[Phplib-trackers] [ phplib-Bugs-473196 ] ac_get_lock() and MySQL

[<no...@so...>]

Bugs item #473196, was opened at 2001-10-20 12:25
You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=403611&aid=473196&group_id=31885

Category: Data storage container
Group: current CVS
Status: Closed
Resolution: Duplicate
Priority: 5
Submitted By: Andreas Stein (steinas)
Assigned to: Chris Johnson (chaska)
Summary: ac_get_lock() and MySQL

Initial Comment:
file: ct_sql.inc
object: ct_sql
method: ac_get_lock()

  function ac_get_lock() {
    if ( "" != $this->database_lock_semaphore ) {
      $query = sprintf("SELECT get_lock('%s')", $this-
>database_lock_semaphore);
      while ( ! $this->db->query($query)) {
        $t = 1 + time(); while ( $t > time() ) { ; }
      }
    }
  }

for MySQL the SQL statement should be:

"SELECT GET_LOCK(str, time_out)" or
"SELECT GET_LOCK(str, 0)"

$db->query($query) is always true even if lock state 
is false

----------------------------------------------------------------------

>Comment By: Chris Johnson (chaska)
Date: 2002-09-25 13:34

Message:
Logged In: YES 
user_id=279065

I've closed this report because report #550100 completely covers 
the situation.  ct_sql.inc as written depends on the database 
being MySQL, and worse, does not even use currently correctly 
get_lock(name, timeout) syntax.  Other databases may not even 
support this locking idea (Postgres does not, Oracle seems to).

----------------------------------------------------------------------

Comment By: Chris Johnson (chaska)
Date: 2002-09-25 13:31

Message:
Logged In: YES 
user_id=279065

I've closed this report because report #550100 completely covers 
the situation.  ct_sql.inc as written depends on the database 
being MySQL, and worse, does not even use currently correctly 
get_lock(name, timeout) syntax.  Other databases may not even 
support this locking idea (Postgres does not, Oracle seems to).

----------------------------------------------------------------------

Comment By: nathan hruby (nhruby)
Date: 2002-07-02 14:02

Message:
Logged In: YES 
user_id=19736

Please see Bug #550100 for further discussion of ac_get_lock.

This is still a valid bug and shoudl be fixed for 7.4-final;
however please know that it will most likely be scrapped
and/or changed a good bit in the next major release.



----------------------------------------------------------------------

Comment By: Richard Archer (richardarcher)
Date: 2002-04-29 05:35

Message:
Logged In: YES 
user_id=279311

What happens if MySQL isn't the underlying database?

This is so broken that the only way I can see to fix it is
to delete all the code from this function and abolish
session/application locks!

I've posted another bug outlining a possible long-term fix.

----------------------------------------------------------------------

You can respond by visiting: 
https://sourceforge.net/tracker/?func=detail&atid=403611&aid=473196&group_id=31885