OASIS Mailing List ArchivesView the OASIS mailing list archive below
or browse/search using MarkMail.

 


Help: OASIS Mailing Lists Help | MarkMail Help

sca-j message

[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]


Subject: RE: [sca-j] Java Callback Simplification Proposal



I'll  resist the temptation to interleave more deeply.

I'm OK with proposals 1, 2 and 3.  For proposal 2, I didn't fully understand the use case involving third parties and mapping callbacks IDs to callback endpoints, but I'm OK with this on simplification grounds.

I'd like to propose a small change to proposal 4.  For the conversational-both-ways case, I think it's better to still mark the callback interface as conversational rather than having this implied from the conversational marking on the forward interface.  I think this is clearer and would allow the callback interface to be used as a regular forward conversational interface, which could somettimes be useful.

I would like to go a step further and eliminate callback IDs in all cases.  I believe this can be done by bringing callbacks and conversations more closely together and using the conversation ID for client-side correlation in cases that are currently treated as non-conversational and where the callback ID is being used.

This approach involves thinking of callbacks as a type of conversation that starts with a forward call and ends when the last callback has been made by the service provider.  This means that all callback interactions would in effect be conversational in both directions (from an interface perspective).  If the client's implementation isn't conversation-scoped, the conversation ID from the callback would be available to the client to use to perform its own correlation, replacing the current use of callbackID for this purpose.

WIth this view of the world, a callback would observe the following pattern:
1. Bidirectional interfaces must be marked as conversational for both the forward and callback interfaces.  The client and service implementations can have any scope.
2. A forward call over a bidirectional interface starts a conversation if it's not already started (this is just the current conversational semantics).
3. All callbacks are part of the same conversation that was started or continued by the forward call.
4. Normal rules for ending this conversation apply.  Typically, the final callback would end the conversation.
5. If the service implementation is conversation-scoped, the infrastructure will automatically perform routing based on the conversation ID passed on the forward call.  Otherwise, the conversation ID is available to the service implementation to perform its own correlation.
6. If the client implementation is conversation-scoped, the infrastructure will automatically perform routing based on the conversation ID passed on the callback.  Otherwise, the conversation ID is available to the client implementation to perform its own correlation.
7. Any correlation that is more complex than the above mechanisms provide must be performed by means of business data passed over forward and callback messages.

Here's an example of how this would work for the simplest "stateless callback" case:
a. Stateless client A makes a forward call F to stateless service B over bidirectional conversational interface X (with callback conversational interface Y).
b. Call F causes the infrastructure to start a conversation and allocate a conversationID C, which is passed on call F.
c. B makes callbacks G and H.  These are routed to new stateless instances of A.  These callbacks carry conversationID C, which can be used by A to perform correlation (in place of callbackID).
d. Call H ends the conversation that was started by call F.  (If this isn't done, the conversation will eventually time out.)

    Simon

Simon C. Nash, IBM Distinguished Engineer
Member of the IBM Academy of Technology
Tel. +44-1962-815156  Fax +44-1962-818999



"Michael Rowley" <mrowley@bea.com>

13/02/2008 14:58

To
Mike Edwards/UK/IBM@IBMGB, "OASIS Java" <sca-j@lists.oasis-open.org>
cc
Subject
RE: [sca-j] Java Callback Simplification Proposal





Mike,
 
I’m glad you liked 3 of the 4 proposed simplifications.   Regarding the one you didn’t like:
 
The proposal:
3. Making ServiceReference.setCallbackID(Object id) illegal and  
ServiceReference.getCallbackID() (assuming #2) return null for  
conversational interactions.

 
Then part of the rationale, and your response to it, were:
...conversations by nature correlate  
invocations between a client and provider instance. If finer-grained  
correlation is needed such as between multiple invocations in the  
same conversation, correlation data can be included as part of the  
service operation data (e.g. “line item number” can be passed as a  
parameter in the example above).

<mje>THIS is where I hear the baby screaming as it flies out with the bathwater.
 
<MR>I don’t know... Claiming we are baby-killers seems a bit severe doesn’t it :-) </MR>

IF you really believe that, then there is no need for a callback ID ever.  All you

do is use correlation data in the messages themselves.  I suggested this to

Michael Rowley not so long ago, with the model of BPEL correlation in mind.  He

was of the opinion that the BPEL correlation approach was a failure and not

to be used as a model for SCA.


There is a good argument to be made for a system managed correlation

mechanism in that it simplifies the life of the programmer.  Forcing the use of

correlation data in the message may well have the consequence of distorting

the form of the messages, since they now *HAVE* to contain data fields that

are "uniquely identifying" - and this must be done for all callback messages.

</mje>

 
<MR>Actually my point on BPEL correlation is a little more subtle.  Having to use correlation to get to the right process instance is indeed too much of a pain.  The system should be able to do that.  However, when you have a complex process and you need to get to the right place within that process, then correlation is OK, and in fact the SCA BPEL spec makes it clear that BPEL correlations can and should continue to be used for this.  Because this is a less-common complex case, it is OK to require some complication in the solution.
 
The situation is analogous for Java.  The infrastructure should be able to route to the right client instance, based on the conversation that is taking place.  If you need finer-grained correlation, that is up to you.
 
Also, it would be difficult or impossible for the system to make any useful guarantees about which callback ID will be received for a conversational interaction.  Imagine this case (pseudocode):
 
{
  setCallbackID(“1”)
  callService()
  setCallbackID(“2”)
  callService()
}
 
receiveCallback() {
getCallbackID();  -- what does this return?
}
 
When multiple callback IDs are set within a single conversation, there is nothing to guarantee that the service provider will keep track of all of the outstanding callback IDs (“1” and “2” in this case) and use the most appropriate one when it is invoking a callback.  Creating such an ability to automatically maintain a set of outstanding callback IDs would be too complex for the value it would provide.
 
Michael
 



From: Mike Edwards [mailto:mike_edwards@uk.ibm.com]
Sent:
Wednesday, February 13, 2008 5:24 AM
To:
OASIS Java
Subject:
Re: [sca-j] Java Callback Simplification Proposal

 

Jim,


Thanks for this detailed proposal.


I've got a few concerns, that I can summarize simply as "the baby being thrown out with the bathwater"

- ie some capabilities that are actually quite important seem to get chopped in the name of simplicity.

I'm left feeling that callbacks in particular are no longer of any use once the proposals are accepted.


Let me try to express what I see as the problems in inline comments as
<mje>...</mje>

Yours,  Mike.

Strategist - Emerging Technologies, SCA & SDO.
Co Chair OASIS SCA Assembly TC.
IBM Hursley Park, Mail Point 146, Winchester, SO21 2JN, Great Britain.
Phone & FAX: +44-1962-818014    Mobile: +44-7802-467431  
Email:  mike_edwards@uk.ibm.com

Jim Marino <jmarino@bea.com>

12/02/2008 18:23


To
OASIS Java <sca-j@lists.oasis-open.org>
cc
 
Subject
[sca-j] Java Callback Simplification Proposal

 


   





All,

Bellow is a proposal from BEA to simplify callback mechanisms in the  
Java C&I specifications. I will work on the appropriate steps of  
getting this into a JIRA so that it can be tracked.

Thanks,
Jim

This proposal aims to simplify using asynchronous communication in  
the SCA Java programming model by:

1. Eliminating callback objects.
  - Removing ServiceReference.get/setCallback()

<mje>So this implies the client MUST implement the callback I/F</mje>



2. Removing the ability for service providers to get callback IDs.
  - Moving CallableReference getCallbackID() to ServiceReference)

<mje>.+1 </mje>


<mje>There is one interesting argument to follow from this.  It is possible to
take the view that callbackID is only of interest to the client, while conversationID

is only really of use to the provider.  Each is being used to pick out state of

some kind, but at opposite ends of the communication. So perhaps we can

eliminate the client having access to the conversation ID?


It would be nice to coalesce the 2 IDs,  The problem is that they have completely

different lifecycles.  The conversationID has some fixed span dictated by the

form of the forward call interface, with specific starting and ending operations.


The callback ID has an indeterminate span, always starting with one forward

call, but with an undetermined number and type of callback operations
being included in its span. Logically it needs to change for each subsequent

forward call.


I'm now really heading off into the weeds, but it occurs to me that there is a
relationship of the callback ID to the concept of "Future" that is used in the Java

concurrency packages.  There, the Future represents the asychronous operation

that is taking place, and can be used to feed back results. The Future is handed

back to the "client" application on the invocation of the forward "service" invocation.


Today, Futures only handle one-shot pieces of asychronous work, with a single

"response" message.  However, you could conceive of stretching this, to permit

something you might call a "CallbackFuture" - which would represent an asynchronous

process with a continuing life and an arbitrary number of "response" messages.

The nasty thing about this is that with our current callback design there is no

obvious end to the life of a CallbackFuture - in other words, the callback interface

does not define the equivalent of an "endsConversation" which indicates completion

of the work of the asynchronous service.  Today, the "end" of the callbacks must

be defined in metadata outside the interface.  Perhaps we need to fix that,

but it will not be easy.</mje>



3. Saying callback IDs are irrelevant in conversations.
  - Making ServiceReference.setCallbackID(Object id) illegal and
    ServiceReference.getCallbackID() (assuming #2) return null for
    conversational interactions.

<mje>Why does having a conversation enable this to be done?

The callback ID I thought allowed a client to mark a service invocation in

a way that callbacks made to the client carried the same ID and that

this enables the client to establish which original request caused a

given response.  


Conversation ID does not do this at all, unless I am mistaken, since

while callback ID can change from one invocation to the next, conversation ID

stays the same for the whole of a conversation.  Thus conversation ID is

useless for distinguishing one callback from another.</mje>



4. Removing the distinction between stateless and conversational
  callback interfaces.
   - Removing the ability to specify conversational on callback
     Interfaces.

<mje>+1</mje>



By simplification, we mean as a primarily reducing the number of  
concepts and APIs a developer must understand to program  
asynchronous  services. This will entail some reduction in  
capability. However,  this reduction, we argue, affects edge-cases  
that can be accommodated through alternative techniques. A secondary  
aspect of simplification concerns implementation. By removing the  
requirement for runtimes to provide infrastructure for  handling edge  
cases that can be accommodated through relatively simple application  
programming techniques,implementing callback capabilities becomes  
fairly straightforward.

<mje>Unfortunately, unless I'm mistaken, some rather central use cases have been

trashed too.</mje>



[RATIONAL]

The rational for proposed items is dealt with individually in this  
section.

1. ServiceReference.setCallback(Object object) and  
ServiceReference.getCallback()
------------------------------------------------------------------------
---------------------------------------------

<mje>+1</mje>


This API exists as a mechanism to route callbacks to instances other  
than an instance of the client component making a forward invocation.  
This provides two capabilities.  Clients do not have to implement a  
callback interface. The API also allows clients to instruct the  
runtime to perform callback dispatching for stateless interactions  
that execute concurrently. A common example of this is inventory  
check. A client component needs to perform an inventory check for  
multiple line items. As an efficiency, it written to invoke the same
inventory service using a non-blocking call multiple times. For each  
invocation, it wishes to dispatch the associated callback to a  
different handler instance it creates.  For example:

ServiceReference<InventoryService> serviceReference = ..

for (LineItem item: purchaseOrder.getLineItems()) {
   CallbackService handler = new CallbackServiceImpl();
   serviceReference.setCallback(handler)
   serviceReference.getService().checkInventory(item);
}

This can be done using the CallableReference.getCallbackID():

// code that conducts the forward invoke:

ServiceReference<InventoryService> serviceReference = ..

for (LineItem item: purchaseOrder.getLineItems()) {
   String id = // generate or assign the callback ID
   // store the callback id and whatever information the callback  
needs
   cache.put(id, handler);
   serviceReference.getService().checkInventory(item);
}

Callback processing can either be done directly by a component  
implementation instance or delegated to CallbackServiceImpl as shown  
below:

// code in the same class that handles the callback:
public void onCallback(InventoryStatus status) {
   String id = requestContext.getServiceReference().getCallbackID();
   CallbackInfo info = .. // retrieve the stored callback
information by using the callback id
   CallbackService handler = new CallbackServiceImpl(info);
   handler.onCallback(status);
}

Some advantages to eliminating get/set/Callback() are:

a. Reduces the number of concepts and APIs a developer will be  
presented with without affecting common functionality. In other  
words, this API is primarily used for edge-cases and removing it  
simplifies the common case.

b. Application migrations and versioning are easier using the  
alternative approach. By using an object instance, runtime  
implementations would be responsible for handling class versioning
for in-process interactions. By avoiding serialization and the use of  
a specific classes, and instead storing callback data,  
CallbackServiceImpl can be versioned and substituted for callbacks  
that have not yet occurred.

c. Runtime performance characteristics are likely to improve. For  
cases where the client is stateless, routing to the specific instance  
set through setCallback() requires that the callback be routed to the  
specific JVM where the instance is hosted. Eliminating routing to a  
particular instance allows the callback to potentially be sent to any  
JVM where the client component is hosted.

d. Eliminates the need for runtime implementations to manage garbage  
collection of callback objects. For example, if a callback is never  
made, the runtime at some point will need to remove the object from  
storage or memory.


2. Moving CallableReference.getCallbackID() to ServiceReference
------------------------------------------------------------------------
--------------------

<mje>+1</mje>


We propose removing getCallbackID() from access by service providers  
as it is cannot be used for correlation in third-party components  
except when the third-party is configured with a callback to the same  
client component. For example in the stateless case:

A---------B--------C

The only way C can callback to A is either by calling back B or  
having B pass a CallableReference to C.  The callback ID is not  
sufficient, in itself, to be used by arbitrary code to send a message  
that qualifies as a callback in this bidirectional exchange.  For  
example, if B has clients other than A, the callback ID doesn’t have  
the information to tell which client it is for.


Moving CallableReference.getCallbackID() to ServiceReference  
simplifies the Java programming model by:

a. Eliminating an API that can only be used in very particular  
circumstances.
b. Removing the need for a runtime implementation to create and  
maintain a distributed domain-wide mapping of callback ids to  
callback endpoints. Note that routing information cannot be encoded  
in the callback id as it can be set by the application.


3. Making ServiceReference.setCallbackID(Object id) illegal and  
ServiceReference.getCallbackID() (assuming #2) return null for  
conversational interactions.
------------------------------------------------------------------------
------------------------------------------------------------------------
----------------------------------------------------------------------
Having both callback ids and conversation ids for bi-directional  
conversational wires is confusing and not needed as the conversation  
id will be used for correlation.


<mje>This is what I challenge - it isn't true if you need to correlate callback responses

to individual requests.  Conversation ID is necessarily the same for all the forward

invocations in a conversation.  This then is useless for distinguishing between
callback messages resulting from different forward invocations.


In fact, I argue that conversationID is almost useless for the client - it's really a
provider-defined thing useful for the provider to access its state data.</mje>



Sometimes a client may wish to  
correlate callback invocations using the callback id. For  
conversational callbacks, correlation using the callback id is  
limited. Since conversations are serial between a particular client  
instance and a provider instance, the additional correlation that may  
be needed is between different invocations to the provider within the  
same conversation. In other words, conversations by nature correlate  
invocations between a client and provider instance. If finer-grained  
correlation is needed such as between multiple invocations in the  
same conversation, correlation data can be included as part of the  
service operation data (e.g. “line item number” can be passed as a  
parameter in the example above).


<mje>THIS is where I hear the baby screaming as it flies out with the bathwater.


IF you really believe that, then there is no need for a callback ID ever.  All you

do is use correlation data in the messages themselves.  I suggested this to

Michael Rowley not so long ago, with the model of BPEL correlation in mind.  He

was of the opinion that the BPEL correlation approach was a failure and not

to be used as a model for SCA.


There is a good argument to be made for a system managed correlation

mechanism in that it simplifies the life of the programmer.  Forcing the use of

correlation data in the message may well have the consequence of distorting

the form of the messages, since they now *HAVE* to contain data fields that

are "uniquely identifying" - and this must be done for all callback messages.

</mje>



This simplifies the Java programming model by clarifying the  
relationship between callback and conversation ids.


4. Removing the ability to specify stateless or conversational  
callback service contracts
------------------------------------------------------------------------
-----------------------------------------------------

<mje>+1</mje>



We propose wires be either conversational or stateless in both  
directions and that it be specified as part of the forward service  
contract. In other words, @Conversational cannot be specified on the  
callback interface definition.  We don't believe having a stateless  
callback for a stateful forward invocation and vice versa makes much  
sense.

---------------------------------------------------------------------
To unsubscribe from this mail list, you must leave the OASIS TC that
generates this mail.  You may a link to this group and all your TCs in OASIS
at:
https://www.oasis-open.org/apps/org/workgroup/portal/my_workgroups.php






 

Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU











Unless stated otherwise above:
IBM United Kingdom Limited - Registered in England and Wales with number 741598.
Registered office: PO Box 41, North Harbour, Portsmouth, Hampshire PO6 3AU








[Date Prev] | [Thread Prev] | [Thread Next] | [Date Next] -- [Date Index] | [Thread Index] | [List Home]