Le mar. 30 juin 2020 à 13:06, Mark Struberg <strub...@yahoo.de.invalid> a écrit :
> Oh shees, yes was on a different track in my head. I think the CDI spec > doesn't mention websockets at all. Probably that's why we don't integrate > with it. > I remember that we once had the discussion and the outcome was that it > might be too expensive to do it every time. And people can use > CDI.current() anyway IF they need it. At least that was what I remember > from the discussions we had in the CDI EG. Not sure if there have been > discussions on the EE umbrella level. > The spec document (~ PDF) is empty but javadoc mentions that EE context looks up in EE context - which means what we want I guess. That said, realisticly it does not sound crazy to say the websocket components have at least injections even if they are not beans as JAXRS, Servlet, JSF and all EE components. In terms of cost it is likely less costly than for servlet components cause a connection lasts way longer than for webapps, likely something to enhance but I'm not sure today the usage of this module. From my small window I saw only custom integrations and rarely so deep integrations so not sure it is worth it. > > LieGrue, > strub > > > Am 30.06.2020 um 10:23 schrieb Romain Manni-Bucau <rmannibu...@gmail.com > >: > > > > Hmm, I'm not following, spoke of websockets, not async requests and as an > > user I expect a tomcat-owb integration module to make both working > together > > otherwise it is pointless to have an integration module not integrating > > technologies. > > Now the spec mentions it is supported in an EE world - we are not there > but > > I guess we should be consistent with it in such a module: > > > https://jakarta.ee/specifications/websocket/1.1/apidocs/javax/websocket/server/ServerEndpointConfig.Configurator.html#getEndpointInstance-java.lang.Class- > > > > I suspect tomcat could use by itself the instance manager to avoid this > > integration need but then it would leak the "release" call - it must be > > compensated by another mechanism since the websocket spec does not give > it. > > > > Anyway, we can open a new thread on the impl, didn't intend to hijack > this > > one. > > > > Le mar. 30 juin 2020 à 09:59, Mark Struberg <strub...@yahoo.de.invalid> > a > > écrit : > > > >> It means we support @RequestScoped and ApplicationScoped beans in Async > >> Requests and events as required by the spec. > >> > >> The rest is afaik not portable. > >> > >> Happy to be proven wrong ;) > >> > >> LieGrue, > >> strub > >> > >> > >>> Am 30.06.2020 um 09:39 schrieb Romain Manni-Bucau < > rmannibu...@gmail.com > >>> : > >>> > >>> If does support means you can write websocket code, sure. > >>> If it means "works with cdi beans" then I think we miss this class: > >>> > >> > https://github.com/apache/tomee/blob/master/tomee/tomee-catalina/src/main/java/org/apache/tomee/catalina/websocket/JavaEEDefaultServerEnpointConfigurator.java > >>> > >>> Romain Manni-Bucau > >>> @rmannibucau <https://twitter.com/rmannibucau> | Blog > >>> <https://rmannibucau.metawerx.net/> | Old Blog > >>> <http://rmannibucau.wordpress.com> | Github < > >> https://github.com/rmannibucau> | > >>> LinkedIn <https://www.linkedin.com/in/rmannibucau> | Book > >>> < > >> > https://www.packtpub.com/application-development/java-ee-8-high-performance > >>> > >>> > >>> > >>> Le mar. 30 juin 2020 à 09:29, Mark Struberg <strub...@yahoo.de.invalid > > > >> a > >>> écrit : > >>> > >>>> our tomcat integration does support websockets afair. > >>>> > >>>> LieGrue, > >>>> strub > >>>> > >>>>> Am 30.06.2020 um 08:20 schrieb Romain Manni-Bucau < > >> rmannibu...@gmail.com > >>>>> : > >>>>> > >>>>> @Gurkan Erdogdu <gurkanerdo...@yahoo.com> the first reason to keep > >>>> tomcat > >>>>> module is that we release it whereas o.a.tomcat:tomcat-owb does not > >> exist > >>>>> for end users today ("you can build from source" is not an option > IMHO > >>>> and > >>>>> justifies to fork in most cases). The main difference in terms of > code > >> is > >>>>> the fact tomcat integration provides a valve for the principal > whereas > >> we > >>>>> only use a filter but I guess it is enough since valve will prevent > to > >>>>> position the filter - = capture of the principal - in the filter > chain > >>>> and > >>>>> can therefore break apps even if it is tempting to make it always win > >> (we > >>>>> shouldn't use a thread local but we don't have much options there). > >> Both > >>>>> impl miss websocket integration - tomee has it - so it looks like > >>>>> tomcat-owb is a fork of our module today, not much so release point > is > >> a > >>>>> blocker for me. > >>>>> > >>>>> With jakarta I guess we can maybe ask tomcat+jetty to get an official > >>>>> servlet components injections and drop all specific code. > >>>>> > >>>>> Last point about the consistency for jetty AND tomcat is also key for > >> me, > >>>>> there is no reason to favor jetty and not tomcat. > >>>>> > >>>>> +1 to drop the version from the module though, it does not make sense > >>>>> anymore - was for 6 -> 7 move IIRC. > >>>>> > >>>>> Romain Manni-Bucau > >>>>> @rmannibucau <https://twitter.com/rmannibucau> | Blog > >>>>> <https://rmannibucau.metawerx.net/> | Old Blog > >>>>> <http://rmannibucau.wordpress.com> | Github < > >>>> https://github.com/rmannibucau> | > >>>>> LinkedIn <https://www.linkedin.com/in/rmannibucau> | Book > >>>>> < > >>>> > >> > https://www.packtpub.com/application-development/java-ee-8-high-performance > >>>>> > >>>>> > >>>>> > >>>>> Le mar. 30 juin 2020 à 00:34, Gurkan Erdogdu < > cgurkanerdo...@gmail.com > >>> > >>>> a > >>>>> écrit : > >>>>> > >>>>>> Hi Remy > >>>>>> > >>>>>>> I would think you should keep "tomcat7" too, it's not really the > same > >>>>>> idea > >>>>>>> as modules/owb. > >>>>>>> > >>>>>> I have looked at both implementations and both are the same purpose, > >>>>>> injection into Servlet related classes and get the current > Principal > >>>> from > >>>>>> the request. In Tomcat/OWB module, its integration is more natural > >> than > >>>> the > >>>>>> Tomcat7 module. > >>>>>> What is the benefit of using Tomcat7 in OWB? > >>>>>> Regards. > >>>>>> Gurkan > >>>>>> > >>>>>> > >>>>>> On Tue, Jun 30, 2020 at 12:33 AM Rémy Maucherat <r...@apache.org> > >>>> wrote: > >>>>>> > >>>>>>> On Mon, Jun 29, 2020 at 1:54 PM Romain Manni-Bucau < > >>>>>> rmannibu...@gmail.com> > >>>>>>> wrote: > >>>>>>> > >>>>>>>> +1 to drop jms module, never saw any usage of it > >>>>>>>> -0.5 for tomcat7, rational being that if we want to do it, we > should > >>>>>> (at > >>>>>>>> the same time) 1. ensure tomcat module is at least 1-1 (not the > >> case I > >>>>>>>> think) + released properly and not just a sandbox and 2. drop > jetty > >>>>>>>> integration too (which can be envisioned since we worked to > >> integrate > >>>>>> OWB > >>>>>>>> in jetty itself) but dropping tomcat7 module without these two > >>>>>> conditions > >>>>>>>> looks like an user regression to me. > >>>>>>>> > >>>>>>> > >>>>>>> I would think you should keep "tomcat7" too, it's not really the > same > >>>>>> idea > >>>>>>> as modules/owb. The main problem is using a version number in the > >>>> module > >>>>>>> name, that creates confusion in the long run and gives the > impression > >>>> it > >>>>>> is > >>>>>>> outdated. Tomcat 7 will be eoled "soon", for example. > >>>>>>> > >>>>>>> Rémy > >>>>>>> > >>>>>>> > >>>>>>>> > >>>>>>>> I guess ee modules can move to tomee too - any other consumer - > with > >>>>>> the > >>>>>>>> relevant adaptations to our codebase? > >>>>>>>> > >>>>>>>> Romain Manni-Bucau > >>>>>>>> @rmannibucau <https://twitter.com/rmannibucau> | Blog > >>>>>>>> <https://rmannibucau.metawerx.net/> | Old Blog > >>>>>>>> <http://rmannibucau.wordpress.com> | Github < > >>>>>>>> https://github.com/rmannibucau> | > >>>>>>>> LinkedIn <https://www.linkedin.com/in/rmannibucau> | Book > >>>>>>>> < > >>>>>>>> > >>>>>>> > >>>>>> > >>>> > >> > https://www.packtpub.com/application-development/java-ee-8-high-performance > >>>>>>>>> > >>>>>>>> > >>>>>>>> > >>>>>>>> Le lun. 29 juin 2020 à 13:38, Gurkan Erdogdu < > >>>> cgurkanerdo...@gmail.com > >>>>>>> > >>>>>>> a > >>>>>>>> écrit : > >>>>>>>> > >>>>>>>>> Hi folks > >>>>>>>>> I would like to discuss to remove the following modules from the > >> OWB > >>>>>>> code > >>>>>>>>> base. > >>>>>>>>> > >>>>>>>>> - webbeans-jms : We introduced this module years ago for JMS but > >>>>>>>> frankly > >>>>>>>>> never see any usage. Also, it was not completed. > >>>>>>>>> - webbeans-tomcat7 : We introduced this modules for Tomcat7 > >>>>>>>> integration > >>>>>>>>> but now it is useless and Tomcat already includes this > integration > >>>>>>>> with > >>>>>>>>> more natural way ( > >>>>>>>>> https://github.com/apache/tomcat/tree/master/modules/owb) > >>>>>>>>> > >>>>>>>>> WDYT? Any objection? > >>>>>>>>> Regards. > >>>>>>>>> Gurkan > >>>>>>>>> > >>>>>>>> > >>>>>>> > >>>>>> > >>>>>> > >>>>>> -- > >>>>>> Gurkan Erdogdu > >>>>>> http://gurkanerdogdu.blogspot.com > >>>>>> > >>>> > >>>> > >> > >> > >
