Even though this is may be a serious issue, I do think that current 1.3.4 is important enough to warrant a release and leave this one for 1.3.5.
The memory issue and the cross session pollution are serious enough that this release is long overdue. Martijn On Fri, Jun 20, 2008 at 10:32 AM, Johan Compagner <[EMAIL PROTECTED]> wrote: > just looked at that get page lock code and i dont see a loophole > We sync over the complete part.. > > I dont really have the time coming 2-3 days to really test it further, > beginning of next week is the earliest > > johan > > > On Fri, Jun 20, 2008 at 10:27 AM, Johan Compagner <[EMAIL PROTECTED]> > wrote: > >> that one is really strange. >> It looks like a concurrency issue but pages cant be accessed by multiply >> threads at once. >> >> For example if you place a breakpoint at one of these breakpoints >> And you just do 2 request in your browser does it get to that place twice? >> I guess not >> >> Maybe the code to get the page lock has a loop that it isnt synced >> completely correctly and that 2 request that come in >> at exact the same time can both go further.. >> But does JMeter does that? Does JMeter do multiply request for the same >> session at the same time? >> I thought it has X number of sessions and then did 1 request per session >> >> johan >> >> >> >> On Fri, Jun 20, 2008 at 10:10 AM, Peter Ertl <[EMAIL PROTECTED]> wrote: >> >>> I would love to see 1.3.4 but I see some serious issue still open (and I >>> don't know how bad it is) >>> >>> https://issues.apache.org/jira/browse/WICKET-1693 >>> >>> seems to affect not only 1.4 but 1.3 as well. I just did a test on current >>> wicket 1.3 and got another bunch of NullPointerExceptions under stress. >>> >>> I uploaded test-project-for-wicket-1.3.tar.gz for reproducing this issue >>> on 1.3 (without generics). >>> >>> you need IDEA and jmeter to use it (hopefully that's not a big problem) >>> >>> I don't know exactly why this concurrency issue happens but in heay-load >>> environments it could be a blocker :-( >>> >>> Cheers >>> Peter >>> >>> >>> >>> Am 19.06.2008 um 14:06 schrieb Frank Bille: >>> >>> >>> [0]: https://issues.apache.org/jira/browse/WICKET-1646 >>>> [1]: https://issues.apache.org/jira/browse/WICKET-1667 >>>> >>>> On Thu, Jun 19, 2008 at 2:03 PM, Frank Bille <[EMAIL PROTECTED]> >>>> wrote: >>>> >>>> I was just about writing a similar mail. To my understanding, the >>>>> two[0][1] >>>>> regression/showstopper issues as been fixed. I don't know of any other >>>>> issues, which needs to be fixed before. >>>>> >>>>> It's nice of you to do the build, but I actually have real, dedicated >>>>> spare >>>>> time on saturday, so perhaps you have some social relations that needs >>>>> attention :-) >>>>> >>>>> Frank >>>>> >>>>> [0] >>>>> [1] >>>>> >>>>> >>>>> On Thu, Jun 19, 2008 at 1:50 PM, Martijn Dashorst < >>>>> [EMAIL PROTECTED]> wrote: >>>>> >>>>> I can spare a couple of cycles to build 1.3.4, but I thought there >>>>>> were some regressions? Is 1.3.4 in releasable state? >>>>>> >>>>>> Martijn >>>>>> >>>>>> -- >>>>>> Become a Wicket expert, learn from the best: http://wicketinaction.com >>>>>> Apache Wicket 1.3.3 is released >>>>>> Get it now: http://www.apache.org/dyn/closer.cgi/wicket/1.3.3 >>>>>> >>>>>> >>>>> >>>>> >>> >> > -- Become a Wicket expert, learn from the best: http://wicketinaction.com Apache Wicket 1.3.3 is released Get it now: http://www.apache.org/dyn/closer.cgi/wicket/1.3.3
