Constreofy is a tool-set that obtains the formal semantics of a Reo network by utilizing constraint satisfaction techniques. It works by constructing a constraint satisfaction problem from the conjunction of the constraints, which describe the behavior of a Reo element and by using REDUCE Algebra system to solve the constraints. Then, it builds an automaton, which visualizes the solutions to the constraint. It uses GraphViz to represent the automaton.
The tool provides API to define a Reo network programmatically.
public ConstraintConnector example() {
ConnectorFactory factory = new ConnectorFactory();
ConstraintConnector connector = factory.prioritySync("a", "b");
connector.add(factory.merger("c", "d", "e"), "c", "b");
connector.add(factory.sync("f", "g"), "f", "a");
connector.add(factory.merger("h", "i", "j"), "h", "g");
return connector;
}It is also possible to define a model in JSON format. The schema for the input format is here. To facilitate the definition of a Reo network in the JSON format, we are building a GUI editor based on PixiJS.
Constreofy works by mapping each Reo element to a binary constraint that encodes its behavior and constructing a constraint from the conjunction of these constraints. Then, it launches the Reduce program that is installed on the machine, loads the redlog package, sets the context to IBALP, (Initial Boolean algebras, basically quantified propositional logic.) using the following commands:
load_package "redlog";
rlset ibalp;
The variables used in the constraints are declared using rlpcvar .
The main constraint is provided to Reduce as constraint_name := constraints .
The command rldnf constraint_name calculated the DNF (Disjunctive Normal Form) of the given constraint.
If the constraint is not data-aware, the solutions to the constraints are extracted from the DNF form in a straight-forward manner.
Otherwise, prior to solving the constraint, the data-aware expressions are replaced by new binary predicates whole validity is obtained from the DNF. A new data-aware constraint is formed and solved using Reduce.
Constreofy constructs an automaton from the solutions and writes it in the dot format, for instance:
bc -> bc [ label="L1:{}"];
empty -> bc [ label="L2:{b2c}"];
bc -> empty [ label="L3:{bc1}"];
empty -> empty [ label="L1:{}"];
}
The automata corresponding to a FIFO1 channel named bc with the source end named b2c and sink end called bc1 is as follows:
The below is a nontrivial Reo network.
which is mapped to the following automata using Constreofy tool
Click to expand the json definition of the above example.
{"nodes":[{"Replicate":{"name":"A",
"ends":[{"name":"A1",
"type":"Source"},
{"name":"A2",
"type":"Sink"}]}},
{"Replicate":{"name":"B",
"ends":[{"name":"B1",
"type":"Source"},
{"name":"B2",
"type":"Sink"},
{"name":"B3",
"type":"Sink"}]}},
{"Route":{"name":"C",
"ends":[{"name":"C1",
"type":"Source"},
{"name":"C2",
"type":"Sink"},
{"name":"C3",
"type":"Sink"},
{"name":"C4",
"type":"Sink"}]}},
{"Route":{"name":"D",
"ends":[{"name":"D1",
"type":"Source"},
{"name":"D2",
"type":"Sink"},
{"name":"D3",
"type":"Sink"},
{"name":"D4",
"type":"Sink"}]}},
{"Route":{"name":"E",
"ends":[{"name":"E1",
"type":"Source"},
{"name":"E2",
"type":"Sink"},
{"name":"E3",
"type":"Sink"},
{"name":"E4",
"type":"Sink"}]}},
{"Merge":{"name":"F",
"ends":[{"name":"F1",
"type":"Source"},
{"name":"F2",
"type":"Source"},
{"name":"F3",
"type":"Sink"}]}},
{"Replicate":{"name":"G",
"ends":[{"name":"G1",
"type":"Source"},
{"name":"G2",
"type":"Sink"}]}},
{"Replicate":{"name":"H",
"ends":[{"name":"H1",
"type":"Sink"},
{"name":"H2",
"type":"Source"}]}},
{"Replicate":{"name":"I",
"ends":[{"name":"I1",
"type":"Source"},
{"name":"I2",
"type":"Sink"}]}},
{"Replicate":{"name":"J",
"ends":[{"name":"J1",
"type":"Source"},
{"name":"J2",
"type":"Sink"},
{"name":"J3",
"type":"Sink"},
{"name":"J4",
"type":"Sink"},
{"name":"J5",
"type":"Sink"},
{"name":"J5",
"type":"Sink"}]}},
{"Replicate":{"name":"K",
"ends":[{"name":"K1",
"type":"Source"},
{"name":"K2",
"type":"Sink"}]}},
{"Replicate":{"name":"L",
"ends":[{"name":"L1",
"type":"Sink"},
{"name":"L2",
"type":"Source"}]}},
{"Replicate":{"name":"M",
"ends":[{"name":"M1",
"type":"Sink"},
{"name":"M2",
"type":"Source"}]}},
{"Replicate":{"name":"N",
"ends":[{"name":"N1",
"type":"Source"},
{"name":"N2",
"type":"Sink"}]}},
{"Replicate":{"name":"O",
"ends":[{"name":"O1",
"type":"Sink"},
{"name":"O2",
"type":"Source"}]}},
{"Replicate":{"name":"P",
"ends":[{"name":"P1",
"type":"Source"},
{"name":"P2",
"type":"Sink"}]}},
{"Replicate":{"name":"Q",
"ends":[{"name":"Q1",
"type":"Source"},
{"name":"Q2",
"type":"Sink"},
{"name":"Q3",
"type":"Sink"},
{"name":"Q4",
"type":"Sink"},
{"name":"Q5",
"type":"Sink"}]}},
{"name":"S",
"ends":[{"name":"S1",
"type":"Source"},
{"name":"S2",
"type":"Sink"}]},
{"name":"T",
"ends":[{"name":"T1",
"type":"Source"},
{"name":"T2",
"type":"Sink"}]},
{"name":"U",
"ends":[{"name":"U1",
"type":"Sink"},
{"name":"U2",
"type":"Source"}]}],
"readers":[{"name":"R1",
"ends":[{"name":"R12",
"type":"Sink"}]},
{"name":"R2",
"ends":[{"name":"R22",
"type":"Sink"}]},
{"name":"R3",
"ends":[{"name":"R32",
"type":"Sink"}]},
{"name":"R4",
"ends":[{"name":"R42",
"type":"Sink"}]}],
"writers":[{"name":"W1",
"ends":[{"name":"W11",
"type":"Source"}]},
{"name":"W2",
"ends":[{"name":"W21",
"type":"Source"}]},
{"name":"W3",
"ends":[{"name":"W31",
"type":"Source"}]}],
"channels":[{"Sync":[{"Source":"A2B"},
{"Sink":"AB1"}]},
{"FIFO":[{"Source":"B2C"},
{"Sink":"BC1"}]},
{"FIFO":[{"Source":"C2D"},
{"Sink":"CD1"}]},
{"Sync":[{"Source":"D2F"},
{"Sink":"DF1"}]},
{"FIFO":[{"Source":"F3G"},
{"Sink":"FG1"}]},
{"FIFO":[{"Source":"B3E"},
{"Sink":"BE1"}]},
{"Sync":[{"Source":"E2F"},
{"Sink":"EF2"}]},
{"SyncDrain":[{"Source":"E4T"},
{"Source":"ET2"}]},
{"Sync":[{"Source":"Q2T"},
{"Sink":"QT1"}]},
{"PrioritySync2":[{"Source":"S2Q"},
{"Sink":"SQ1"}]},
{"FIFO":[{"Source":"Q5P"},
{"Sink":"QP1"}]},
{"Lossy":[{"Source":"Q3O"},
{"Sink":"QO2"}]},
{"Lossy":[{"Source":"Q4H"},
{"Sink":"QH2"}]},
{"SyncDrain":[{"Source":"O1D"},
{"Source":"OD4"}]},
{"SyncDrain":[{"Source":"H1C"},
{"Source":"HC4"}]},
{"SyncDrain":[{"Source":"C3M"},
{"Source":"CM1"}]},
{"SyncDrain":[{"Source":"D3U"},
{"Source":"DU1"}]},
{"Lossy":[{"Source":"J3M"},
{"Sink":"JM2"}]},
{"Lossy":[{"Source":"J6U"},
{"Sink":"JU2"}]},
{"PrioritySync1":[{"Source":"I2J"},
{"Sink":"IJ1"}]},
{"FIFO":[{"Source":"J2K"},
{"Sink":"JK1"}]},
{"Sync":[{"Source":"J4L"},
{"Sink":"JL2"}]},
{"FIFO":[{"Source":"J5N"},
{"Sink":"JN1"}]},
{"SyncDrain":[{"Source":"E3L"},
{"Source":"EL1"}]}],
"connections":[{"one":"W11",
"two":"I1"},
{"one":"I2",
"two":"I2J"},
{"one":"W21",
"two":"A1"},
{"one":"A2B",
"two":"A2"},
{"one":"AB1",
"two":"B1"},
{"one":"B2C",
"two":"B2"},
{"one":"B3E",
"two":"B3"},
{"one":"BC1",
"two":"C1"},
{"one":"C2D",
"two":"C2"},
{"one":"E3L",
"two":"C3"},
{"one":"A2B",
"two":"HC4"},
{"one":"CD1",
"two":"D1"},
{"one":"D2F",
"two":"D2"},
{"one":"D3U",
"two":"D3"},
{"one":"OD4",
"two":"D4"},
{"one":"BE1",
"two":"E1"},
{"one":"E2F",
"two":"E2"},
{"one":"E3L",
"two":"E3"},
{"one":"E4T",
"two":"E4"},
{"one":"DF1",
"two":"F1"},
{"one":"EF2",
"two":"F2"},
{"one":"F3G",
"two":"F3"},
{"one":"FG1",
"two":"G1"},
{"one":"R32",
"two":"G2"},
{"one":"H1C",
"two":"H1"},
{"one":"QH2",
"two":"H2"},
{"one":"A2B",
"two":"A2"},
{"one":"J2K",
"two":"J2"},
{"one":"J3M",
"two":"J3"},
{"one":"J4L",
"two":"J4"},
{"one":"J5N",
"two":"J5"},
{"one":"J6U",
"two":"J6"},
{"one":"JK1",
"two":"K1"},
{"one":"R12",
"two":"K2"},
{"one":"CM1",
"two":"M1"},
{"one":"JM2",
"two":"M2"},
{"one":"JN1",
"two":"N1"},
{"one":"R22",
"two":"N2"},
{"one":"O1D",
"two":"O1"},
{"one":"QO2",
"two":"O2"},
{"one":"A2B",
"two":"A2"},
{"one":"QP1",
"two":"P1"},
{"one":"R42",
"two":"P2"},
{"one":"W31",
"two":"S1"},
{"one":"S2Q",
"two":"S2"},
{"one":"QT1",
"two":"T1"},
{"one":"ET2",
"two":"T2"},
{"one":"DU1",
"two":"U1"},
{"one":"JU2",
"two":"U2"}
]}Wanna get involved? There are a few things you can help to improve the tool.
- Dockerize the tool for easy installation
- Implement a graphical editor for defining input Reo models (work in progress)
![dependabot[bot] avatar](https://avatars.githubusercontent.com/in/29110?v=4 "dependabot[bot]")
React
Typescript
Django
Laravel
Facebook
Microsoft
Google
Alibaba
D3
Tencent