Universal Data Transport
Learn what UDT is and how it differs from processing and physical transport.
Universal Data Transport (UDT)
Universal Data Transport (UDT) is the data model and exchange contract for representing and moving Sinapsis DataContainer values between compatible systems. Its central principle is:
Standardize the container. Optimize the payload.
Templates work with semantic DataContainer objects. When those objects stay in one execution environment, they may be passed directly as Python objects. When data needs to cross a boundary, UDT can encode it into a wire representation and reconstruct it on the receiving side.
flowchart LR
A[DataContainer] --> E[UDT encoding]
E --> W[Wire bytes]
W --> T[HTTP / Queue / TCP / other transport]
T --> D[UDT decoding]
D --> B[DataContainer]
Green-bordered nodes reference Sinapsis templates. Hover or focus them to fetch documentation.
View Mermaid source
flowchart LR
A[DataContainer] --> E[UDT encoding]
E --> W[Wire bytes]
W --> T[HTTP / Queue / TCP / other transport]
T --> D[UDT decoding]
D --> B[DataContainer]
UDT does not require all payloads to share one codec. A DataFrame may use Arrow; NumPy arrays can use binary representation; and stable custom message schemas may use Protobuf as a payload codec, not as a replacement for the container contract. Large data can be referenced through an artifact store.
Distinguish the layers: the Agent model specifies the workflow; a Template performs computation; UDT specifies data representation/exchange; and physical transport moves the encoded bytes.
Read UDT Architecture for the specification and Sinapsis Protocol Quickstart for the Python implementation.
Published from Git revision 7b7ba25d627b