yes and no, depend what you do of it, the point is if you base your app on CDI (as much as possible I mean) and it starts to be common, you can put logic in these producers, typically wrapping of requests/responses (can be easier than using filters) and in this case this is often not 1-1 replacement. I know it is doable but needs to update the app and can break "big apps" where you aggregate multiple parts.
Having a namespace should be a best practise IMHO. Romain Manni-Bucau Twitter: @rmannibucau Blog: http://rmannibucau.wordpress.com/ LinkedIn: http://fr.linkedin.com/in/rmannibucau Github: https://github.com/rmannibucau 2014/1/4 Gerhard Petracek <gerhard.petra...@gmail.com>: > @romain: > i don't see an issue here - if you add the ds-servlet-module, you just drop > your own producers (which overlap and should do the same anyway). > > regards, > gerhard > > > > 2014/1/4 Romain Manni-Bucau <rmannibu...@gmail.com> > >> well in fact I saw a lot of cdi 1.0 app producing http* objects >> without qualifier cause it was missing in cdi so conflicts can occurs >> and are quite common >> Romain Manni-Bucau >> Twitter: @rmannibucau >> Blog: http://rmannibucau.wordpress.com/ >> LinkedIn: http://fr.linkedin.com/in/rmannibucau >> Github: https://github.com/rmannibucau >> >> >> >> 2014/1/4 Gerhard Petracek <gerhard.petra...@gmail.com>: >> > we had no qualifier for FacesContext (in codi, seam3,...). since it used >> to >> > be a common producer, we saw "compatibility issues". >> > however, with a proper documentation (how to veto one of them), no user >> > (i'm aware of) had a real issue with it and for the majority it was >> easier >> > to use (because there wasn't an issue at all). >> > >> > regards, >> > gerhard >> > >> > >> > >> > 2014/1/4 Mark Struberg <strub...@yahoo.de> >> > >> >> >> >> >> >> The question for me is: are there already known producers for it or is >> >> there any spec which introduces this? >> >> In that case a custom qualifier is always a good idea imo. Otherwise we >> >> would face different behaviour on different containers. They most times >> >> behave different... >> >> I just would like to avoid possible incompatibilities. And for this a >> >> Qualifier certainly works great - without much additional complexity. >> >> >> >> Does all the needed detection + veto really pay off? How do you know you >> >> are running in an environment which already has such a producer >> registered? >> >> This is not easy to accomplish! >> >> >> >> >> >> Thus I'm for keeping it. >> >> >> >> >> >> LieGrue, >> >> strub >> >> >> >> >> >> >> >> >> >> >________________________________ >> >> > From: Gerhard Petracek <gerhard.petra...@gmail.com> >> >> >To: dev@deltaspike.apache.org >> >> >Sent: Saturday, 4 January 2014, 12:57 >> >> >Subject: Re: Servlet Module - Do we really need @Web? >> >> > >> >> > >> >> >+1 for a veto in case of cdi 1.1. >> >> > >> >> >@external producers: >> >> >we can document it (how users can veto e.g. producers, if they see any >> >> >overlap). >> >> >however, deltaspike shouldn't add complexity just because there might >> be a >> >> >custom producer (for the same). >> >> > >> >> >regards, >> >> >gerhard >> >> > >> >> > >> >> > >> >> > >> >> >2014/1/4 Christian Kaltepoth <christ...@kaltepoth.de> >> >> > >> >> >> @John: Actually the Servlet module provides more than what CDI 1.1 >> adds. >> >> >> For example the event propagation and the recently added "WebStorage" >> >> for >> >> >> the resource loading and so on. So people may want to add the Servlet >> >> >> module even in a CDI 1.1 container. >> >> >> >> >> >> I'm also +0 for that. Of cause it would be nice to get rid of @Web. >> For >> >> the >> >> >> CDI 1.1 case we could actually veto our produces as Thomas suggested. >> >> But >> >> >> what about other portable extensions that may have producers for >> >> @Default. >> >> >> Say I'm using CDI 1.0 and also have Solder on the classpath. I think >> >> Solder >> >> >> is still a common dependency of some libraries, correct? In some >> regard >> >> it >> >> >> is nice to have a custom "namespace" for the producers. >> >> >> >> >> >> >> >> >> >> >> >> 2014/1/3 Thomas Andraschko <andraschko.tho...@gmail.com> >> >> >> >> >> >> > Because our customers have different servers (tomcat7 and even 6, >> >> >> > glassfish, jboss), so it would be a great enhancement for product >> >> >> > development. >> >> >> > >> >> >> > >> >> >> > 2014/1/3 John D. Ament <john.d.am...@gmail.com> >> >> >> > >> >> >> > > If you're in servlet 3.1/CDI 1.1 you don't even need the servlet >> >> >> > > module (so why include it as a dependency?) >> >> >> > > >> >> >> > > On Fri, Jan 3, 2014 at 1:09 PM, Romain Manni-Bucau >> >> >> > > <rmannibu...@gmail.com> wrote: >> >> >> > > > -0 both injections can be different depending on containers >> using >> >> >> some >> >> >> > > > advanced stuff out of ee but affecting ee lifecycle (at least >> in >> >> >> > tomcat) >> >> >> > > > but your proposal sounds acceptable. >> >> >> > > > Le 3 janv. 2014 17:58, "Thomas Andraschko" < >> >> >> > andraschko.tho...@gmail.com> >> >> >> > > a >> >> >> > > > écrit : >> >> >> > > > >> >> >> > > >> Hi, >> >> >> > > >> >> >> >> > > >> IMHO @Web is somehow annoying. >> >> >> > > >> HttpServlet e.g. is always "web", so @Web is just a overhead >> and >> >> >> > doesn't >> >> >> > > >> look nice. >> >> >> > > >> >> >> >> > > >> Can't we just veto the producers if CDI1.1 is available? >> >> >> > > >> The code would be the same with CDI 1.0 + DS, CDI 1.1 without >> or >> >> >> with >> >> >> > > DS. >> >> >> > > >> >> >> >> > > >> Regards, >> >> >> > > >> Thomas >> >> >> > > >> >> >> >> > > >> >> >> > >> >> >> >> >> >> >> >> >> >> >> >> -- >> >> >> Christian Kaltepoth >> >> >> Blog: http://blog.kaltepoth.de/ >> >> >> Twitter: http://twitter.com/chkal >> >> >> GitHub: https://github.com/chkal >> >> >> >> >> > >> >> > >> >> >>