Skip to main content
OfficeIMO Docs Search documentation/

Validation and Support Evidence

Edit on GitHub

Understand what source, automated tests, generated API reference, package artifacts, and deployment smoke tests each prove.

OfficeIMO support is reported at the level that was actually validated. A project file proves that a component exists; it does not prove every workflow on every runtime. A passing unit test proves its scenario; it does not prove a published package or a NativeAOT binary.

Evidence ladder

EvidenceWhat it provesWhat it does not prove
Project metadataOwnership, dependencies, target frameworks, package intentRuntime behavior or output fidelity
Generated API referencePublic types and members in a built assemblyThat a particular workflow was exercised
Unit and integration testsThe asserted scenario on the tested framework and hostEvery document variant or deployment mode
Generated artifact readbackThe file can be reopened and key content survivesPixel-perfect parity with every Office client
Packaged-artifact smoke testThe consumer path works outside project referencesPublic feed availability unless downloaded from that feed
Platform smoke testThe scenario executed on that OS and runtime identifierOther architectures or unexercised code paths
NativeAOT publish and executionThe tested dependency graph compiled and ran nativelyBlanket AOT support for unrelated packages and workflows

How the documentation uses evidence

Conceptual pages describe the intended workflow and link into generated reference for exact APIs. The component catalog is generated from all production project files. PSWriteOffice counts come from its module manifest and fail validation if a cmdlet belongs to no documentation family or more than one total is reported.

The NativeAOT page is maintained against checked-in executable scenarios for authoring, extraction, and rendering. A repository script publishes each application, runs semantic assertions, and can write a machine-readable result. Package-wide trim and AOT analyzer builds complement those runtime checks without being presented as a substitute for them.

Validate your own document corpus

Document formats are broad and real files combine features in ways synthetic samples do not. Before production:

  • select representative source files, including damaged and edge-case inputs;
  • assert semantic outcomes such as text, table data, comments, fields, formulas, pages, or signatures;
  • capture structured diagnostics and decide which codes block delivery;
  • reopen generated files with OfficeIMO and, where relevant, the target desktop or web client;
  • preserve artifacts from failures so a regression can be reproduced.

The repository tests and examples are a starting point. Your accepted fidelity policy remains part of the application contract.