Bertrand Delacretaz wrote:
-----Original Message----- From: Bertrand Delacretaz [mailto:[EMAIL PROTECTED] Sent: Sunday, February 08, 2004 9:49 AM To: [EMAIL PROTECTED] Subject: Re: DO NOT REPLY [Bug 26753] - Persistent store or cache corruption?
Le Samedi, 7 f�v 2004, � 23:44 Europe/Zurich, Geoff Howard a �crit :
[EMAIL PROTECTED] wrote:
> http://nagoya.apache.org/bugzilla/show_bug.cgi?id=26753
Persistent store or cache corruption? ------- Additional Comments From [EMAIL PROTECTED] 2004-02-07 18:34 ------- Afaik, this is a known Jisp bug which should be fixed according to Scott with Jisp 3.0. We could try to switch to Jisp 3.0 after the release and see what happens.
Are you saying after the release because Jisp 3.0 would require more changes in the excalibur/cocoon code? From what I read about this bug, I'd say it's important enough to try to fix it before the release if at all possible, no?
Releasing with this bug worries me as well. In my tests I've seen either garbage or wrong pages (ie another page than the one I did request) being served, this is a serious problem.
The Cocoon core is known to be very stable and reliable, we don't want to give a bad impression here IMHO. I'd consider it a blocker and would vote against the release unless there is at least a workaround.
Now, in general I totally agree with you, but I fear that this bug is in there for a long time, which means all 2.1.x releases have already this bug (perhaps I'm wrong). Afaik, this bug happens only on some environments and we never managed to exactly find out, what the real problem is (perhaps a specific JDK version on a specific OS, or a combination, don't know). Anyways, I would say: if this bug is already in 2.1.3, we could release 2.1.4 with it as well. We fixed a lot of other problems, so at least 2.1.4 is not worse than 2.1.3. I don't know the impact of switching to 3.0 of Jisp. It is a dot zero release which could cause other problems. So, if we switch to 3.0 we need imho a longer testing period which would in my view delay the 2.1.4 release.
I didn't realize this was not a universal problem. The impression I had was that this was consistently repeatable given the same actions in any installation.
Carsten, as I can't find any information on the Jisp aspect of this, do you have any links or hints where to find clues on the environment dependencies? If this is limited to an OS or JDK version, it would be very helpful to consider that.
Still, I agree that in the end we may need to mark this as a known issue if it can't be resolved quickly, which appears to be the case.
Geoff
