Change control dependencies determine which software artifacts require authorization, who may approve modifications, and how one change propagates through related work products. The documented frameworks do not describe a single universal process. Instead, they distinguish levels of control by configuration item, assign owners and change authorities, and require traceability relationships to be mapped before impact can be assessed. A quality software team must therefore treat change control as a set of explicit dependency decisions rather than a generic approval workflow. The practical question is not whether to control changes, but which items, authorities, and relationships the control procedure actually names. This guide examines those distinctions using NASA, Berkeley Lab, IEEE, and FDA documentation, then proposes a hypothetical validation scenario to test the dependency logic.
Configuration Items and Levels of Control
NASA guidance states that each identified configuration item has a defined level of control for making changes to that item [1]. Documentation might have one level of change control, while source code might require a different level [1]. This means the dependency structure begins with classification: an item must be identified as a configuration item before any control level can be assigned. The guidance also requires procedures to designate the levels of control through which each identified software configuration item is required to pass [1]. Without that designation, a change request cannot be routed to the correct authority.
Berkeley Lab's safety software quality assurance requirements add a labeling dimension. A labeling system for configuration items must uniquely identify core configuration items, identify changes by revision or version identifier, and provide the ability to uniquely identify each approved configuration of the revised software that is available for use [2]. This labeling requirement creates a dependency between change control and deployment boundaries: an approved configuration is only meaningful if it can be distinguished from prior and subsequent versions. The inventory of related systems is maintained at the system-owner or functional-owner level, while a broader inventory of safety software systems is maintained by the Office of Contract Assurance [2].
Ownership and Change Authority
NASA documentation states that each configuration item or class of items has a defined owner who is responsible for authorizing changes to that item, such as the software Change Control Board [1]. Ownership is therefore not merely administrative; it is the locus of authorization. The same guidance notes that levels of control may be affected by the change originator. A change request from outside the organization, such as from customers or end-users, may require different authority for approval than changes requested by software developers [1]. This originator dependency means the same technical modification can follow different approval paths depending on who initiates it.
Change authority examples include Change Control Boards, Change Authorization Boards, Engineering Change Boards, peer review teams, and project managers [1]. The approval authority may be determined by complexity, cost, risk, or effect of the change [1]. NASA guidance also recommends including software assurance and safety personnel to ensure safety impact is considered, and involving stakeholders based on the expertise and authority required to disposition the change [1]. Berkeley Lab similarly requires that changes to software be systematically proposed, evaluated, documented, and approved so that impact and rationale are carefully assessed prior to updating the software baseline [2].
A Hypothetical Validation Scenario for Dependency Testing
Consider a suggested scenario: a quality management system spreadsheet used to record production test results receives a change request to add a validation rule that restricts a numeric field. The change originator is an end user in manufacturing, not the original developer. Under NASA guidance, this originator distinction may require a different approval authority than a developer-initiated change [1]. The configuration item owner, perhaps a quality systems Change Control Board, must first authorize the change [1]. The evaluation then requires assessing impact to other baseline products, including any traceability matrix linking the spreadsheet to test documentation [3].
The regression dependency becomes visible when the team must determine whether archived test cases from the original validation are available. FDA guidance states that failure to archive previous test documentation can significantly increase revalidation effort [7]. The team would also need to verify that the specification is updated before the change is made [7]. This scenario is a suggestion for testing the dependency logic, not a claim that any specific tool or team has performed this validation. It illustrates how ownership, originator, traceability, archival, and specification dependencies intersect in a single change request.
Dependency Supply Chains and Traceability
The Software Engineering Body of Knowledge describes software products as heavily reliant on dependencies, including internal and external commercial or open-source dependencies [4]. Package managers such as Maven in Java and NPM in JavaScript are widely used to automate installing, upgrading, configuring, and removing dependencies [4]. The direct and indirect dependencies of software products constitute a dependency supply chain network [4]. This network view means a change to one component can affect artifacts that do not directly reference it, because indirect dependencies propagate through the chain.
Traceability is the mechanism for making those propagation paths visible. The SWEBOK states that work products may share dependency relationships such as uses, implements, and tests [4]. As software is developed, managed, maintained, or extended, these traceability relationships must be mapped and controlled to demonstrate the software requirements' consistency with the software model and the many work products [4]. Traceability enables change analysis once the software is developed and released because relationships to software work products can easily be traversed to assess change impact [4]. NASA guidance similarly lists evaluating impact to other baseline products, such as design, tests, and documentation, and notes that traceability matrices are helpful here [3].
Evaluation Boundaries and Regression Dependencies
NASA's change evaluation guidance requires assessing impact to schedule and cost, including making and testing the change and regression testing the software [3]. It also requires evaluating impact to functions and features, interfaces, and system resource requirements [3]. Software assurance checks that the project tests any code that has changed and runs a set of regression tests to see that the change has not caused problems anywhere else in the software system [6]. If the software is safety-critical, a full set of regression tests must be run to ensure there was no impact on the safety-critical functions [6]. This creates a dependency between change criticality and test scope.
FDA guidance adds an archival dependency. The effort to fully validate a change is dependent upon the degree to which validation of the original software was documented and archived [7]. Test documentation, test cases, and results of previous verification and validation testing need to be archived if they are to be available for performing subsequent regression testing [7]. Failure to archive this information for later use can significantly increase the level of effort and expense of revalidating the software after a change is made [7]. The FDA also states that specifications should be updated before any maintenance and software changes are made [7], making specification currency a precondition for controlled change.
Risk-Based Rigor and Unresolved Questions
FDA's Computer Software Assurance guidance describes a risk-based approach to establish confidence in the automation used for production or quality management systems, and identifies where additional rigor may be appropriate [5]. The guidance is nonbinding and represents current thinking rather than legally enforceable responsibilities [5]. A Manufacturing Execution System failure may disrupt operations but not affect the process parameters established to produce a safe and effective device [5]. This distinction means the dependency between a change and patient safety, not the software category alone, determines the appropriate level of control.
Several questions remain unresolved in the source documentation. NASA guidance does not specify a universal threshold for when a change requires a product-level Change Control Board versus a lower authority; it directs projects to define levels based on criticality and range of effect [1]. The SWEBOK notes that requirements change control includes an optional impact analysis stage to more thoroughly examine benefits and costs of a requested change [4], but does not mandate when that stage becomes necessary. Berkeley Lab states that changes to previously accepted software will be subject to the same level of control [2], yet the documentation does not enumerate every possible control level. Teams must therefore define these boundaries in their own configuration management plan rather than import them from a generic template.
Sources & references
- View Source · swehb.nasa.gov
- Safety Software Quality Assurance Requirements · it.lbl.gov
- View Source · swehb.nasa.gov
- Guide to the Software Engineering Body of Knowledge · ieeecs-media.computer.org
- Computer Software Assurance for Production and Quality Management System Software · www.fda.gov
- SWE-080 - Track and Evaluate Changes - SW Engineering Handbook Ver C - Global Site · swehb.nasa.gov
- General Principles of Software Validation - Final Guidance for Industry and FDA Staff · www.fda.gov