Set validation before import, generation, and publication

Which validations should be performed before import, generation, and publication?

Before import, format, schema, and identity are checked; before generation, functional completeness and permissible values ​​are verified. Before publication, result, link, rights, security, and presentation checks are performed to ensure that erroneous artifacts never silently advance to the next state.

Late Detection

• Late Detection – An input error only becomes visible in the final content, resulting in time-consuming traceability and partial publication.

• Over-validation – Overly restrictive rules block permissible new cases and lead to manual workarounds outside the process.

• Warning overload – Numerous inconsequential warnings make critical errors difficult to detect during the review.

Diagnostic case: "Late detection"

A CSV file passes the coding check but contains an unknown category and later a broken source link. Category validation stops the generation of the row; the link check then blocks the finished artifact because the target URL did not previously exist.

Blocking based on risk

1. Rules are assigned to the import, transformation, generation, and publication boundaries based on the required information.

2. Each rule receives an error level, owner, correction note, and a positive and negative test case.

3. Error patterns and manual overrides are evaluated to adjust rules that are too strict or missing.

Early Responsibility

• Early Responsibility – A rule is located at the earliest boundary where all necessary information is reliably available.

• Blocking based on risk – Critical errors halt the data record or batch, while clearly marked warnings allow for justified release.

• Traceable Findings – Error reports specify the object, field, rule, actual value, and possible correction without unnecessarily duplicating sensitive data.

Traceable Findings

How "Validate Before Any External Impact" relates to related decisions

Controlling Automations with Unique IDs and Status Values answers the next practical question: How do IDs and status values ​​prevent duplicate or lost processing?

Information architecture for platforms with multiple user roles continues the thought with another question: How does an information architecture connect common content with different user roles?

If you want to put "Validate Before Any External Impact" into practice, you can refer to Robust Website Systems which focuses on "Data Pipelines and Observability" and "Early Responsibility."

Conclusion: Validate Before Any External Impact

Validation belongs at every boundary where new certainty or risk arises. Early, concrete findings prevent costly error propagation.

Sources and Further Information

The primary sources define the business framework for "Validate before any external effect."

• Input Validation Cheat Sheet — OWASPOWASP recommends early syntactic and semantic checks, allowlisting, and server-side validation at trusted system boundaries.

• JSON Schema SpecificationJSON Schema provides a standard for describing and validating data types, required fields, value ranges, and structural contracts in a machine-readable way.