[Nepomuk] Crash in nepomukqueryservice when running multiple instances of QueryServiceClient

[Sebastian Trüg]

There is no need for a workaround. We will simply fix the service. :)
All we need is a good test case to trigger the bug. Could you extract
that from your code?

Cheers,
Sebastian

On 04/21/2010 09:40 PM, Dario Freddi wrote:
> Hello people,
> 
> as I told Sebastian earlier today, I discovered a severe bug in 
> nepomukqueryservice. With telepathy-kde, I need to run several 
> QueryServiceClient (at least two) at the same time. Note that the application 
> has a single thread and every call to each qsc is asynchronous (I'm always 
> using sparqlQuery()).
> 
> If I instantiate and start more than a QSC, the dbus service crashes all the 
> time when calling Soprano::Client::ClientModel::executeQuery() and accessing 
> the mutex which is inside it. This situation is always reproducible in a short 
> time when instantiating at least 3 QSC.
> 
> To test some stuff, I created a workaround that I attached. I created a queue 
> that starts a query at a time, waits for its listing to finish, and then 
> starts the other queries in queue. This works perfectly for the first round, 
> but queryservice always crashes afterwards when trying to pick up changes. 
> Probably such a solution needs to be implemented right in queryservice, 
> letting one query at a time hit soprano.
> 
> If you want to test this yourself, you can apply this merge request: 
> http://reviewboard.kde.org/r/3650/ which also includes my "workaround"
> 
> 
> 
> 
> _______________________________________________
> Nepomuk mailing list
> Nepomuk at kde.org
> https://mail.kde.org/mailman/listinfo/nepomuk