On Fri, 2019-02-22 at 12:41 +0100, Karl Pauls wrote:
> On Fri, Feb 22, 2019 at 12:35 PM Robert Munteanu <[email protected]>
> wrote:
> > On Fri, 2019-02-22 at 10:47 +0000, David Bosschaert wrote:
> > > Hi Robert,
> > > 
> > > Regarding the question in the subject line "Why does loading
> > > classes
> > > reach
> > > into the service registry?"
> > > This is most likely to see if there are any registered weaving
> > > hooks
> > > which
> > > can weave the class bytes.
> > > 
> > > Best regards,
> > > 
> > > David
> > > 
> > > [1]
> > > https://osgi.org/specification/osgi.core/7.0.0/framework.weavinghooks.html
> > 
> > Interesting. So I guess your advice would be "don't do non-trivial
> > stuff in static initializers since they might break your class?".
> 
> No. While that probably is good advice in general, this doesn't seem
> right. We might be looking at a bug in felix. A classload should not
> fail because the thread that instigated it was interrupted. Can you
> reproduce this?

I have only seen this once and it seems very timing-dependant. I filed

  https://issues.apache.org/jira/browse/FELIX-6067

for further discussion.

Robert

Reply via email to