Javax.servlet Module

[Carsten Ziegeler <cz...@apache.org>]

I'm still wondering if we need our own version of the javax.servlet
implementation. I know that the spec mandates 2.1.

We just ran into compilation problems because of two different
servlet-apis on the classpath. This is due to the problem that our
version uses "org.apache.felix" as the group id whereas the servlet api
uses "javax.servlet". Therefore mvn can't tell that this is the same
library just with different versions. So as we use 2.3 features of the
servlet-api and as maven wanted to put the 2.1 version before the 2.3
version in the classpath, compilation failed.

Now, of course, it's possible to exclude this dependency here and there,
but it has to be excluded nearly with every felix artifact and is a nice
pitfall.

So, can't we just get rid off this? Or can we at least use the groupId
javax.servlet to make things easier?

Carsten
-- 
Carsten Ziegeler
cziegeler@apache.org

Re: Javax.servlet Module

["Richard S. Hall" <he...@ungoverned.org>]

Carsten Ziegeler wrote:
> I'm still wondering if we need our own version of the javax.servlet
> implementation. I know that the spec mandates 2.1.
>
> We just ran into compilation problems because of two different
> servlet-apis on the classpath. This is due to the problem that our
> version uses "org.apache.felix" as the group id whereas the servlet api
> uses "javax.servlet". Therefore mvn can't tell that this is the same
> library just with different versions. So as we use 2.3 features of the
> servlet-api and as maven wanted to put the 2.1 version before the 2.3
> version in the classpath, compilation failed.
>
> Now, of course, it's possible to exclude this dependency here and there,
> but it has to be excluded nearly with every felix artifact and is a nice
> pitfall.
>
> So, can't we just get rid off this? Or can we at least use the groupId
> javax.servlet to make things easier?
>   

Hmm. I was thinking the opposite...changing 2.1 to use 
"org.apache.felix" for its groupId...

I think we could go in either direction, however.

-> richard