Bug #360 Severe memory leak in MySQL++ Result copy operator

• Submitted: 2 May 2003 13:52
• Modified: 3 May 2003 9:00
• Reporter: Jeremy Linton
• Status: Closed
• Category: MySQL++
• Severity: S1 (Critical)
• Version: 1.7.9
• OS: Linux (Linux)
• CPU Architecture: Any

[2 May 2003 13:52] Jeremy Linton

Description:
There is a bad memory leak in the MySQL++ Result copy operator. It is irritated when an empty result set is assigned to a non empty result set.

How to repeat:
code like the following

Result list_of_volumes;
while (1)
{
  query << "LOCK TABLES Volumes Read";
  list_of_volumes = query.store();
  query << "select * from Volumes";
  list_of_volumes = query.store();
}

Suggested fix:
reverse the purge() and blank result check code in ResUse::copy(const ResUse& other)

Initial code:
void ResUse::copy(const ResUse& other) {
  if (!other.mysql_res) {
     mysql_res=0; _types=0; _names=0; return;
  }
  if (initialized)
    purge();
.....

Change to 

void ResUse::copy(const ResUse &other)
{
   //maybe check for this==other...?
   //
   // clear the current copy...
   if (initialized) {
     purge();
   }
   //blank result set?
   if (!other.mysql_res) {
      mysql_res=0; 
      _types=0; 
      _names=0; 
      return;
   }

[3 May 2003 9:00] MySQL Verification Team

SQLQuery::operator<< 

was not designed to handle commands like "LOCK TABLES".

But, still this patch will be applied in a little bit modified format.

Thank you.

Temporary Workaround:

use Connection::exec() method to run commands like "LOCK TABLES"