6 messages in com.mysql.lists.bugsNew bug in query cache with MySQL 4.0...
FromSent OnAttachments
Fournier Jocelyn [Presence-PC]19 Dec 2001 14:12 
Michael Widenius19 Dec 2001 20:11 
SDiZ (UHome)20 Dec 2001 04:33 
Michael Widenius20 Dec 2001 14:19 
SDiZ (UHome)20 Dec 2001 22:14 
Michael Widenius21 Dec 2001 05:04 
Subject:New bug in query cache with MySQL 4.0.1 ?
From:Michael Widenius (mon@mysql.com)
Date:12/19/2001 08:11:56 PM
List:com.mysql.lists.bugs

Hi!

"Fournier" == Fournier Jocelyn <Fournier> writes:

Fournier> Hi, Fournier> I noticed a strange behaviour with the query cache.

Fournier> How-to-repeat :

Fournier> I have the following query in my cache :

Fournier> SELECT haut,bas FROM forumpubhardwarefr WHERE cat='7';

Fournier> If I execute this command without performing "use MyDatabase", it
doesn't Fournier> matter, it executes the query. (No "No Database Selected" message) Fournier> I wonder what would be the cache behaviour if there were two tables
with the Fournier> same name in two different databases...

This is a bug that need to be fixed at once.

Sanja, can you please take a look at this. The used database should be part of the query cache key!

Fournier> Moreover, after performing a "flush query cache", I noticed a
corruption in Fournier> data stored in some table :

Fournier> The query in the cache :

Fournier> |<center> Fournier> <!-- DEBUT / Cybermonitor / START -->

Fournier> <script language="javascript"> Fournier> <!-- Fournier> CM_RUBRIQUE = "forum"; Fournier> CM_CLIENT = "harc @c @ Fournier> CM_SECTION1 = "Blabla";

Fournier> The query not cached :

Fournier> |<center> Fournier> <!-- DEBUT / Cybermonitor / START -->

Fournier> <script language="javascript"> Fournier> <!-- Fournier> CM_RUBRIQUE = "forum"; Fournier> CM_CLIENT = "hardware"; Fournier> CM_SECTION1 = "Blabla";

Fournier> So there is still a problem with flush query cache, which corrupt the
data Fournier> :(

I have tried to repat this, but I haven't yet succeded!

Fournier> I will try to make a testcase as soon as I have time...

Thanks! I would really appreciate if you could do this ! I wouldn't like to release the current code if I knew that it's likely that there is a fatal bug in it!

Regards, Monty