Module Glossary

Page Maps

graph LR
  family["Reproducible Research"]
  program["Deep Dive DVC"]
  section["Truthful Pipelines Declared Dependencies"]
  page["Module Glossary"]
  capstone["Capstone evidence"]

  family --> program --> section --> page
  page -.applies in.-> capstone

flowchart LR
  orient["Orient on the page map"] --> read["Read the main claim and examples"]
  read --> inspect["Inspect the related code, proof, or capstone surface"]
  inspect --> verify["Run or review the verification path"]
  verify --> apply["Apply the idea back to the module and capstone"]

This glossary belongs to Module 04: Truthful Pipelines and Declared Dependencies in Deep Dive DVC.

Use it to keep the module language stable while you move between the core lessons, the worked example, the exercises, and capstone review.

How to use this glossary

Read the directory index first. Return here when a term starts to feel vague or when a review discussion needs sharper language.

The goal is not extra theory. The goal is a shared vocabulary for explaining why a DVC pipeline reruns, skips, records evidence, or becomes deceptive.

Terms in this directory

Term	Meaning in this directory
declared dependency	A file or directory listed in deps because the command reads it and its content can influence the result.
declared output	An artifact listed in outs because the stage owns it and DVC should record or restore its content identity.
declared parameter	A reviewed control value listed in params, usually from params.yaml, that should influence rerun decisions.
stage contract	The reviewable promise that a command uses declared inputs and controls to produce declared outputs.
truthful stage	A stage whose declaration matches the real reads, controls, writes, and ownership boundary of the command.
deceptive stage	A stage that looks organized but hides a real influence or artifact boundary from the declared graph.
DVC graph	The directed execution structure created from stage declarations and artifact edges.
DAG	A directed acyclic graph; in this module, the shape DVC uses to understand stage order and dependency flow.
staleness	The condition where current declared state no longer matches recorded lock evidence, so rerun is needed.
false rerun	A visible inefficiency where a stage reruns even though the meaningful result would not have changed.
stale output	A correctness risk where a result should change but DVC has no declared reason to rerun the producing stage.
hidden influence	Any real input, control, environment fact, or artifact dependency that affects a result without being declared or otherwise reviewed.
lock evidence	The recorded dependency hashes, parameter values, command text, and output identities in dvc.lock.
producer	The stage that owns and declares an output artifact.
consumer	A stage that reads another stage's output as a dependency.
shared intermediate	An artifact produced once and consumed by more than one downstream stage.
fan-out	A graph shape where one output feeds multiple consumers.
fan-in	A graph shape where one stage depends on multiple upstream artifacts.
output ownership	The rule that one stage should be responsible for a declared artifact boundary.
scratch artifact	A temporary file created during execution that should not become a declared output unless downstream review depends on it.
broad dependency	A dependency declaration, such as an entire directory, that may cause reruns for changes unrelated to the stage result.
narrow dependency	A dependency declaration that names the actual files or directories the command reads.
control surface	The set of parameter values that reviewers expect to change deliberately and compare across runs.
graph refactoring	A change to stage shape, names, boundaries, or artifact paths that should preserve the provenance story.
provenance story	The explanation of which declared inputs and controls produced which declared outputs.

Stable review questions

Use these questions when the module feels abstract:

• What does this stage promise?
• Which files does the command really read?
• Which control values should be reviewed as parameters?
• Which artifacts does the stage own?
• What would make this stage stale?
• What evidence should dvc.lock record after a rerun?
• Is this surprise a false rerun or a stale-output risk?
• Does this refactor make the provenance story clearer or harder to explain?