Hi, Thanks for your answer. I have created a JIRA issue with Quickstart: https://issues.apache.org/jira/browse/WICKET-6692
-- Daniel On Sat, Aug 3, 2019 at 10:05 PM Sven Meier <s...@meiers.net> wrote: > > Hi Daniel, > > can you create a quickstart and attach it to a Jura issue please? > > Thanks > Sven > > On 01.08.19 10:57, Daniel Stoch wrote: > > Correction to my previous message (I have debugged this more precisely): > > > > AbstractWebSocketProcessor.onClose method is called in both Jetty > > versions (not only in a newer one). > > The difference is in connection: > > - in Jetty 9.4.12.v20180830 connection is closed > > - in Jetty 9.4.18.v20190429 connection is still open > > > > So message is broadcasted only in a newer version. > > > > -- > > Daniel > > > > On Thu, Aug 1, 2019 at 10:25 AM Daniel Stoch <daniel.st...@gmail.com> wrote: > >> Hi, > >> > >> We are using web sockets (with wicket-native-core) on many pages in > >> our application. After upgrade a Jetty from 9.4.12.v20180830 to a > >> newer version 9.4.18.v20190429, I have found a different behaviour in > >> application: > >> > >> When user navigates to another page, a websocket connection is closed > >> and AbstractWebSocketProcessor.onClose method is called. This causes > >> broadcasting a message to connected page: > >> > >> broadcastMessage(new ClosedMessage(getApplication(), getSessionId(), > >> key)); > >> > >> and leads to page deserialisation (from PageStore). > >> I think something was changed in a new version of Jetty, because in > >> the previous used version this was not called. Maybe because of this: > >> https://github.com/eclipse/jetty.project/issues/3835 > >> https://github.com/eclipse/jetty.project/commit/2383bf4974ba7d82109cedfc4a8e7693d106abf0 > >> > >> I believe that now it works correctly (as was designed) and onClose > >> should be called. But I wonder how it can affect performance: almost > >> every page navigation causes page deserialization (when it should > >> occur only for back button or when some web socket message comes and > >> application need to process it). > >> > >> Maybe this message should be send only when needed or maybe I should > >> not care and current behaviour does not affect performance? > >> > >> -- > >> Best regards, > >> Daniel Stoch > > --------------------------------------------------------------------- > > To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org > > For additional commands, e-mail: users-h...@wicket.apache.org > > > > --------------------------------------------------------------------- > To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org > For additional commands, e-mail: users-h...@wicket.apache.org > --------------------------------------------------------------------- To unsubscribe, e-mail: users-unsubscr...@wicket.apache.org For additional commands, e-mail: users-h...@wicket.apache.org
