Hi Marc, Thanks for the investigation. I've fixed the regressions in SVN trunk and the build should pass again. It does on my machine.
Best regards, Jerome -- Restlet ~ Founder and Technical Lead ~ http://www.restlet.org Noelios Technologies ~ http://www.noelios.com -----Message d'origine----- De : Marc Portier [mailto:[email protected]] Envoyé : vendredi 3 septembre 2010 09:03 À : [email protected] Objet : Re: stream closed errors On 08/28/2010 06:44 PM, Jerome Louvel wrote: > Hi Freya, > > This is probably related to the Message#finalize() which is called at GC time > and closes the underlying entity streams. The problem with Tomcat and > probably Jetty as well is that those streams might have been reused for other > call, thus provoking some side-effects. > > I've just removed Message#finalize() method in SVN trunk to remove this > potential side effet. Could you try again with a recent snapshot and let us > know if it helps? > Jerome, Sorry for not answering earlier. Own investigation in http://dev.outerthought.org/trac/outerthought_kauri/ticket/349 shows that this message#finalize should be the solution indeed. I'm working towards testing the current trunk and will let you know. However, the build of trunk isn't currently working, so I am working towards fixing that first and will create patches allong the way. First one is at: http://restlet.tigris.org/issues/show_bug.cgi?id=1175 regards, -marc= > Best regards, > Jerome > -- > Restlet ~ Founder and Technical Lead ~ http://www.restlet.org Noelios > Technologies ~ http://www.noelios.com > > > > > > -----Message d'origine----- > De : Thierry Boileau [mailto:[email protected]] > Envoyé : vendredi 27 août 2010 11:56 > À : [email protected] > Objet : RE: stream closed errors > > Hello Freya, > > I have no answer yet, just some thoughts. >> From what I notice in the code source of Jetty, this exception happens when >> the output stream is actually closed (but I can't see in the source code why >> this would happen), and when the writing operation is blocked for a too long >> time (for example the client take time to read the already sent bytes). I >> read also in the javadocs that instances of the HttpOutput class can be >> reused (I can't imagine this happens before the writing has been fullfilled). > > What kind of clients consume the responses? > > Best regards, > Thierry Boileau > > ------------------------------------------------------ > http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447&dsMessageId > =2652015 > > ------------------------------------------------------ > http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447&dsMessageId > =2652435 > ------------------------------------------------------ http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447&dsMessageId=2655037 ------------------------------------------------------ http://restlet.tigris.org/ds/viewMessage.do?dsForumId=4447&dsMessageId=2655086
