Skip to content

MATLAB and GNU Octave

The +opensysml client calls the service using Connect-JSON. It is source distributed for MATLAB R2019b+ and GNU Octave 7+; use it when model workflows are already built around either environment.

Install

Add the package directory to the MATLAB or Octave path:

addpath('client/matlab')

MATLAB uses matlab.net.http; Octave uses a curl subprocess.

First model

parseSource accepts inline SysML. Check the parsed model before evaluating the mass of the Demo::sedan subject:

addpath('client/matlab')

source = sprintf(['package Demo {\n' ...
    '  part def Vehicle {\n' ...
    '    attribute mass default = 1500.0;\n' ...
    '  }\n' ...
    '  part sedan : Vehicle {\n' ...
    '    attribute :>> mass = 1800.0;\n' ...
    '  }\n' ...
    '}']);

conn = opensysml.connect();
cleanup = onCleanup(@() conn.close());
model = opensysml.parseSource(conn, source, 'name', 'demo.sysml');
if ~model.ok()
    disp(opensysml.diagnostics(model))
    error('The model has errors.')
end

mass = opensysml.evaluate(model, 'mass', 'subject', 'Demo::sedan');
fprintf('%.1f\n', mass);

1800.0

Connection can start a private child or connect to an external address; Model also supports symbol lookup, instances, verification, execution, analysis, editing, conversion and document queries.

Service source

MATLAB never downloads a service binary. Build or install sysml-grpc and provide it through OPENSYSML_GRPC_BINARY, the shared cache or $PATH; see the service-binary reference. In Octave builds without Java, connect to an external service instead of starting a child.

Parse diagnostics are available from the model, and failures use typed MException identifiers under the opensysml:* namespace.

Next steps