Test Orchestration Platform for Service-oriented, Heterogeneous, Adaptive, Polyglot Enterprise systems - a single Java reference architecture that drives Web, Mobile, SAP GUI, Oracle EBS, REST/SOAP, SQL/NoSQL, batch, and messaging layers through one reflection-based dispatch contract.
Most frameworks specialize in a single layer - a web UI, a REST API, or a database - and stitch cross-layer business workflows together with brittle glue scripts. In regulated, ERP-heavy domains, one business transaction can span SAP GUI, Oracle EBS, a CRM API, a database write, a batch job, and a messaging event - with no framework treating them as one orchestrated test.
Test Orchestration Platform for Service-oriented, Heterogeneous, Adaptive, Polyglot Enterprise systems - a Java reference architecture that unifies execution across every layer under one reflection/MethodHandle-based dispatch mechanism.
Fig. 1 - Scenario-driven orchestration over heterogeneous service components, resolved dynamically at runtime.
Fig. 2 - "Create and Bill Health Policy," six steps, zero hardcoded coupling between Policy and Billing.
| Capability | Robot FW | Cucumber | Karate | Citrus | TOPSHAPE |
|---|---|---|---|---|---|
| Keyword-driven dispatch | Yes (static lib) | Glue code | DSL-based | XML/Java DSL | Reflective, no static map |
| Gherkin/BDD authoring | Optional | Yes | No | No | No (YAML/JSON/Excel) |
| REST/SOAP API testing | Via library | Via library | Native | Native | Native (RestAssured/SAAJ) |
| SQL/NoSQL DB validation | Via library | Via library | Limited | Native | Native (JDBC + Mongo) |
| Legacy desktop/ERP UI | No | No | No | No | Native (UFT Developer) |
| Batch / messaging triggers | Via library | Via library | Limited | Native | Native (shell/REST/JSch) |
| Shared cross-layer context | Partial | Partial | Partial | Partial | Yes (single TestContext) |
| License | Apache 2.0 | MIT | MIT | Apache 2.0 | Apache 2.0 |
TOPSHAPE has not yet run against a live enterprise system. Verification to date combines line-by-line manual code review with an independent Python cross-check of the two algorithmically non-trivial routines (ServiceResolver resolution, ParamResolver substitution) - both matched expected behaviour. Defect-escape rate, authoring time, and maintenance-effort metrics are explicitly future work, not asserted claims.
for End-to-End Testing Across Heterogeneous Enterprise Platforms