On Thursday 26 April 2012 06:53 PM, Laird Nelson wrote:
> On Thu, Apr 26, 2012 at 4:00 AM, Sahoo <sanjeeb.sahoo_at_oracle.com
> <mailto:sanjeeb.sahoo_at_oracle.com>> wrote:
>
> Didn't you notice the group id of the artifact which had version
> number 3.1.1? It must be org.glassfish. There is no universal API
> jar produced for servlet spec. The API classes are a part of
> implementation of the spec. So, we at glassfish have produced api
> jars that exclusively belong to our namespace, compiled the way we
> like. They match version number of our project as they are all
> released together.
>
>
> Sure, just as JBoss and Geronimo (and others) have done.
>
> Having said that some of us believe that since we are the Java EE
> RI, we can claim namespaces like javax.blah.blah and produce jars
> like javax.servlet:javax.servlet:3.0.
>
>
> Yes, that was what I was hoping. I was hoping that these version
> numbers or at least the versioning scheme would follow a predictable
> pattern.
I was the one who had changed most of our API jars to use GlassFish
project specific groupId and version matching the release train they are
part of. They definitely gave us a predictable pattern. The reason I had
done so is that in my opinion, RI for a particular javax API can change
over time and I didn't think maven central checked to make sure only RI
project can use javax groupId. I still believe that unless there is
stringent enforcement by maven central about who can release an artifact
with javax.something as groupId, it is incorrect to use them. This is my
personal opinion.
Sahoo