://photos.msn.com/support/worldwide.aspx
 

Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.

[bkoz]

Synopsis: basic_string seems to fail to use template-defined allocator.

State-Changed-From-To: feedback->analyzed
State-Changed-By: bkoz
State-Changed-When: Thu Apr  4 11:51:05 2002
State-Changed-Why:
    I would like to get resolution on this before gcc-3.1 ships. 
    
    To do this, it looks like somebody will have to download post-1.35, and the file stltest.cxx and run it, try to find out where it breaks.
    
    -benjamin

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5492

Re: libstdc++/5492: basic_string seems to fail to use template-defined allocator.

[rodrigc]

Synopsis: basic_string seems to fail to use template-defined allocator.

State-Changed-From-To: open->feedback
State-Changed-By: rodrigc
State-Changed-When: Sun Jan 27 10:15:16 2002
State-Changed-Why:
    Can you follow the steps outlined in http://gcc.gnu.org/bugs.html
    and submit a testcase, preferably preprocessed source?
    It is a bit too difficult to try to reproduce the
    problem which you describe in the PR without going through
    a lot of steps.

http://gcc.gnu.org/cgi-bin/gnatsweb.pl?cmd=view%20audit-trail&database=gcc&pr=5492