tagdepends

Tagdepends is a conceptual approach to dependency management that uses metadata tags attached to software components to express requirements, capabilities, and context. It treats tags as first-class descriptors for features, environments, and platform constraints, enabling resolution to account for more than fixed version pins alone.

Key concepts include tags, tag providers (components that supply capabilities), tag requirements (tags asserted by dependents), and a dependency resolver that selects a compatible set of components by matching requirements to provisions under defined policies. Tagsets may be combined with traditional version constraints to produce a hybrid resolution strategy. Conflicts are detected when no component can satisfy a given tag requirement or when two components offer incompatible provisions.

Usage typically involves annotating components with a set of tags, declaring tag-based requirements for dependents, and running a resolver that outputs a consistent build or deployment graph. This approach is common in build systems, deployment pipelines, plugin ecosystems, and feature-flag driven configurations where permutations would be impractical to enumerate.

Example: a library component may declare provides: tags "db-support" and "python3.11", and requires: tags "linux" and "openssl". A consumer that enables "db-support" and targets "linux" will select this library if the environment satisfies all required tags and any version or capability constraints are met.

Benefits include modularity, flexible feature composition, and easier support for multi-environment deployments. Drawbacks include tagging overhead, possible ambiguity if tags are under-specified, and performance challenges for large graphs.

Tagdepends is described as an evolving concept rather than a single standard. It draws on practices from feature flags, tag-based configuration, and modern dependency resolution, and is discussed in open-source communities and academic literature as a way to manage combinatorial dependency scenarios.