[icedtea-web] RFC: PR618 - Can't install OpenDJ, JavaWebStart fails with Input stream is null error.

[Dr Andrew John Hughes]

On 14:42 Tue 25 Jan     , Omair Majid wrote:
> Hi,
> 
> The attached patch fixes PR618. The issue is that jars marked as lazy 
> are not currently searched for resources other than classes. So loading 
> classes from lazy jars works, but loading anything else (zip files in 
> the case of the bug report) fails.
> 
> The attached patch fixes findResources to use lazy jars if needed.
> 
> URLClassLoader invokes findResource() as a part of its getResource() 
> implementation. Instead of duplicating the addition of lazy jars in 
> getResource(), getResource() was removed, and findResource added 
> instead. findResource() now delegates to findResources(), so there is 
> only one place where an actual search for resources is performed.
> 
> Any concerns or comments?
> 
> Cheers,
> Omair

> diff -r 64da2a80df88 netx/net/sourceforge/jnlp/runtime/JNLPClassLoader.java
> --- a/netx/net/sourceforge/jnlp/runtime/JNLPClassLoader.java	Tue Jan 25 10:19:20 2011 -0500
> +++ b/netx/net/sourceforge/jnlp/runtime/JNLPClassLoader.java	Tue Jan 25 14:42:07 2011 -0500
> @@ -39,6 +39,7 @@
>  import java.util.LinkedList;
>  import java.util.List;
>  import java.util.Map;
> +import java.util.NoSuchElementException;
>  import java.util.Random;
>  import java.util.TreeSet;
>  import java.util.Vector;
> @@ -1086,27 +1087,42 @@
>       * Finds the resource in this, the parent, or the extension
>       * class loaders.
>       */
> -    public URL getResource(String name) {
> -        URL result = super.getResource(name);
> -
> -        for (int i = 1; i < loaders.length; i++)
> -            if (result == null)
> -                result = loaders[i].getResource(name);
> -
> -        return result;
> +    @Override
> +    public URL findResource(String name) {
> +        try {
> +            return findResources(name).nextElement();
> +        } catch (NoSuchElementException e) {
> +            return null;
> +        } catch (IOException e) {
> +            return null;
> +        }

Would it be more efficient to check if there are more elements first
rather than throwing and catching an exception?

>      }
>  
>      /**
> -     * Finds the resource in this, the parent, or the extension
> +     * Finds the resources in this, the parent, or the extension
>       * class loaders.
>       */
>      @Override
>      public Enumeration<URL> findResources(String name) throws IOException {
> +
> +        List<URL> resources = findResourcesBySearching(name);
> +
> +        // if not found, load all lazy resources; repeat search
> +        while (resources.size() == 0 && addNextResource() != null) {
> +            resources = findResourcesBySearching(name);
> +        }
> +
> +        return Collections.enumeration(resources);
> +    }
> +
> +    private List<URL> findResourcesBySearching(String name) throws IOException {
>          Vector<URL> resources = new Vector<URL>();
>  
>          for (int i = 0; i < loaders.length; i++) {
>              Enumeration<URL> e;
> -
> +            // TODO check if this will blow up or not
> +            // if loaders[1].getResource() is called, wont it call getResource() on
> +            // the original caller? infinite recursion?
>              if (loaders[i] == this)
>                  e = super.findResources(name);
>              else
> @@ -1116,7 +1132,7 @@
>                  resources.add(e.nextElement());
>          }
>  
> -        return resources.elements();
> +        return resources;
>      }

Is findResources the only consumer of the resources collection?  If so, it would be
more efficient to use an ArrayList or LinkedList and avoid the implicit synchronisation
of Vector.

What is the reason for returning a List from findResourcesBySearching rather than just
an Iterator/Enumeration?  AFAICS, you just create an enumeration from it anyway and
hasNext()/hasNextElement() would tell you if there are no entries.

>  
>      /**


-- 
Andrew :)

Free Java Software Engineer
Red Hat, Inc. (http://www.redhat.com)

Support Free Java!
Contribute to GNU Classpath and IcedTea
http://www.gnu.org/software/classpath
http://icedtea.classpath.org
PGP Key: 94EFD9D8 (http://subkeys.pgp.net)
Fingerprint = F8EF F1EA 401E 2E60 15FA  7927 142C 2591 94EF D9D8