On Mon, 2009-03-23 at 20:52 +0100, Oleg Kalnichevski wrote: > Thorsten Scherler wrote: > > On Sun, 2009-03-22 at 00:07 +0100, Oleg Kalnichevski wrote: > >> Oleg Kalnichevski wrote: > > ... > >> The problem is not caused by the HTTP transport. The > >> MultiThreadedTaskMaster appears to have synchronization issues and > >> should also be reviewed for thread-safety. > > > > Ok, do you have any method in special as possible candidates? Will now > > look into it and hopefully find the problem. > > > > salu2 > > I have looked at MultiThreadedTaskMaster quite extensively and got an > impression that it simply did not attempt to synchronize access to its > instance variables at all despite the fact they could be accessed from > multiple threads of execution. I'll try to find time to propose a fix > for the issue, but cannot really tell when I can manage to get around to > doing that. >
Thank you very much Oleg, looking forward. I ATM preparing some small example for ApacheCon and your recent change shows the sequential capacity so that is good and brings back a running example. salu2 -- Thorsten Scherler <thorsten.at.apache.org> Open Source <consulting, training and solutions> -- Thorsten Scherler <scherler.at.gmail.com> Open Source <consulting, training and solutions>
