Topic · requirements diagram

Requirements diagram

Gist

A requirements diagram is the spec tree as it sits in the repo this commit, not a SysML drawing from last quarter. Proof prints it as Mermaid with proof diagram hierarchy --format mermaid. Sparx still draws. Jama still authors.

proof diagram hierarchy --format mermaid

Keep SysML if you model «satisfy» and «derive» in a CASE tool. Keep Sparx if the programme wants a canvas. Stdout is not a drawing surface, and a node count is not a proof of the function.

Request demo Last stale drawing vs this tree

01 · The drawing that outlived the shall

A box in last quarter's export is not a live parent.

The merge stayed green because nobody regenerated the picture from the files that actually failed the gate.

On this install SYS-REQ-010 was retired from specs/system. The Sparx requirements diagram still showed it as a parent with two «derive» edges. Reviewers argued about the picture. The YAML no longer named the id.

proof diagram hierarchy --format mermaid does not open a canvas. It walks the configured SpecTree and prints one node per spec level, parent-child edges, and the requirement count on each node. Unsupported --format values are rejected. There is no legacy flow subcommand.

The split of one parent into owned children lives on requirements decomposition. The blast-radius graph for a single id lives on software change impact analysis as proof diagram impact. This URL is the tree, not the split and not the impact walk.

proof diagram hierarchy --format mermaid
proof diagram components --format mermaid
proof diagram impact SYS-REQ-001 --format mermaid
proof validate

On this install the teaching miss is a stale export: the diagram still named a retired parent, and CI never asked whether the picture matched the tree.

02 · The exhibit

Same repo. A stale box, or a count from this commit.

The last drawing still showed SYS-REQ-010. The hierarchy command will not print a level that is not in the manifest. Click the tabs.

This push

export Sparx REQ diagram, Q2
box SYS-REQ-010 still a parent
tree id retired from specs/system

Merge stayed green

Finding

No check asked whether the picture named an id the YAML no longer carries. Reviewers argued about «derive» edges on a retired parent.

Stale canvas

This push

Still last week's YAML. The command will not invent a box for an id that is not in the SpecTree. Fix the files, then reprint.

Keep the YAML

Proof

Ask proof diagram hierarchy --format mermaid
Kind system 11, software 40, interface 8

retired id is absent

Same repo. A stale box, or a count from this commit. Click the tabs.

Invariant	What they do	What Proof does
SysML Requirement Diagram	A static structural diagram of «requirement» blocks and «satisfy» / «derive» / «refine» edges. Wikipedia and sysml.org own that definition.	Not SysML. Proof does not emit «satisfy» stereotypes or a SysML XMI file. Keep the CASE tool if the programme is modeled that way.
Mermaid requirementDiagram	Hand-authored Mermaid syntax for requirement nodes and typed links, documented by mermaid.ai.	Not that grammar. Proof prints a hierarchy or impact flowchart from the live tree. You paste stdout into a renderer.
Sparx / Visual Paradigm / PTC	A canvas where an analyst draws requirements and traces inside the model.	Not a canvas. Stdout only. `--format` other than mermaid is rejected.
Component diagram	UML boxes for callers and callees.	`proof diagram components` prints documented interface edges. Different subcommand. Not this H1.
Impact diagram	A blast-radius picture for one change.	`proof diagram impact SYS-REQ-001` on software change impact analysis. Suspect-link styling lives there. Not this H1.
Jama / DOORS	A hierarchy view of objects in the ALM, including a retired id until someone deletes the row.	Jama still authors. IBM DOORS stays a mention on Proof vs Jama. Proof reprints from the repo.
A count is a proof	Twelve boxes on the diagram means twelve shalls hold in the code.	A node count is how many YAML files sit at that spec level this commit. It is not a proof of the function.

Reprint after the tree changes. If the hierarchy is empty, run proof validate before arguing with the renderer. Impact diagrams need refreshed trace links or the edges will lie the same way the Sparx export did.

proof validate
proof diagram hierarchy --format mermaid
proof diagram components --format mermaid
proof diagram impact SYS-REQ-001 --format mermaid

We have not run Sparx, SysML, and Proof on the same frozen corpus, and we have not claimed this Mermaid is a SysML Requirement Diagram or a proof of the Go. The loss is named, not scored.

03 · The honest loss

Stdout is not a CASE tool.

Proof reprints the SpecTree. It does not author «satisfy» edges, qualify a drawing tool, or prove the code. Jama still authors.

The command does not split a parent into owned children. That H1 lives on requirements decomposition as proof req decompose. It does not walk the blast radius of one id. That H1 lives on software change impact analysis. It does not import ReqIF. That H1 lives on ReqIF. There is no flow subcommand for retired inline state machines.

Unsupported formats fail closed. An empty tree is a manifest problem, not a renderer problem. Jama still authors.

Request demo How the install is bought

04 · Nearby questions

What people type next.

What is a SysML Requirement Diagram? sysml.org and Wikipedia own that SERP as «requirement» blocks and typed traces. Same cluster as this URL only when the tree is software shalls in a repo. Not a twin for a CASE model.

What is a component diagram? Callers and callees. proof diagram components --format mermaid on this install. Not this H1.

What is an impact diagram? Downstream requirements and artifacts for one id. software change impact analysis as proof diagram impact.

What is requirements decomposition? One parent split into owned children. requirements decomposition as proof req decompose.

What is a requirement hierarchy? Parent and child files in the graph. Same cluster as this URL.

Is Proof a Sparx alternative? No. Sparx still draws. Proof fails closed on a bad format and reprints from YAML. Jama still authors.