Support for master data sequences #114
Description
Support a new type of sequence to deal with master data replication. Some headers will be used to delimit the dataset.
It should ideally be possible to publish a collection of messages and have the Silverback producer automatically add the necessary headers.
The consumer should recognize the dataset and create an ISequence to deal with it as a unit of work, where the subscriber recognizes when all messages have been received (subscribing to the IMessageStreamEnumerable, that will complete when the entire set has been consumed).
The headers should be as follows.
| Header | Description |
|---|---|
| x-dataset-last | Marks the last dataset element in this partition. |
| x-dataset-first-offset | Starting offset of the dataset in this partition. Present in all records except the first one of each partition. |
| x-dataset-version | Optional. Monotonic increasing number identifying the dataset version. |
| x-dataset-index | Sequence number, within a version. Starts with zero. |
| x-dataset-count | Optional. Total number of records. |