On 8/09/2015 11:26 PM, Greg Trasuk wrote:
That’s the current state.  Changing (e.g. by enforcing a builder pattern or 
something) would add unneeded complexity for the user if you ask me.

Yes, I agree, a constructor is suitable, keep it simple.

In fact, the fact that the Entry fields are non-final is the way Jini knows it’s a field 
and not a constant.  i.e. in the AbstractEntry docs, it specifically says "The entry 
fields of an Entry are its public, non-primitive, non-static, non-transient, non-final 
fields."

There's no code that inspects the field and checks whether it's final, not within River that I'm aware of, feel free to show me where if I'm wrong :). The reason these fields are non final, is so they can be set with reflection, this is a detail of the Entry specification we should look at changing.

Entry's are not subject to the usual serialization rules. All fields in an Entry in superclass to subclass order are stored in an EntryRep (appended). At present these are set after construction using reflection, however a constructor that accepts an array parameter, will allow the child most class to pass that array up through all constructors to reconstruct the Entry without using reflection (magnitudes faster performance wise).

Eliminating the use of reflection during deserialization will increase performance, and immutability increases scalability.

Where do we use entries?  Typically in doing Service registrations and lookups. 
 Most users can understand that the Entry is going to be serialized and used 
remotely.
Clients of Javaspaces and utility classes too, like JoinManager and ServiceDiscoveryManager.
Add to documentation?  Sure.  Change the API?  Maybe add an optional 
builder-style object.  (e.g. create a StatusBuilder that acts as a bean and 
generates Status entries).  But even so, it seems like a small enough part of 
using Jini to not bother much with.

Due to the way the current Entry spec works, you can't add a field to an Entry without breaking compatibility with subclassses.

So the proposed change would be a new public constructor, that accepts an array, containing field object values in the order that fields occur and that all fields be final.

I'd also propose including a warning that once published the order and number of fields in an Entry should not be changed if it can be extended, otherwise if there's a chance that additional fields might need to be appended at a later date, a reccommendation that the class be made final, so it can't be subclassed.

So in other words, the following changes will break an Entry:

  1. A change in the class heirarchy.
  2. A change in the order or number or type of fields.

An Entry is best thought of an interface definition for a defined group of objects in serial form.



Another way to think about it is that Entries don’t mean much until you send 
them somewhere.  So you need to make sure they’re all setup and everything they 
refer to is stable when you send them.  To me, that’s a user-understanding item 
that’s difficult to enforce through code.

But because Entry fields are mutable and unsynchronized, and used in utility classes such as ServiceDiscoveryManager, it becomes difficult to manage when Entry's are passed around between threads. To submit an Entry to a service, it is passed to a thread pool where it becomes a serialized method invocation.

To ensure the changes made by one thread are visible to another, the Entry must be published safely, otherwise the changes can't be guaranteed visible between threads. There is no synchronization on Entry field access, so this makes it very difficult to reason about an Entry, one strategy I've adopted is defensive copying, however all this does, is guarantee that any modifications made by client code, doesn't affect a copy in ServiceDiscoveryManager (for example), there is a memory usage cost, because internal copies can't be shared with client ServiceItemFilter's.

Reasoning about shared state becomes much simpler if Entry's are immutable and the upside is scalability and performance improves, especially for clients of lookup services deserializing a lot of Entry objects.

/**
 * An EntryRep contains the fields of an Entry packaged up for
 * transmission between client-side proxies and the registrar server.
 * Instances are never visible to clients, they are private to the
 * communication between the proxies and the server.
 * <p>
 * This class only has a bare minimum of methods, to minimize
 * the amount of code downloaded into clients.
 *
 * @author Sun Microsystems, Inc.
 *
 */
class EntryRep implements Serializable, Cloneable {

    private static final long serialVersionUID = 2L;

    /**
     * The Class of the Entry converted to EntryClass.
     *
     * @serial
     */
    public EntryClass eclass;
    /**
     * The codebase of the entry class.
     *
     * @serial
     */
    public String codebase;
    /**
     * The public fields of the Entry, each converted as necessary to
     * a MarshalledWrapper (or left as is if of known java.lang immutable
     * type).  The fields are in super- to subclass order.
     *
     * @serial
     */
    public Object[] fields;

    /**
     * Converts an Entry to an EntryRep.  Any exception that results
     * is bundled up into a MarshalException.
     */
    public EntryRep(Entry entry) throws RemoteException {
    EntryClassBase ecb = ClassMapper.toEntryClassBase(entry.getClass());
    eclass = ecb.eclass;
    codebase = ecb.codebase;
    try {
        EntryField[] efields = ClassMapper.getFields(entry.getClass());
        fields = new Object[efields.length];
        for (int i = efields.length; --i >= 0; ) {
        EntryField f = efields[i];
        Object val = f.field.get(entry);
        if (f.marshal && val != null)
            val = new MarshalledWrapper(val);
        fields[i] = val;
        }
    } catch (IOException e) {
        throw new MarshalException("error marshalling arguments", e);
    } catch (IllegalAccessException e) {
        throw new MarshalException("error marshalling arguments", e);
    }
    }

    /**
     * Convert back to an Entry.  If the Entry cannot be constructed,
     * null is returned.  If a field cannot be unmarshalled, it is set
     * to null.
     */
    public Entry get() {
    try {
        Class clazz = eclass.toClass(codebase);
        EntryField[] efields = ClassMapper.getFields(clazz);
        Entry entry = (Entry)clazz.newInstance();
        for (int i = efields.length; --i >= 0; ) {
        Object val = fields[i];
        EntryField f = efields[i];
        Field rf = f.field;
        try {
            if (f.marshal && val != null)
            val = ((MarshalledWrapper) val).get();
            rf.set(entry, val);
        } catch (Throwable e) {
            if (e instanceof IllegalArgumentException) {
            // fix 4872566: work around empty exception message
            String msg = "unable to assign " +
                ((val != null) ?
                "value of type " + val.getClass().getName() :
                "null") +
                " to field " + rf.getDeclaringClass().getName() +
                "." + rf.getName() + " of type " +
                rf.getType().getName();
            e = new ClassCastException(msg).initCause(e);
            }
            RegistrarProxy.handleException(e);
        }
        }
        return entry;
    } catch (Throwable e) {
        RegistrarProxy.handleException(e);
    }
    return null;
    }

    /**
     * We don't need this in the client or the server, but since we
     * redefine equals we provide a minimal hashCode that works.
     */
    public int hashCode() {
    return eclass.hashCode();
    }

    /**
     * EntryReps are equal if they have the same class and the fields
     * are pairwise equal.  This is really only needed in the server,
     * but it's very convenient to have here.
     */
    public boolean equals(Object obj) {
    if (obj instanceof EntryRep) {
        EntryRep entry = (EntryRep)obj;
        if (!eclass.equals(entry.eclass) ||
        fields.length != entry.fields.length)
        return false;
        for (int i = fields.length; --i >= 0; ) {
        if ((fields[i] == null && entry.fields[i] != null) ||
            (fields[i] != null && !fields[i].equals(entry.fields[i])))
            return false;
        }
        return true;
    }
    return false;
    }

    /**
     * Deep clone (which just means cloning the fields array too).
     * This is really only needed in the server, but it's very
     * convenient to have here.
     */
    public Object clone() {
    try {
        EntryRep entry = (EntryRep)super.clone();
        entry.fields = (Object[])entry.fields.clone();
        return entry;
    } catch (CloneNotSupportedException e) {
        throw new InternalError();
    }
    }

    /**
     * Converts an array of Entry to an array of EntryRep.  If needCodebase
     * is false, then the codebase of every EntryRep will be null.
     */
public static EntryRep[] toEntryRep(Entry[] entries, boolean needCodebase)
    throws RemoteException
    {
    EntryRep[] reps = null;
    if (entries != null) {
        reps = new EntryRep[entries.length];
        for (int i = entries.length; --i >= 0; ) {
        if (entries[i] != null) {
            reps[i] = new EntryRep(entries[i]);
            if (!needCodebase)
            reps[i].codebase = null;
        }
        }
    }
    return reps;
    }

    /** Converts an array of EntryRep to an array of Entry. */
    public static Entry[] toEntry(EntryRep[] reps) {
    Entry[] entries = null;
    if (reps != null) {
        entries = new Entry[reps.length];
        for (int i = reps.length; --i >= 0; ) {
        entries[i] = reps[i].get();
        }
    }
    return entries;
    }
}

Reply via email to