Now, while I think that a portlet.xml which does not use use the
namespace is not valid (wrt the spec), it sounds reasonable to
support such descriptors.
And I also agree that this support makes sense in Pluto.
The only question I have which cases we support :)
The example below mentions: <portlet-app id="charttest" version="1.0">.
This can be supported easily.
Two questions:
- I guess the version attribute is still mandatory, right?
- What happens if someone is using <portlet-app id="charttest"
version="2.0">? Would we support that as well?
Regards
Carsten
Woonsan Ko wrote:
> Hi there,
>
> During testing portlet deployments with some useful PAs from jp.sf.pal
> portlet repository, I found that the current pluto-2 deployment component
> failed to deploy a PA when the descriptor of the PA does not have any
> namespace uri definition with the following exceptions:
>
> java.io.IOException: unexpected element
> (uri:"", local:"portlet-app"). Expected elements are
> <{http://java.sun.com/xml/ns/portlet/portlet-app_1_0.xsd}portlet-app>,<{http://java.sun.com/xml/ns/portlet/portlet-app_2_0.xsd}portlet-app>
>
> The portlet.xml is like the following:
>
> <?xml version="1.0" encoding="UTF-8"?>
> <portlet-app id="charttest" version="1.0">
> <snip/>
> </portlet-app>
>
> It's simply because PortletAppDescriptorServiceImpl is using explicit schema
> based JAXB unmarshalling. (.../portlet-app_1_0.xsd or .../portlet-app_2_0.xsd)
>
> However, if we allow empty namespaced descriptor as well, then I think it
> would be very helpful to deploy some old portlet-1.0 based existing portlet
> applications.
>
> To do this without any side effect, we can just copy the portlet 1.0 JAXB
> package, "o.a.p.container.om.portlet10.impl", to a new package (such as
> "o.a.p.container.om.portlet10.emptyns.impl"?), with empty namespace
> annotations and add it into the JAXB context as a third option in the
> PortletAppDescriptorService implementation.
>
> By the way, this feature has been recently committed in
> Jetspeed-2.2.1-SNAPSHOT, but I'm now thinking it would be better to move this
> to pluto.
> You can see the issue (JS2-1061) and the related discussion here:
> http://www.nabble.com/On-the-current-strong-validation-during-PA-deployment-td25261119.html
>
> If there's no objection, I hope to add this to pluto trunk.
> What do you think?
>
>
> Kind regards,
>
> Woonsan
>
>
>
>
>
--
Carsten Ziegeler
cziege...@apache.org
