Liking cljdoc? Tell your friends :D

Overarch Usage Overview

The usage of Overarch is twofold. On the one hand, Overarch provides an open data format to describe the concepts, functional requirements, architecture, and design of software systems. On the other hand it is a tool to transform the description into diagrams, reports or other representations.

Overarch can be used as a CLI tool to convert specified models and diagrams into different formats, e.g. the rendering of diagrams in PlantUML or the conversion of the data to JSON. It can also be used to query the model by criteria and generate artifacts based on the result with templates.

Use Cases of Overarch

Rationale for Overarch

Modelling and vizualizing a software system or a system landscape is useful for thinking, communication, documentation and onboarding collegues. The systems can be modelled with different level of details and vizualized with different views.

To model a system you can either use a modelling tool or a diagramming tool. Most modelling tools are commercial and lock in your models or not comprehensive. Modelling is also often decoupled from development.

Most diagramming tools don't use a model or don't separate the model information from the layout. The scope of most diagramming tools is limited to diagrams only. With diagramming tools like Excalidraw or Gliffy it's quite easy to draw some diagrams for a system. But changing existing diagrams can be cumbersome when diagram elements have to be placed and moved manually. This also limits the scope and size of the diagrams that can be created and maintained.

Some problems most of these tools often face is versioning and collaboration. By using a text based representation of the model and the views, a development workflow with version control, e.g. git can be applied to modelling.

Some tools like PlantUML or Structurizr provide a DSL to model a system or to describe diagrams. While PlantUML operates at the diagram level and is basically a text based diagramming tool with auto layout, structurizr is a modelling tool which also features a graphical editor. But to use the model in a different context, the DSL has to be parsed.

Overarch addresses these issues by providing a text based data format for the specification of the model and the views of the model. By using plain data literals like sets and maps, no DSL parser is needed to instanciate the model.

Overarch is free, open and extensible

  • provides model information as plain data
  • models are readable, discoverable, composable, reusable and extensible
  • model queries, vizualization, artifact generation

Command Line Interface

Overarch CLI
   
   Reads your model and view specifications and renders or exports
   into the specified formats.

   For more information see https://github.com/soulspace-org/overarch

Usage: java -jar overarch.jar [options].

Options:

  -m, --model-dir PATH              models     Models directory or path
  -r, --render-format FORMAT                   Render format (all, graphviz, markdown, plantuml)
  -R, --render-dir DIRNAME          export     Render directory
      --[no-]render-format-subdirs  true       Use subdir per render format
  -x, --export-format FORMAT                   Export format (json, structurizr)
  -X, --export-dir DIRNAME          export     Export directory
  -w, --watch                       false      Watch model dir for changes and trigger action
  -s, --select-elements CRITERIA               Select and print model elements by criteria
  -S, --select-references CRITERIA             Select model elements by criteria and print as references
      --select-views CRITERIA                  Select and print views by criteria
  -T, --template-dir DIRNAME        templates  Template directory
  -g, --generation-config FILE                 Generation configuration
  -G, --generation-dir DIRNAME      generated  Generation artifact directory
  -B, --backup-dir DIRNAME          backup     Generation backup directory
      --[no-]model-warnings         true       Returns warnings for the loaded model
      --[no-]model-info             false      Returns infos for the loaded model
      --plantuml-list-sprites       false      Lists the loaded PlantUML sprites
  -h, --help                                   Print help
      --debug                       false      Print debug messages

CLI Examples

To render all views for all models, use

> java -jar ./target/overarch.jar -r all

or

> java -jar ./target/overarch.jar -r all --debug

To render all views for all models with a directory watch to trigger rerendering on changes, use

> java -jar ./target/overarch.jar -r all -w --debug

To export the models to JSON, use

> java -jar ./target/overarch.jar -x json

To query the model for all containers, use

> java -jar ./target/overarch.jar -s '{:el :container}'

or

> java -jar ./target/overarch.jar -S '{:el :container}'

Editor/IDE Integration

Visual Studio Code

With the addition of a few extensions, Visual Studio Code is a lightweight modelling environment. Install the following extensions in VS Code: Calva Extension PlantUML Extension Graphviz Preview Extension Graphviz Language Extension

With this setup you get an editor for the EDN files with code completion, syntax check and syntax highlighting.

Model editing

You also get integrated previews of the rendered PlantUML diagrams and the ability to generate image files in different formats (e.g. PNG, SVG, PDF, ...) directly from within Visual Studio Code.

IntelliJ

Modelling

Overarch supports modelling the functional requirements, the architecture and the design of the system under description.

Modelling a system with overarch should provide a value for the project and this guides the selection of model elements and supported abstractions and views.

Overarch currently supports the following kinds of models

  • Concept Models to model the concepts used for the system and their relations
  • Use-Case Models to model the scope of the system with the supported use cases
  • Architecture Models to model the logical architecture of the system with different levels of detail
  • Code Models to model the design of the system, e.g. the domain model, and the structure of the implementation
  • State-Machine Models to model relevant states and transitions of the system
  • Deployment Models to model the physical architecture of the system
  • Organization Models to model organizational structures, collaborations and responsibilities for parts of the system

The model contains all the elements relevant in the architecture of the system. Models are specified in the Extensible Data Notation (EDN).

EDN Basics

The Extensible Data Notation EDN is a data notation with a rich set of literals for scalar and composite data types. It is also a subset of the Clojure language textual format. Therefore Clojure plugins/extensions for editors or IDEs provide syntax checking/highlighting and code completion.

Compared to JSON, which only supports literals for numbers, strings, vectors (arrays) and maps, EDN provides a richer set of data literals, e.g. integer and floating point numbers, big integers and decimals, strings, symbols, keywords, UUIDs and instants of time. It also provides literals for list, vectors, sets and maps.

The following literals are used in Overarch models and views.

Strings

Strings are used e.g. as names and descriptions of model elements and for the title of views.

"This is a string"

"This is
a multiline
string"

Keywords

Keywords are used as keys in the maps for model elements and views. They are also used as identifiers for model elements and views.

Keywords start with a colon (:), have an optional namespace followed by a slash (/) and a mandatory name, e.g. :namespace/name.

Keywords should be prefixed with a namespace to avoid collisions with keywords for other models, which is especially relevant for identifiers or for custom keys in the model elements and views. Namespaces may have different parts, separated by a period (.), e.g. :org.soulspace/overarch.

:keyword
:namespaced/keyword
:my.namespaced/keyword

Unprefixed keywords and the namespace 'overarch' for map keys are reserved for overarch. Please use your own prefix if you want to add custom information to the maps in the model.

Sets

Sets are unordered collections of elements without duplicates. They are used as top level collections for the model elements and views. They are also used as a container for the children of model elements.

#{"a" "b" "c"}

Maps

Maps are associative collections of key/value pairs. They are used to describe the attributes of model elements and views.

{:firstname "John" :lastname "Doe" :age 42}

Vectors

Vectors are ordered collections of elements which may contain an element multiple times. They are used for the elements as content of a view because the ordering of the elements may be relevant for the rendering of the view (e.g. in PlantUML).

[1 2 3 4]
["John" "Doe"]

As you can see in the example models, all collection literals can be nested.

For more information see the EDN specification.

Examples

The model and diagram descriptions of the C4 model banking example can be found in models/banking folder:

If you have a Clojure environment in some editor or IDE, just use it. If not, try Visual Studio Code with the Calva and PlantUML extensions. With this setup you get an editor for the EDN files with code completion, syntax check and syntax highlighting.

Model editing

Models

You can split your model into separate EDN files, which might be reasonable for big systems. Overarch can recursively read all models from a directory or search path so you are quite free in structuring your model files.

The top level element in each model EDN file is a set which contains the top level model elements. Model elements are denoted as maps in the EDN file.

All model elements have at least two keys, :el for the type of the element and :id for the identifier. The identifiers should be namespaced keywords, so that different models can be composed without collisions of the identifiers.

Model Nodes

Model Nodes describe the elements of the different kind of models for the system.

Common Keys of Model Nodes

keytypevaluesdescription
:elkeywordsee model elementstype of the model node
:idkeywordnamespaced idid of the model node
:namestringshort namename of the model node
:descstringshort descriptiondescription of the model element, to be rendered in diagrams
:docmultiline stringlonger documentationdocumentation of the model element, not to be rendered in diagrams but textual output
:maturitykeyword:proposed, :deprecatedthe maturity of the model element
:tagsset of stringse.g. #{"critical"}some tags which can be used in element selection
:ctset of mapsmodel nodesthe children of the model node

Model Relations

Relations describe the connections and interactions of the nodes.

Common Keys of Relations

keytypevaluesdescription
:elkeyworde.g. :rel, :requesttype of the relation
:idkeywordnamespaced idid of the relation
:fromkeywordnamespaced idid of the referrer node
:tokeywordnamespaced idid of the referred node
:namestring name of the relation
:descstring description of the relation
:docmultiline stringlonger documentationdocumentation of the model element, not to be rendered in diagrams but textual output
:maturitykeyword:proposed, :deprecatedthe maturity of the model element
:tagsset of stringse.g. #{"critical"}some tags which can be used in element selection

References (:ref)

References refer to a model element with the given id. They are primarily used to refer to the model elements to include in views. They can also be used to refer to model elements in other model elements, e.g. to split a huge hierarchical systems into multiple EDN files.

References can have other keys, which are merged with the referred element in context of the reference. For example you can mark an internal system as external in the context of a specific view by adding :external true to the reference.

Boundaries

Boundaries group related elements and are normally rendered as a dashed box in a view. There are currently 4 types of boundaries, two of them implicit.

The implicit boundaries are the system boundary and the container boundary. They are not modelled explicitly but are rendered for referenced systems and containers in specific views. A system boundary is rendered, when an internal system with containers as content is referenced in a container view or component view. Likewise a container boundary is rendered for a referenced container in a component view.

The explicit boundaries are enterprise boundary and context boundary. These are explicitly modelled. An enterprise boundary {:el :enterprise-boundary} can be used to group systems by enterprise or company. A context boundary {:el :context-boundary} can be used to group concepts, containers or components by some common context, especially by domain contexts in the sense of domain driven design.

Architecture Model

Overarch supports elements for C4 architecture models.

Logical Data Model for the Architecture Model Elements

Architecture and Deployment Model Elements

Additional Keys for Architecture Model Nodes

keytypevaluesdescription
:subtypekeyword:database, :queuespecific role of the model node
:externalbooleantrue, falsedefault is false
:techstring technology of the model node

Person (:person)

Persons are internal or external actors of the system.

{:el :person
 :id :banking/personal-customer
 :name "Personal Banking Customer"
 :desc "A customer of the bank, with personal banking accounts."}

System (:system)

A System is the top level element of the C4 model an can contain a set of containers. Systems can be internal or external to the project context. The structure of internal systems is modelled with containers.

Container (:container)

A container is a part of a system. It represents a process of the system (e.g. an executable or a service). Containers are composed of a set of components.

Component (:component)

A component is unit of software, which lives in a container of the system.

Relations

Relations describe the connections and interactions of the parts of a view.

kindsync/asyncdependencydescription
:requestsynctruesynchrounous request
:responsesyncfalseresponse to a synchronous request
:sendasynctrueasynchronous point-to-point message
:publishasynctrueasynchronous broadcast message (via broker, topic, queue)
:subscribeasynctruesubscribtion to an asynchronous broadcast message (via broker, topic, queue)
:dataflowunspecifiedunspecifiedflow of data independent of the call semantic
:relunspecifiedunspecifiedunclassified relation

Additional Keys for Architecture Model Relations

keytypevaluesdescription
:techstringe.g. "REST"technology of the relation

Example

Example (exerpt from the banking model containing context and container level elements):

#{{:el :person
  :id :banking/personal-customer
  :name "Personal Banking Customer"
  :desc "A customer of the bank, with personal banking accounts."}
 {:el :system
  :id :banking/internet-banking-system
  :name "Internet Banking System"
  :desc "Allows customers to view information about their bank accounts and make payments."
  :ct #{{:el :container
        :id :banking/web-app
        :name "Web Application"
        :desc "Deliveres the static content and the internet banking single page application."
        :tech "Clojure and Luminus"}
       {:el :container
        :id :banking/single-page-app
        :name "Single-Page Application"
        :desc "Provides all of the internet banking functionality to customers via their web browser."
        :tech "ClojureScript and Re-Frame"}
       {:el :container
        :id :banking/mobile-app
        :name "Mobile App"
        :desc "Provides a limited subset of the internet banking functionality to customers via their mobile device."
        :tech "ClojureScript and Reagent"}
       {:el :container
        :id :banking/api-application
        :name "API Application"
        :desc "Provides internet banking functionality via a JSON/HTTPS API."
        :tech "Clojure and Liberator"}
       {:el :container
        :subtype :database
        :id :banking/database
        :name "Database"
        :desc "Stores the user registration information, hashed authentication credentials, access logs, etc."
        :tech "Datomic"}}}
 {:el :system
  :id :banking/mainframe-banking-system
  :external true
  :name "Mainframe Banking System"
  :desc "Stores all the core banking information about customers, accounts, transactions, etc."}
 {:el :system
  :id :banking/email-system
  :external true
  :name "E-mail System"
  :desc "The internal Microsoft Exchange email system."}

 ; Context level relations 
 {:el :rel
  :id :banking/personal-customer-uses-internet-banking-system
  :from :banking/personal-customer
  :to :banking/internet-banking-system
  :name "Views account balances and makes payments using"}
 {:el :rel
  :id :banking/internet-banking-system-uses-email-system
  :from :banking/internet-banking-system
  :to :banking/email-system
  :name "Sends e-mail using"}
 {:el :rel
  :id :banking/internet-banking-system-using-mainframe-banking-system
  :from :banking/internet-banking-system
  :to :banking/mainframe-banking-system
  :name "Gets account information from, and makes payments using"}
 {:el :rel
  :id :banking/email-system-sends-mail-to-personal-customer
  :from :banking/email-system
  :to :banking/personal-customer
  :name "Sends e-mail to"}} 

Deployment Model

Overarch also supports elements for C4 deployment models.

Logical Data Model for the Deployment Model Elements

Deployment Model Elements

Node (:node)

A node is a unit in a deployment view. Nodes represent parts of the infrastructure in which the containers of the system are deployed. They can contain a set of other nodes or containers.

Deployment model relations

relation typedescription
:linkA link between two nodes of the deployment model, e.g. two virtual networks
:deployed-toA deployment relation between a container and a node in the deployment model

:rel

Concept Model

A concept model captures relevant concepts of the domain(s) of the system. The concepts could be part of the ubiquous language of the systems domain.

A concept model can contain the concepts of the domain and the high level elements of the architecture model, e.g. the persons (actors), external systems and the system itself with it's containers.

Logical Data Model for the Concept Model Elements

Concept Model Elements

Concepts (:concept)

A concept which is relevant for the domain of the system. The description should document the meaning of the concept.

Concept Model Relations (:is-a, :has, :rel)

Concepts can be related with other concepts.

relation typedescription
:is-athe :from node is a specialization of the :to node
:hasthe :from node is a part or attribute of the :to node
:relunclassified relation between the nodes

Example

See example concept model.

Use Case Model

A use case model captures the functionality a system is suposed to deliver. High level use cases provide an overview of this functionality and may link to business processes, domain stories and arcitectural elements.

As such they provide a pivot for the traceability from business processes into the design of the system.

The elements of the use case model are mainly borrowed from the UML use case model so prior knowledge of UML modelling applies here.

Logical Data Model for the Use Case Model Elements

Use Case Model Elements

Example

Example Use Case Model

Use Cases (:use-case)

A use case describes the goal of an actor in the context of the system described. The goal can be a concrete user goal, a high level summary of user goals or a subfunction of a user goal. This is captured by the :level key.

keytypevaluesdescription
:levelkeyword:summary :user-goal :subfunctionspecific role of the element
:ext-pointsstring extension points of a use case

Actors (:person, :system, :container, :actor)

Persons, systems and containers from the architecture model should be used as actors in the use case model to provide a connection between the architecture model and the use case model.

You can use the :actor element to model actors not present as persons or systems in the architectural model, but this should be avoided if possible. A reason for an :actor element might be the introduction of a time actor to model the scheduling of use cases.

Relations (:uses :include :extends :generalizes)

Relations connect actors to the use cases or use cases with other use cases. Use case models support different kinds of relations.

kinddescription
:usesa use case element uses another use case element (e.g. an actor uses a use case or a use case uses an external system)
:includea use case includes the functionality of another use cases
:extendsa use case extends the functionality of another use case
:generalizes

State Machine Model

A state model describes a state machine which can be used to model the states a system component can be in and the transition from one state to the next state based on the events the system receives as input.

The elements of the class model are mainly borrowed from the UML class model so prior knowledge of UML modelling applies here.

Logical Data Model for the State Machine Model Elements

State Machine Elements

Example

Example State Model

State Machine (:state-machine)

A state machine is the root element for a state machine view. It contains the set of states and transistions as value of the :ct key.

States (:state, :start-state, :end-state)

A simple state machine has at least one start state, some normal states to model the different states a system can be in, and at least one end state.

A start state starts the state machine and an end state terminates the state machine.

States can be compound, they can have an internal state machine. This is modelled as a set of states and transitions in the :ct key, analog to the state machine itself.

Transitions (:transition)

A transition connects two states and models the input that leads to the transition from the current state (:from) to the next state (:to).

Forks and Joins (:fork-state, :join-state)

You can split a transition to trigger multiple new states with a fork state. A fork has a single input transition and multiple output transitions.

To join multiple transitions after a fork a join state is used. A join has multiple input transitions and a single output transition.

Code Model

A code model captures the static structure of the code.

The abstraction level of a code model is not very high compared to the actual implementation. Therfore modelling and updating a complete code model is not of much value. But code models of the domain can be very valuable as a means of communication between domain experts and developers to shape and document the domain model for a bounded context.

The elements of the code model are mainly borrowed from the UML class model so prior knowledge of UML modelling applies here.

Logical Data Model for the Code Model Elements

Code Model Elements

Packages/Namespace (:package, :namespace)

Packages and namespaces provide a hierarchical structure for the organisation of the elements of the code model.

Packages and namespaces are treated the same, so use what suits your system best.

Interfaces/Protocols (:interface, :protocol)

Interfaces and protocols specify related methods. Interfaces also provide a type for the static type system.

Interfaces and protocols are treated the same, so use what suits your system and inplementation language best.

Class (:class)

A class in object orientation is a typed element that encapsulates state and behaviour. The state is modelled with fields, the behaviour with methods.

In functional programming, you can use classes to model the values of your system.

Enumeration (:enum)

An enumeration is a typed enumeration of values.

Field (:field)

A field is part of the state of a class.

Method (:method)

A method is part of the behaviour of a class or an interface.

Function (:function)

A function is a first class element in functional programming. It has input parameters and calculates results.

Relations (:association :aggregation, :composition :inheritance :implementation :dependency)

Organization Model

The responsibility model captures the organizational structure of the system architecture.

Logical Data Model for the Organization Model Elements

Organization Model Elements

Organization Structure (:organization :org-unit)

An organization contains organizational units (e.g. branches and departments) and organizational units can cointain other organizational units. You can enhance the org-unit node with attributes specific for your organization (e.g. support mailboxes) and use these attributes in the templates to generate documentation or other artifacts.

Relations (:responsible-for, :collaborates-with)

The responsible-for relation captures the responsibility of an organizational unit for architecture or deployment nodes (e.g. a system or an Azure subscription).

Model Element Selection By Criteria

Model elements can be selected based on criteria. Criterias are given as a map where each key/value pair specifies a criterium for the selection. An element is selected, if it matches all criteria in the map (logical conjunction).

Criterias can also be given as a vector of criteria maps. An element is selected, if it is selected by any of the critria maps (logical disjunction).

Keys

keytypeexample valuesdescription
:key?vector[:tech true]elements for which the check for the key returns the value (useful for custom keys)
:keyvector[:tech "Clojure"]elements for which the lookup of the key returns the value (useful for custom keys)
:elkeyword:systemelements of the given type
:elsset of keywords#{:system :person}elements with one of the given types
:namespacestring"org.soulspace"elements with the given id namespace
:namespacesset of strings#{"org.soulspace"}elements with one of the given id namespaces
:namespace-prefixstring"org"elements with the given id namespace prefix
:from-namespacestring"org.soulspace"relations with the given id namespace of the from reference
:from-namespacesset of strings#{"org.soulspace"}relations with one of the given id namespaces of the from reference
:from-namespace-prefixstring"org"relations with the given id namespace prefix of the from reference
:to-namespacestring"org.soulspace"relations with the given id namespace of the from reference
:to-namespacesset of strings#{"org.soulspace"}relations with one of the given id namespaces of the from reference
:to-namespace-prefixstring"org"relations with the given id namespace prefix of the from reference
:id?booleantrue, falseelements for which the id check returns the given value
:idkeyword:org.soulspace/overarchthe element with the given id
:fromkeyword:org.soulspace/overarchrelations with the given from id
:tokeyword:org.soulspace/overarchrelations with the given to id
:subtype?booleantrue, falsenodes for which the subtype check returns the given boolean value
:subtypekeyword:queuenodes of the given subtype
:subtypesset of keywords#{:queue :database}nodes of one of the given subtypes
:maturity?booleantrue, falseelements for which the maturity check returns the given boolean value
:maturitykeyword:proposed, :deprecatedelements of the given maturity
:maturitiesset of keywords#{:implemented :deprecated}elements of the given maturities
:external?booleantrue, falseelements of the given external state
:name?booleantrue, falseelements for which the name check returns the given value
:namestring/regex"Overarch CLI"elements for which the name matches the given value
:desc?booleantrue, falseelements for which the description check returns the given value
:descstring/regex"CLI" "(?i).CLI."elements for which the description matches the given value
:doc?booleantrue, falseelements for which the documentation check returns the given value
:docstring/regex"CLI" "(?i).CLI."elements for which the documentation matches the given value
:tech?booleantrue, falseelements for which the technology check returns the given value
:techstring"Clojure"elements of the given technology
:techsset of strings#{"Clojure" "Java"}elements with one or more of the given technologies
:all-techsset of strings#{"Clojure" "Java"}elements with all of the given technologies
:tags?booleantrue, falseelements for which the tags check returns the given value
:tagstring"critical"elements with the given tag
:tagsset of strings#{"Clojure" "Java"}elements with one or more of the given tags
:all-tagsset of strings#{"Clojure" "Java"}elements with all of the given tags
:refers?booleantrue, falsenodes for which the check for refers returns the given value
:referred?booleantrue, falsenodes for which the check for referred returns the given value
:refers-tokeyword:org.soulspace/overarchnodes which refer to the node with the given id
:referred-bykeyword:org.soulspace/overarchnodes which are referred by the node with the given id
:child?booleantrue, falsenodes for which the check for child returns the given value
:child-ofkeyword:org.soulspace/overarchnodes which are children of the node with the given id
:descendant-ofkeyword:org.soulspace/overarchnodes which are descendants of the node with the given id
:parent?booleantrue, falsenodes for which the check for children returns the given value
:parent-ofkeyword:org.soulspace/overarchnode which is the parent of the node with the given id
:ancestor-ofkeyword:org.soulspace/overarchnodes which are ancestors of the node with the given id

Views

To show model elements in diagrams or in textual representations you can define different kind of views. The kind of view defines the visual rendering of the elements and the kind of elements and relations that are shown.

Overarch supports different types of views and renderings. For architecture and deployment models, C4 diagrams can be rendered. Use case models, state machine models and code models can be rendered as UML diagrams. Structure diagams can be used to show the hierarchical structure of organisations, systems or deployments.

Overarch supports the description of all C4 core and supplementary diagrams as views. The core C4 views form a hierarchy of views. See c4model.com for the rationale and detailed information about the C4 model and diagrams.

Overarch also supports conceptual views as part of the documentation of the system. Conceptual views can be used in early stages of the development project, when the requirements and the architecture are not yet fixed, to get an overview of the system to be designed. They can also be used to document the relevant concept ofthe domain of the system for discussion, onboarding and learning. Concepts should also be part of the glossary, as well as actors, systems and the applications and containers developed for the system.

Logical Data Model for the View Model Elements

View Model Elements

View Content Specification

In a specific view you can reference the model elements you want to include in this view. A Model element can be included in as many views as you want, but the element has to match the expected kinds of model elements to be shown. For example, a system landscape view renders person and system elements but no use cases or state machines, even if they are referenced in the view. Please consult the models for the model and view elements.

Model elements can be referenced directly via a :ref. They can also be selected via model criteria. Either References for selected nodes or nodes for selected references can be included automatically.

The views can reference elements from the model as their content. The content of a view should be a list instead of a set because the order of elements may be relevant in the rendering of a view.

Keys

keytypevaluesdescription
:elkeywordsee viewstype of the view
:idkeywordnamespaced idused for export file name
:titlestring rendered title
:specmapsee view specsrendering customization (e.g. styling)
:ctlistmodel refs (or elements)view specific keys possible

System Context View (:context-view)

Shows the system in the context of the actors and other systems it is interacting with. Contains users, external systems and the system to be described.

System Context View rendered with PlantUML

Container View (:container-view)

Shows the containers (e.g. processes, deployment units of the system) and the interactions between them and the outside world. Contains the elements of the system context diagram and the containers of the system to be described. The system to be described is rendered as a system boundary in the container diagram.

Container View rendered with PlantUML

Component View (:component-view)

Shows the components and their interactions inside of a container and with outside systems and actors.

Component View rendered with PlantUML

Code View (:code-view)

A code view is used to show the design of parts of the software. You can use it e.g. to model a domain and to communicate the model with domain experts.

Normally it is not neccessary to model the whole code base, the level of abstraction for implementation details is usually not high enough to justify modelling implemeted code. Also the speed of change in the code is most likely to high and renders a code model obsolete fast. If you want to visualize existing code, you can use the features of your IDE to generate a diagram of it. Maybe you can generate a code model from the existing code base.

On the other hand it can be useful to create a view of code not yet implemented. An UML class view can be used to model a domain or communicate a design pattern.

The data models shown here in the documentation are examples of code-views.

System Structure View (:deployment-structure-view)

The system structure view shows hierarchical composition of a system.

System Landscape Views (:system-landscape-view)

The system landscape view shows a high level picture, a broader view of the system landscape and the interactions of the systems.

System Landscape View rendered with PlantUML

Dynamic Views (:dynamic-view)

Shows the order of interactions. The relations get numerated in the given order and the nuber is rendered in the diagram. The :index attribute can be used on relations or refs to relations to set the number for a relation in the view.

If the relations are describing interactions specific for the diagram instead of general architectural relations (e.g. interfaces) of the model elements, it is ok to to specify the relations in the content of the view.

Deployment View (:deployment-view)

The deployment view shows the infrastucture and deployment of the containers of the system.

Deployment View rendered with PlantUML

Deployment Structure View (:deployment-structure-view)

The deployment structure view shows hierarchical composition of the infrastructure and deployments of the system.

State Machine Views (:state-machine-view)

A state machine view is used to show the different states a component can be in. It also shows the transitions between these states based on the input events the component receives.

Use Case View (:use-case-view)

A use case view is used to show the actors of the system under design and their goals using this system.

Use Case View rendered with PlantUML

Concept View (:concept-view)

The concept view is a graphical view. It shows the concepts as a concept map with the relations between the concepts.

Organization Structure View (:organization-structure-view)

The organization structure view shows hierarchical composition of a organization.

Glossary View (:glossary)

The glossary view is a textual view. It shows a sorted list of elements with their type and their descriptions.

View Content Selection and Customization

Views can be customized with the :spec key. View specs may include general directives for a view or directives for specific renderers (e.g. PlantUML).

keytypeexample valuesdescription
:includekeyword:relations :relatedspecify automatic includes (work in progress)
:selectionmap or vector{:namespace "banking"}select the content by criteria (see Model Element Selection)
:layoutkeyword:top-down, :left-rightrendering direction
:linetypekeyword:orthogonal, :polygonaldifferent line types for relations
:sketchbooleantrue, falsevisual clue for sketches
:stylessetsee Stylingvisual customization of elements
:themeskeywordid of the themetheme containing styles

Selection

With the :selection key a criteria map or a vector of criterias can be specified. The matching elements will be included in the view. This feature can be used to create 'dynamic' views that always contain the latest model content matching the criteria. See section Model Element Selection by Criteria for details and banking views for examples.

Includes

With the :include key elements can be automatically included in a view. The default behaviour is :referenced-only which only includes the referenced elements.

With the value :relations all relations to the referenced elements will be automatically included.

With the value :related all elements participating in the referenced relations will be automatically included in addidtion to the referenced elements.

Preference Rules for View Content

Criteria based selection, direct element references and includes can be combined in a view. First the selection is merged with the references in such a way, that key overrides and additions on references are preserved. Then the included elements are calculated and merged. This merge also preserves the key overrides and additions made on the references.

Therefore you can select the content with the :selection and :include keys and customize the rendering with direct references in the :ct vector of the view.

Styling

Overarch supports custom styles for elements. For an example see views.edn.

Keys

keytypevaluesdescription
:idkeywordnamespaced idused to reference styles
:forkeyword:rel, :elementelement type to be styled
:line-stylekeyword:dashed, :dotted, :boldline style for relations
:line-colorhex rgb#0000FF for bright blueline color for relations
:border-colorhex rgb#FF0000 for bright redborder color for nodes
:text-colorhex rgb#003300 for dark greentext color for names and descriptions
:legend-textstring meaningful text to show in legend

Rendering

Views can be rendered into different formats via the Overarch CLI tool. Views can also be rendered via templates, which allows full control over the output and allows new output formats without changes in Overarch itself (see Template Based Artifact Generation).

This section describes the rendering of views via the Overarch CLI tool.

PlantUML

The specified views C4 architecture and UML views can be rendered to PlantUML diagram specification (*.puml files). These can be rendered into different formats (e.g. SVG, PNG, PDF) with PlantUML.

You can specify PlantUML specific directives with the :plantuml key of a view spec.

   :spec {:plantuml {:sprite-libs [:azure :devicons]}}

Keys

keytypeexample valuesdescription
:node-separationinteger50 (for 50 pixels)separation between nodes
:rank-separationinteger250 (for 250 pixels)separation between ranks
:sprite-libsvectorsprite-lib keywordsused to render sprites for techs, order defines precedence of the libs
:skinparamsmap{"monochrome" "true"}render generic skinparams (as skinparam <key> <value>)

Sprite Support

Overarch supports PlantUML sprites to show a visual cue of the technology in the elements of a diagram. It does so by matching the value of the :tech key of a model element to the list of sprites. You can also directly add a :sprite key to the reference of a model element in a view. The explicit :sprite value takes precedence over the :tech value.

The list of sprites contains sprites of the PlantUML standard library, e.g. sprites for AWS and Azure. The mapping files from tech name to sprite reside in resources/plantuml.

Currently the following keys for sprite libs are supported:

  • :awslib14
  • :azure
  • :cloudinsight
  • :cloudogu
  • :devicons
  • :devicons2
  • :font-awesome-5
  • :logos

The command line interface supports the option --plantuml-list-sprites which prints the (long) list of sprite mappings.

Rendering PlantUML diagrams in VS Code

The Visual Studio Code PlantUML Extension allows previewing and exporting these diagrams right from the IDE.

PlantUML preview

PlantUML plugins also exists for major IDEs and build tools (e.g. IntelliJ, Eclipse, Maven, Leiningen).

GraphViz

The concept view can be exported as a concept map to a GraphViz *.dot file.

Rendering GraphViz diagrams

For GraphViz there are a few Visual Studio Code extensions available that allow previews of the generated Graphviz files.

The images can be created with the dot executable, which resides in the bin directory of the GraphViz installation.

You can specify Graphviz directives with the :graphviz key in a view spec. Currently only the configuration of the layout engine is supported.

Keys

keytypevaluesdescription
:enginekeyworde.g. :dot, :neato, :sfdpthe graphviz layout engine to use

Markdown

Markdown is used to render textual representations of the views. You can use converters to generate other formats like HTML or PDF from markdown.

You can specify Markdown directives with the :markdown key in a view spec.

Keys

keytypevaluesdescription
:referencesbooleantrue, falserender references for nodes
:diagrammap{:format "png"}render diagram image

Exports

JSON

The model and view descriptions can be exported to JSON to make their content available to languages for which no EDN implementation exists. The export converts each EDN file in the model directory to JSON.

Structurizr (experimental)

Structurizr is a tool set created by Simon Brown. The Structurizr export creates a workspace with the loaded model and views.

As Structurizr currently only supports the C4 architecture model and views, only these elements will be included in the Structurizr workspace.

Template Based Artifact Generation

Overarch can generate artifacts for model elements and views via templates. The use cases of the tempates range from reports up to automatic code generation. Overarch supports forward engineering protected areas for manually written content in generated artifacts

Generation Configuration

You can configure the generation of artifacts with an EDN file. The configuration contains a vector of generation context maps. A generation context map specifies a selection of model elements, a template to use, how the template should be applied, and where the resulting artifact should be created.

keytypevaluesdefaultdescription
:selectionCRITERIA{:el :system} Criteria to select model elements
:view-selectionCRITERIA{:el :context-view} Criteria to select views
:templatePATH"report/node.cmb" Path to the template relative to the template dir
:engine:keyword:comb:combsciThe template engine to use (currently just :comb and :combsci)
:encodingstring"UTF-8""UTF-8"The encoding of the result artifact
:per-elementbooleantrue/falsefalseApply the template for each element of the selection
:per-namespacebooleantrue/falsefalseApply the template for each namespace of the selection
:subdirstring"report" Subdirectory for generated artifact under the generator directory
:namespace-prefixstring"src" Prefix to the namespace to use as path element
:base-namespacestring Base namespace to use as path element
:namespace-suffixstring"impl" Suffix to the namespace to use as path element
:prefixstring"Abstract" Prefix for the filename
:base-namestring Base of the filename
:suffixstring"Impl" Suffix for the filename
:extensionstring"md" "clj" "java" Extension for the filename
:filenamestring"README.md" Specific filename to use
:id-as-namespacebooleantrue/falsefalseUse the element id as the namespace for path generation
:id-as-namebooleantrue/falsefalseUse the name part of the element id as the name for path generation
:protected-areasstring"PA" Marker for protected areas in the template/artifact
:debugbooleantrue/falsefalsePrint parsed template on error

You can add additional (namespaced) keys to the generation context map, which are available in via the ctx symbol in the template.

The model elements, to which a template is applied, are selected via criteria with the selection key. A template can be applied to the collection of selected elements or to each element of the collection.

Templates can also be applied on views selected by criteria with the :view-selection key. The view selection also returns a collection of views, even if there is only one view selected. So use the :per-element key to enable the generation on a view level.

Example Templates

Example templates can be found in the templates folder. The provided templates are usable as is, but they also provide a starting point for organization or project specific templates.

As the model is extensible via custom keys and this information is provided for the templates, custom templates can make use of all those custom information in the model. For example, if containers have a custom key :my-org/repository-url, you can use this information in the documentation templates to provide a link to the repository of each container.

Example Generation Config File

[;; Report for all systems in the banking namespace
 {:selection {:namespace "banking" :el :system} ; selection criteria for the model elements
 :template "node-report.md.cmb"  ; relative path of the template to apply
 :title "Banking Systems Report" ; title of the report
 :engine :comb                   ; the template engine to use (currently only :comb)
 :encoding "UTF-8"               ; artifact encoding
 :per-element false              ; apply the template for each element of the selection or on the selection as a whole
 :subdir "reports"               ; subdirectory for generated artifact
 ; :namespace-prefix ""          ; prefix for the namespace of the generated artifact
 :base-namespace "systems"       ; base namespace of the generated artifact
 ; :namespace-suffix ""          ; suffix for the namespace of the generated artifact
 ; :prefix ""                    ; prefix for the name of the generated artifact
 :base-name "systems-report"     ; base name of the generated artifact
 ; :suffix ""                    ; suffix for the name of the generated artifact
 :extension "md"                 ; extension of the generated artifact
 :id-as-namespace false          ; use the name as the namespace of the artifact
 ; :protected-area "PA"          ; protected area prefix
 }

;; Report for the REST interfaces in the model
{:selection {:el :model-relation :techs #{"REST"}} ; selection criteria for the model elements
 :template "rel-report.md.cmb"   ; relative path of the template to apply
 :title "REST Interface Report"  ; title of the report
 :engine :comb                   ; the template engine to use (currently only :comb)
 :encoding "UTF-8"               ; artifact encoding
 :per-element false              ; apply the template for each element of the selection or on the selection as a whole
 :subdir "reports"               ; subdirectory for generated artifact
; :namespace-prefix ""           ; prefix for the namespace of the generated artifact
 :base-namespace "interfaces"    ; base namespace of the generated artifact
; :namespace-suffix ""           ; suffix for the namespace of the generated artifact
; :prefix ""                     ; prefix for the name of the generated artifact
 :base-name "systems-report"     ; base name of the generated artifact
; :suffix ""                     ; suffix for the name of the generated artifact
 :extension "md"                 ; extension of the generated artifact
 :id-as-namespace false          ; use the name as the namespace of the artifact
; :protected-area "PA"           ; protected area prefix
}]

Overarch CLI

The relevant CLI options for template based artifact generation are

  -m, --model-dir PATH           models     Models directory or path
  -T, --template-dir DIRNAME     templates  Template directory
  -g, --generation-config FILE              Generation configuration
  -G, --generation-dir DIRNAME   generated  Generation artifact directory
  -B, --backup-dir DIRNAME       backup     Generation backup directory

Example using a config file in the current directory and default directories

java -jar overarch.jar -g gencfg.edn

Comb Template Engine

Overarch uses the Comb template syntax by James Reeves.

Comb is a simple templating system for Clojure. You can use Comb to embed fragments of Clojure code into a text file.

Syntax

Clojure fragments in a template are demarkated with <% and %>. You can embed clojure code as an expression, where the result of the execution is included in the resulting artifact. You can also embed the clojure code as a control structure, where the result of the execution of the control structure is not included in the resulting artifact, only the template text or other expressions inside of the control structure.

Expressions

1 + 2 = <%= (+ 1 2) %>

Result:

1 + 2 = 3

Control structures

foo<%
(dotimes [x 3]
%> bar<%
)
%>

Result:

foo bar bar bar

API for Templates

In the comb templates you can use most of the functions of the clojure.core namespace. Additionally the functions of clojure.set and clojure.string are provided under the aliases set and str, e.g. set/union or str/join.

Overarch also provides functions to query and navigate the model under the m alias, e.g. m/resolve-element.

Security Considerations

Comb templates evaluated with :comb are compiled and can contain arbitrary clojure code, which gets evaluated in the context of the overarch process. Be aware of this fact and review templates accordingly, especially when using templates from external sources.

When the comp templates are evaluated by the :combsci engine, they are interpreted with Babashka SCI, the small clojure interpreter. This has many advantages:

  • sandboxed evaluation
  • control over the the available clojure namespaces and symbols
  • independent from the method of providing and starting overarch

As the template code is interpreted with SCI and not compiled, the generation might be a bit slower then with using compiled templates. But the advantages outweight the performance penalty by far.

Protected Areas

Protected areas are used to protect manually inserted text in generated artifacts. For example, when generating source code from a code model element, maybe only the signature of the function may be generated. The body of the function may have to be inserted by a programmer.

When regenerating the source code artifact, you don't want the manually inserted code to be deleted or overridden, but preserved.

Given this class node from a model

{:el :class
 :id :model/calc
 :name "Calc"
 :ct [{:el method
       :name "square"
       :type "double"
       :visibility :public
       :ct [{:el :field
             :name "x"
             :type "double"}
           ]}
     ]}

and a template like

public class <%= (:name e) %> {

  <% (doseq [m (:ct e)] %>
  public <%= (:type m) %> <%= (:name m) %>(<%
  (doseq [p (:ct m)] %> (:type m) %> <%= (:name m) %>, <%)%>) {
    // PA-BEGIN(square-impl)
    <%= (:square-impl protected-areas)%>
    // PA-END(square-impl)
  }
  <%)%>
}

On the first generation pass, the generated file will look like

public class Calc {

  public double square(double x) {
    // PA-BEGIN(square-impl)
    // PA-END(square-impl)
  }
}

After manually inserting the implementation, the artifact looks like

public class Calc {

  public double square(double x) {
    // PA-BEGIN(square-impl)
    return x * x;
    // PA-END(square-impl)
  }
}

On regeneration, the content of the protected area is parsed by the generator before applying the template and reinserted by the template.

So after regeneration the artifact still looks like

public class Calc {

  public double square(double x) {
    // PA-BEGIN(square-impl)
    return x * x;
    // PA-END(square-impl)
  }
}

Can you improve this documentation? These fine people already did:
Ludger Solbach & Solbach, Ludger (FDP4_EXTERN)
Edit on GitHub

cljdoc is a website building & hosting documentation for Clojure/Script libraries

× close