Re: Introduction
On 17/08/16 17:32, Andy McCright wrote: Sure, please do. FYI, in CXF 3.1.7 we have JAX-RS MessageBodyReader/Writer cache supported something that Neal Hu and myself worked upon. MediaType cache has also been implemented earlier (with thanks to Alessio). Interesting - that is where we have found the most success as well. In 3.0.3 and 3.0.9, we found a lot of time was spent determining the appropriate MBR/MBW for a given request/response. I'll take a look at those changes to see if they could be ported back to the the 3.0.X stream. Rookie question, but what is the major difference between the 3.0.X stream and 3.1.X stream? Does 3.1.X require Java 8 (or higher)? 3.0.x supports Java 6+, 3.1.x - Java 7+, master (3.2.0-SNAPSHOT) - Java 8+ Cheers, Sergey Thanks, Andy On Wed, Aug 17, 2016 at 10:48 AM, Sergey Beryozkin wrote: Hi Andy Welcome, thanks for the introductions, some comments below On 17/08/16 15:56, Andy McCright wrote: Hi All, I wanted to introduce myself. My name is Andy McCright and I'll be taking over Iris Deng's role as IBM's advocate for CXF and JAX-RS. Please pass our thanks to Iris, who is still a CXF committer :-) I'm planning to join the JSR 370 expert group as well. Will be good, the group does need new people joining, we already have Alessio Soldano representing our friends from RedHat (RestEasy) there joining recently too. My background has been mainly in WebSphere development. I've worked on few different areas including the kernel/bootstrap/OSGi implementation, the classloading mechanism, trace and logging, the EJB container, etc. I've taken over the JAX-RS role due to some recent reorganizations. So far, my only involvement with CXF has been internal to IBM. We've been tweaking a few things in our WebSphere Liberty implementation. Our main concern lately has been in the area of performance. We've seen about a 20% degradation of performance when taking a very simple JAX-RS 1.1-compliant app (running with the Apache Wink implementation) and running it on CXF 3.0.3 - using Liberty's jaxrs-2.0 feature. I know this is not an "apples-to-apples" comparison, and I suspect that much of the difference is time spent on function specific to JAX-RS 2.0. That said, we have also seen another ~10% drop in performance from 3.0.3 to 3.0.9. I am hoping to contribute some performance fixes to the community in the near future. I am still pretty new to this code base, so I'll be asking for some help with code reviews and suggestions, etc. Sure, please do. FYI, in CXF 3.1.7 we have JAX-RS MessageBodyReader/Writer cache supported something that Neal Hu and myself worked upon. MediaType cache has also been implemented earlier (with thanks to Alessio). I believe one of your colleagues opened an issue earlier related to a CXF Bus synchronization issue and said they'd evaluate a fix. Looking forward to working with you, Looking forward to working with you too Sergey Andy -- Sergey Beryozkin Talend Community Coders http://coders.talend.com/
Re: Introduction
> Sure, please do. FYI, in CXF 3.1.7 we have JAX-RS MessageBodyReader/Writer cache supported something that Neal Hu and myself worked upon. MediaType cache has also been implemented earlier (with thanks to Alessio). Interesting - that is where we have found the most success as well. In 3.0.3 and 3.0.9, we found a lot of time was spent determining the appropriate MBR/MBW for a given request/response. I'll take a look at those changes to see if they could be ported back to the the 3.0.X stream. Rookie question, but what is the major difference between the 3.0.X stream and 3.1.X stream? Does 3.1.X require Java 8 (or higher)? Thanks, Andy On Wed, Aug 17, 2016 at 10:48 AM, Sergey Beryozkin wrote: > Hi Andy > > Welcome, thanks for the introductions, some comments below > On 17/08/16 15:56, Andy McCright wrote: > >> Hi All, >> >> I wanted to introduce myself. My name is Andy McCright and I'll be taking >> over Iris Deng's role as IBM's advocate for CXF and JAX-RS. >> > > Please pass our thanks to Iris, who is still a CXF committer :-) > > I'm planning >> to join the JSR 370 expert group as well. >> >> Will be good, the group does need new people joining, we already have > Alessio Soldano representing our friends from RedHat (RestEasy) there > joining recently too. > > My background has been mainly in WebSphere development. I've worked on few >> different areas including the kernel/bootstrap/OSGi implementation, the >> classloading mechanism, trace and logging, the EJB container, etc. I've >> taken over the JAX-RS role due to some recent reorganizations. >> >> So far, my only involvement with CXF has been internal to IBM. We've been >> tweaking a few things in our WebSphere Liberty implementation. Our main >> concern lately has been in the area of performance. We've seen about a >> 20% >> degradation of performance when taking a very simple JAX-RS 1.1-compliant >> app (running with the Apache Wink implementation) and running it on CXF >> 3.0.3 - using Liberty's jaxrs-2.0 feature. I know this is not an >> "apples-to-apples" comparison, and I suspect that much of the difference >> is >> time spent on function specific to JAX-RS 2.0. That said, we have also >> seen another ~10% drop in performance from 3.0.3 to 3.0.9. >> >> I am hoping to contribute some performance fixes to the community in the >> near future. I am still pretty new to this code base, so I'll be asking >> for some help with code reviews and suggestions, etc. >> >> Sure, please do. FYI, in CXF 3.1.7 we have JAX-RS > MessageBodyReader/Writer cache supported something that Neal Hu and myself > worked upon. MediaType cache has also been implemented earlier (with thanks > to Alessio). > I believe one of your colleagues opened an issue earlier related to a CXF > Bus synchronization issue and said they'd evaluate a fix. > > > Looking forward to working with you, >> > > Looking forward to working with you too > > Sergey > > >> Andy >> >> >
Re: Introduction
Hi Andy Welcome, thanks for the introductions, some comments below On 17/08/16 15:56, Andy McCright wrote: Hi All, I wanted to introduce myself. My name is Andy McCright and I'll be taking over Iris Deng's role as IBM's advocate for CXF and JAX-RS. Please pass our thanks to Iris, who is still a CXF committer :-) I'm planning to join the JSR 370 expert group as well. Will be good, the group does need new people joining, we already have Alessio Soldano representing our friends from RedHat (RestEasy) there joining recently too. My background has been mainly in WebSphere development. I've worked on few different areas including the kernel/bootstrap/OSGi implementation, the classloading mechanism, trace and logging, the EJB container, etc. I've taken over the JAX-RS role due to some recent reorganizations. So far, my only involvement with CXF has been internal to IBM. We've been tweaking a few things in our WebSphere Liberty implementation. Our main concern lately has been in the area of performance. We've seen about a 20% degradation of performance when taking a very simple JAX-RS 1.1-compliant app (running with the Apache Wink implementation) and running it on CXF 3.0.3 - using Liberty's jaxrs-2.0 feature. I know this is not an "apples-to-apples" comparison, and I suspect that much of the difference is time spent on function specific to JAX-RS 2.0. That said, we have also seen another ~10% drop in performance from 3.0.3 to 3.0.9. I am hoping to contribute some performance fixes to the community in the near future. I am still pretty new to this code base, so I'll be asking for some help with code reviews and suggestions, etc. Sure, please do. FYI, in CXF 3.1.7 we have JAX-RS MessageBodyReader/Writer cache supported something that Neal Hu and myself worked upon. MediaType cache has also been implemented earlier (with thanks to Alessio). I believe one of your colleagues opened an issue earlier related to a CXF Bus synchronization issue and said they'd evaluate a fix. Looking forward to working with you, Looking forward to working with you too Sergey Andy
Introduction
Hi All, I wanted to introduce myself. My name is Andy McCright and I'll be taking over Iris Deng's role as IBM's advocate for CXF and JAX-RS. I'm planning to join the JSR 370 expert group as well. My background has been mainly in WebSphere development. I've worked on few different areas including the kernel/bootstrap/OSGi implementation, the classloading mechanism, trace and logging, the EJB container, etc. I've taken over the JAX-RS role due to some recent reorganizations. So far, my only involvement with CXF has been internal to IBM. We've been tweaking a few things in our WebSphere Liberty implementation. Our main concern lately has been in the area of performance. We've seen about a 20% degradation of performance when taking a very simple JAX-RS 1.1-compliant app (running with the Apache Wink implementation) and running it on CXF 3.0.3 - using Liberty's jaxrs-2.0 feature. I know this is not an "apples-to-apples" comparison, and I suspect that much of the difference is time spent on function specific to JAX-RS 2.0. That said, we have also seen another ~10% drop in performance from 3.0.3 to 3.0.9. I am hoping to contribute some performance fixes to the community in the near future. I am still pretty new to this code base, so I'll be asking for some help with code reviews and suggestions, etc. Looking forward to working with you, Andy