Parsing on demand

An input message can be of any length. To improve performance of message flows, a message is parsed only when necessary to resolve the reference to a particular part of its content. If none of the message content is referenced within the message flow (for example, the entire message is stored in a database by the DataUpdate node, but no manipulation of the message content takes place), the message body is not parsed.

If a parser can parse an input bit stream on demand, instead of immediately parsing the entire bit stream, the Parse Timing property of a message flow node controls the on-demand behavior of the parser.

You can set the Parse Timing property to On Demand (the default), Immediate, or Complete.

On Demand causes partial parsing to occur. When fields in the message are referenced, as much of the message is parsed as is necessary to completely resolve the reference. Therefore, fields might not be parsed until late in the message flow, or never. This restriction applies to both the message body and the message headers.

Immediate and Complete both override partial parsing and parse the entire message, including any message headers, except when the MRM parser encounters an element with a complex type with Composition set to Choice or Message that cannot be resolved at the time; for example, the content must be resolved by the user in ESQL. If Composition is set to Choice, the data is added to the message tree as an unresolved item, and parsing continues with the next element. If Composition is set to Message, parsing terminates at that point. The only difference in behavior between Immediate and Complete occurs when MRM validation is enabled.

The Parse Timing property also gives you control over how MRM message validation interacts with partial parsing. Refer to Validation properties for a full description.

The Parse Timing property has no effect on the serialization of output messages.