[
https://issues.apache.org/jira/browse/JCR-4458?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16918656#comment-16918656
]
Woonsan Ko commented on JCR-4458:
---------------------------------
Hi [~reschke],
Yes, the changes fix the test case #4 in my last comment in JCR-4460: "With
custom context path". And the other changes in RepositoryStubImpl are
improvements to allow servlet path configurations as well.
And the reason why I ruled out the option of "changing/extending APIs in a way
that would avoid the thread local" is that I thought another approach I took in
https://github.com/apache/jackrabbit/pull/84 is hard to understand in
maintenance perspective and is by itself kind of indirect inference of context
path anyway. So, I thought it seems better to directly access and use the
request to determine the context path and easier to maintain..
Regards,
Woonsan
> When JcrRemotingServlet deployed on non-root context, AclResource Webdav
> request fails
> --------------------------------------------------------------------------------------
>
> Key: JCR-4458
> URL: https://issues.apache.org/jira/browse/JCR-4458
> Project: Jackrabbit Content Repository
> Issue Type: Bug
> Affects Versions: 2.18.2
> Reporter: Woonsan Ko
> Assignee: Julian Reschke
> Priority: Major
> Fix For: 2.20
>
> Attachments: JCR-4458.diff
>
>
> If {{org.apache.jackrabbit.server.remoting.davex.JcrRemotingServlet}} is
> configured in a non-root web application, the contextPath of which is "/cms"
> for example with the servletPath, "/server", then
> {{javax.jcr.Session#importXML(...)}} fails from a JCR client based on
> JCR/WebDAV. In other words, {{#importXML(...)}} fails from a JCR {{Session}}
> using a repository which can be created like the following for JCR over
> WebDAV:
> {code}
> String repositoryAddress = "http://localhost:8080/cms/server";;
> Jcr2davRepositoryFactory factory = new Jcr2davRepositoryFactory();
> Map<String, String> params = new HashMap<String, String>();
> params.put(JcrUtils.REPOSITORY_URI, repositoryAddress);
> Repository repository = factory.getRepository(params);
> // ...
> {code}
> It seems like that {{Session#importXML(...)}} call invokes an AclResource
> Webdav request first on the specific resource path, but
> {{org.apache.jackrabbit.webdav.jcr.version.report.JcrPrivilegeReport#init(DavResource,
> ReportInfo)}} does not remove the contextPath, "/cms" for example, when
> determining the resoucrePath.
> Unlike the {{JcrPrivilegeReport}},
> {{org.apache.jackrabbit.webdav.WebdavRequestImpl#getHrefLocator(String,
> boolean)}} seems to remove the contextPath properly.
--
This message was sent by Atlassian Jira
(v8.3.2#803003)
