Hi Sebastien,

Thanks for investigating. Re. the  not stopping  issue...


> I suggest to not take risks as the problem existed in previous releases
> and only release the fix in release 1.2.


+1


> While debugging I came across a more serious issue with the 1.1 distro,
> described in TUSCANY-1954 [2].

 <http://issues.apache.org/jira/browse/TUSCANY-1953>
> [2] http://issues.apache.org/jira/browse/TUSCANY-1954
>

I've seen this issue for all of the releases we've done to date. It hasn't
been impacting the correct running of the samples as far as I can tell and
is a product of the slightly strange situation of

1 - shipping both host-tomcat and host-jetty in the manifest/all jar
2 - providing no way of choosing which to use

Each host is registered by its activator, e.g, JettyRuntimeModuleActivator;

servletHosts.addServletHost(server);

The servlet host is chosen in, for example, the ws-axis binding, when the
provider factory (Axis2BindingProviderFactory) is created.

this.servletHost = servletHosts.getServletHosts().get(0);

The order in which the Activators are processed, and hence which one as
returned by get(0), can vary from run to run not least because the service
providers are manipulated via a HashSet as they are read from the classpath.


I suggest we choose one or the other for the all/manifest jar.

Simon

Reply via email to