On 09/17/2012 07:16 AM, Jos Snellings wrote:
Hi Thorsten,

I believe having encountered this problem once.
However, it is not systematic.

Hi Joe,

I think I pinned down the problem.

We use java.net.urlHandler which means we only get init once. I am ATM reviewing the servlet handler and here we do the resolving part in the Connection part.

I think the blockcontextHandler needs to loose the this.blockcontext part, this would need to be resolved in the actual BlockContextURLConnection to resolve against only for the current servlet specific registered blocks.

salu2


Jos

On Mon, Sep 17, 2012 at 12:06 AM, Thorsten Scherler (JIRA) <j...@apache.org <mailto:j...@apache.org>> wrote:


        [
    
https://issues.apache.org/jira/browse/COCOON3-105?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13456689#comment-13456689
    ]

    Thorsten Scherler commented on COCOON3-105:
    -------------------------------------------

    The problem lies in in
    
org.apache.cocoon.blockdeployment.BlockContextURLStreamHandlerFactory.createURLStreamHandler
    here we create a  BlockContextURLStreamHandler extends
    URLStreamHandler

    Regarding URLStreamHandler
    /**
     * The abstract class <code>URLStreamHandler</code> is the common
     * superclass for all stream protocol handlers. A stream protocol
     * handler knows how to make a connection for a particular protocol
     * type, such as <code>http</code>, <code>ftp</code>, or
     * <code>gopher</code>.
     * <p>
     * In most cases, an instance of a <code>URLStreamHandler</code>
     * subclass is not created directly by an application. Rather, the
     * first time a protocol name is encountered when constructing a
     * <code>URL</code>, the appropriate stream protocol handler is
     * automatically loaded.*/

    Which is exactly the problem in our case. Once the
    URLStreamHandler is setup for the first blockcontext as protocol
    the second servlet will directly use the java.net.URLStreamHandler
    and use that protocol.

    Meaning if you start tomcat with both apps deployed the first
    request of the first app decides which blockcontext will be
    associate with the block context protocol

    That happens in BlockContextURLStreamHandlerFactory where
    createURLStreamHandler is only called once with the first request
    the second then uses the old block context.


    > webapp fails if on the same servlet container is a c2.2.1 or
    other c3 webapp running
    >
    
-------------------------------------------------------------------------------------
    >
    >                 Key: COCOON3-105
    >                 URL:
    https://issues.apache.org/jira/browse/COCOON3-105
    >             Project: Cocoon 3
    >          Issue Type: Bug
    >          Components: cocoon-webapp
    >    Affects Versions: 3.0.0-beta-1
    >            Reporter: Thorsten Scherler
    >            Priority: Blocker
    >
    > I noticed that you cannot run 2 c3 based war in a tomcat.
    > To reproduce:
    > - seed parent via archetype
    > - seed block in parent via archetype
    > - seed block2 in parent via archetype
    > - seed webapp in parent via archetype
    > - seed webapp2 in parent via archetype
    > where webapp depends on block one and webapp2 depends on block2.
    > My sample was:
    > [INFO] Reactor Summary:
    > [INFO]
    > [INFO] myparent ..........................................
    SUCCESS [1.163s]
    > [INFO] myblock ...........................................
    SUCCESS [3.611s]
    > [INFO] mywebapp ..........................................
    SUCCESS [1.924s]
    > [INFO] myblock2 ..........................................
    SUCCESS [1.498s]
    > [INFO] mywebapp2 .........................................
    SUCCESS [1.230s]
    > [INFO]
    ------------------------------------------------------------------------
    > Now take a tomcat (I used 6) and first deploy the mywebapp. You
    can copy it before you start to webapp or if you have it enable
    deploy it on a running instance. You should see the welcome page
    under something like http://localhost:8080/mywebapp-1.0-SNAPSHOT/
    > side note: http://localhost:8080/mywebapp-1.0-SNAPSHOT will
    throw a StringIndexOutOfBoundsException but that is another ticket
    I guess.
    > Now if you deploy the second webapp on a running instance it
    will deploy without problem but requesting
    > http://localhost:8080/mywebapp2-1.0-SNAPSHOT/
    > will return a blank page and in
    > /.../tomcat/work/Catalina/localhost/mywebapp-1.0-SNAPSHOT/cocoon.log
    > you find:
    > 2012-09-13 22:12:46,056 ERROR http-8080-1
    org.apache.cocoon.servlet.XMLSitemapServlet - Can't initialize the
    RequestProcessor correctly.
    >
    org.apache.cocoon.sitemap.SitemapBuilder$SitemapBuilderException:
    Can't build sitemap from blockcontext:/myblock2/sitemap.xmap
    >       at
    org.apache.cocoon.sitemap.SitemapBuilder.build(SitemapBuilder.java:70)
    ~[cocoon-sitemap-3.0.0-beta-1-SNAPSHOT.jar:3.0.0-beta-1-SNAPSHOT]
    >       at
    
org.apache.cocoon.servlet.RequestProcessor.initializeSitemap(RequestProcessor.java:203)
    ~[cocoon-servlet-3.0.0-beta-1-SNAPSHOT.jar:3.0.0-beta-1-SNAPSHOT]
    > ...
    > Caused by: java.lang.RuntimeException: There is no block
    'myblock2' deployed. The available blocks are
    
{myblock=file:/home/thorsten/src/apache/apache-tomcat-6.0.20/work/Catalina/localhost/mywebapp-1.0-SNAPSHOT/blocks/myblock/}.
    >       at
    
org.apache.cocoon.blockdeployment.BlockContextURLConnection.getConnection(BlockContextURLConnection.java:76)
    ~[cocoon-block-deployment-1.2.1.jar:1.2.1]
    >       at
    
org.apache.cocoon.blockdeployment.BlockContextURLConnection.getInputStream(BlockContextURLConnection.java:56)
    ~[cocoon-block-deployment-1.2.1.jar:1.2.1]
    >       at java.net.URL.openStream(URL.java:1010) ~[na:1.6.0_30]
    >       at
    org.apache.cocoon.sitemap.SitemapBuilder.build(SitemapBuilder.java:65)
    ~[cocoon-sitemap-3.0.0-beta-1-SNAPSHOT.jar:3.0.0-beta-1-SNAPSHOT]
    >       ... 46 common frames omitted
    > As you see the blockcontext from the 2nd app is the one from the
    first EVEN if they are deployed as 2 different webapps!
    > Now stop the tomcat and start again.
    > Depending which app you request on a fresh stared tomcat that
    one will work the other will present a blank page and the log will
    say something like:
    > Caused by: java.lang.RuntimeException: There is no block
    'myblock' deployed. The available blocks are
    
{myblock2=file:/home/thorsten/src/apache/apache-tomcat-6.0.20/work/Catalina/localhost/mywebapp2-1.0-SNAPSHOT/blocks/myblock2/}.
    > In this case I requested the 2nd first.
    > Originally I found out because we have a client that has some c3
    and a c2.2.1 app (not) running aside. So in case you create a
    2.2.1 webapp from the archetype as described
    http://cocoon.apache.org/2.2/1159_1_1.html and use it instead of
    the c3 2nd webapp you will get similar results.
    > If you start first with the 1st c3 and then deploy the c2.2 on
    the run then you can actually see both working ONLY if you first
    request the c3 and then deploy and then see the c2. In case you do
    not request the c3 prior it will not work once you requested the
    c2 (which maybe present interesting for the cause of the problem).
    > Now shutdown and start with both deployed the c2.2 always works
    and the c3 not.
    > I see the problem for our client coming when we introduced
    >
    
<listener-class>org.apache.cocoon.blockdeployment.BlockDeploymentServletContextListener</listener-class>
    > The main observation is that the c2 one seems to much more
    presistence but that can come the way of invocation (on-demand vs.
    startup). Anyway the blockcontext should never be shared between
    two different servlets.

    --
    This message is automatically generated by JIRA.
    If you think it was sent incorrectly, please contact your JIRA
    administrators
    For more information on JIRA, see:
    http://www.atlassian.com/software/jira




--
The doctrine of human equality reposes on this: that there is no man
really clever who has not found that he is stupid.
        -- Gilbert K. Chesterson




--
Thorsten Scherler <scherler.at.gmail.com>
codeBusters S.L. - web based systems
<consulting, training and solutions>

http://www.codebusters.es/

Reply via email to