Just a second...

Int64 topics

A topic that provides data in 64-bit integer format. The data is transmitted in a binary form for increased efficiency. Int64 topics are stateful: each topic stores a 64-bit integer on the Diffusion™ server .

Why use an int64 topic?

An int64 topic enables you to explicitly type the data that you send through the topic as a 64-bit integer.

Int64 topics support null data values.

The value of the topic is transmitted as Concise Binary Object Representation (CBOR) . For more information about CBOR , see http://cbor.io/.

The value of the topic is accessible both as a 64-bit integer and CBOR .

Properties of an int64 topic

When you create an int64 topic you can specify the following properties in the topic specification:
DONT_RETAIN_VALUE
If set to true, the latest value of the topic is not retained by the Diffusion server or the client that publishes it. New clients that subscribe do not receive an initial value. No value will be returned for fetch operations that select the topic.

For time series topics, if DONT_RETAIN_VALUE is set to true, time series events are still retained, but the latest value is not stored separately.

The DONT_RETAIN_VALUE property is useful for applications like a feed of news items, or for values that are only valid at the moment of publication. You can combine this with VALIDATE_VALUES.

Using DONT_RETAIN_VALUE reduces the topic memory footprint, but disables delta streaming. Using an update stream will send complete values to the server, instead of deltas. Disabling delta streaming is likely to increase the bandwidth used unless subsequent values are unrelated.

This property replaces the obsolete stateless topic type which was removed in Diffusion 6.2.

TIDY_ON_UNSUBSCRIBE
If set to true, when a session unsubscribes from the topic, any updates for the topic that are still queued for the session are removed.

There is a performance overhead to using this option as the client queue must be scanned to find topic updates to remove, however it may prove useful for preventing unwanted data being sent to sessions. This property is disabled by default.

VALIDATE_VALUES
If set to true, the topic rejects updates that would create invalid instances of the topic's data type.

If set to anything other than true, no validation is performed and all values are streamed to subscribing clients whether they are valid data or not.

Validation has a performance overhead and is disabled by default.

Note: If validation is disabled and the value provided is not valid, the client might produce errors or other unexpected behavior. The exact error varies depending on the client platform. To avoid this, use the client-side validation method provided by the Diffusion API.
CONFLATION
Used to set a conflation policy for the topic. Conflation can merge or discard topic updates to reduce server memory and network data usage.
The supported values are:
  • off
  • conflate
  • unsubscribe
  • always
The default is "conflate".
Note: Time series topics do not allow a conflation policy of 'conflate' or 'always'. They default to 'off'. You can set a conflation policy of 'unsubscribe'.

See Using conflation for details of the policies.

OWNER
Used to set a security principal as the owner of the topic. The principal receives READ_TOPIC, MODIFY_TOPIC and UPDATE_TOPIC privileges.

The format of the property value is:

$Principal is "name"

where name is the name of the principal.

REMOVAL
Used to set a topic removal policy.

The property is an expression which defines the conditions which will trigger automatic removal of the topic or of a set of topics specified with an optional selector.

See Removing topics automatically for details of the expression format.

PERSISTENT
Used to disable persistence for a topic.

If set to "false", the topic will not be persisted, even if persistence is enabled for the server.

Considerations when using an int64 topic

Deltas are not available for data published to int64 topics.

Because of a limitation of the JavaScript® platform, the validity of the values converted to an Int64 from a Number or converted to a Number from an Int64 can only be guaranteed for values up to 253 - 1. Converting between a String and an Int64 has fully guaranteed precision.