Issues for Notification Service RTF mailing list

List of issues (green=resolved, yellow=pending Board vote, red=unresolved)

List options: All ; Open Issues only; or Closed Issues only

Issue 2153: CosNotifyChannelAdmin::ConsumerAdmin::obtain_notification_X_supplier oper
Issue 2460: EventTypeRepository class issue
Issue 2501: Premature in eliminating Txn interfaces?
Issue 2512: suspend_connection/resume_connection messages - issue
Issue 3175: Notification service: admin MyChannel attribute
Issue 3517: Section 3.4.17. Page:136
Issue 3518: Issue with pushing events
Issue 3519: The interface on EventChanel does not have the ability to get the ChannelId
Issue 3521: The Supplier Proxy & Consumer Proxy interfaces don't get ID's
Issue 6343: Notification Service IDL issues

Issue 2153: CosNotifyChannelAdmin::ConsumerAdmin::obtain_notification_X_supplier oper (notif-service-rtf)

Summary: The CosNotifyChannelAdmin::ConsumerAdmin::obtain_notification_X_supplier
 operation, when asked for a proxy for a ClientType of ANY_EVENT,
 returns a proxy that is unuasable by event service clients.
 
 The reason is that the CosNotifyChannelAdmin::ProxyXSupplier interface
 returned inherits only from CosEventChannelAdmin::XSupplier, and then
 redefines the connect operation that was available in CosEventChannelAdmin::
 ProxyXSupplier as connect_any_x_supplier. The interface can never be
 narrowed to CosEventChannelAdmin::ProxyXSupplier, and therfore is
 unusable by the Event Service style client.
 
 

:

Summary: In Appendix A it is stated that "the EventTypeRepository class only has
 one instance in any given implementation of the Repository". At the same
 time, there is a package interface and a package factory that implies
 that it"s possible to create multiple repositories (unless the
 NotificationTypesPackage interface is expected to always returns the
 same objects, no matter how many packages one have created).
 

Summary: I feel there is a flaw in the model described in the current Notification 
 Service spec for supporting Transactional Notification.
 

Summary: 
  The  suspend_connection/resume_connection messages do not describe what should happen  to messages that are queued up.  Are new messages rejected for the duration  between the disconect and resume?  Are messages that were queued up  flushed?  How does the Persistent Event Reliablity QoS effect the way both  are handled? size=2>  Along  the same lines, what happens to events when a consumer disconnects?  I  assume events in this case would be lost.

The notification service administration objects support an attribute
called MyChannel that returns the event channel that first created the
administration object. The value type of this attribute is EventChannel.
For typed administration objects, the typed channel can not be returned
because a TypedEventChannel is not compatible with an EventChannel. One
solution is to change the value type of this attribute to a generic
Object, which can be narrowed to the correct type.

I see the following features lacking in CosNotification spec.  What is the process on having this raised as 
an issue, and if all agree add it to the spec in the later revisions.  Any info is appreciated.

1. Section 3.4.17.  Page:136. The EventChannel interface in CosNotification, does not allow to retrieve
its ChannelId.  This is unlike the interface on the admins which allow to get the adminids?  It would be 
very handy to retrieve the channel id given event channel.  I also do not see a way to retrieve 
proxyId's given a consumer proxy or supplier proxy.

2. Publishers are allowed to push events for the types they have not offered using offer_change(). 
This can allow publishers to never advertise the event types they are offerring.  For Example, as 
a consumer it can not tell what are the event types that will be offered on the channel. 

Can we require publishers to always call offer_change() on the new event types before pushing them to 
the event channel?  If this behavior breaks existing functionality, then how about having push_xxx()
check to see if the event type is offered, if not add it to the aggregate list?

1. The interface on EventChanel does not have the ability to get the ChannelId?  This 
is quite contrary to the ability on the admin interfaces to retrieve the adminId's. 
Given a proxy to EventChannels's it will be very handy to retrieve the channelId's.

The interface on EventChanel does not have the ability to get the ChannelId?  This 
is quite contrary to the ability on the admin interfaces to retrieve the adminId's. 
Given a proxy to EventChannels's it will be very handy to retrieve the channelId's. The Supplier Proxy & Consumer Proxy interfaces also do not have the ability to get 
the ID's.

just came across Notification Service IDL where
you are mentioned as contact person:


When I  "grep"  on  all  IDL files I notice that 
orb.idl is not included. 


I notice further that file 


 CosNotifyFilter.idl


in line 105 makes use of TypeCode:


  readonly attribute CORBA::TypeCode value_type; 


However section 3.19 "CORBA Module" of current 
CORBA standard 3.0.2 reads as:


>> Names defined by the CORBA specification are in a module named CORBA. In an
>> OMG IDL specification, however, OMG IDL keywords such as Object must not be
>> preceded by a "CORBA::" prefix. Other interface names such as TypeCode are not
>> OMG IDL keywords, so they must be referred to by their fully scoped names (e.g.,
>> CORBA::TypeCode) within an OMG IDL specification.
>> For example in:
>> #include <orb.idl>
>> module M {
>> typedef CORBA::Object myObjRef; // Error: keyword Object scoped
>> typedef TypeCode myTypeCode; // Error: TypeCode undefined
>> typedef CORBA::TypeCode TypeCode;// OK
>> };
>> The file orb.idl contains the IDL definitions for the CORBA module. Except for
>> CORBA::TypeCode, the file orb.idl must be included in IDL files that use names
>> defined in the CORBA module. IDL files that use CORBA::TypeCode may obtain its
>> definition by including either the file orb.idl or the file TypeCode.idl.


In my humble opinion, "your" Notification IDL is
not correct.


Perhaps  you  have  some time left to comment on
this?