Unclear parse error message when missing equation

[Peter Aronsson]

When you forget to add the equation keyword for an equation section you get a strange parse error message.

model Test

Real x;

x = 1;

end Test;

Gives:
No viable alternative near token 'x'

It would be nice if the error message gave a hint that perhaps 'equation' was missing.

[Peter Aronsson]

MathCore track: ​http://intranet/trac/mathmodelica/ticket/3258

[Martin Sjölund]

I'll see if we can do something about this, but it probably won't say that equation was missing since this looks more like it should be corrected to:
x var = 1; or ty x = 1;

Maybe one could try to add equation keyword in an extra backtracking phase, but I fear that would slow down the parser quite a bit or make it ambiguous

[Peter Aronsson]

Some other examples:

model test "Missing 'equation'"

Pin p,n;

connect(p,n);
end test;

model test2 "Missing 'algorithm'"

Real x;

x := x+1;
end test2;

model test3 "Missing 'algorithm'"

Real x;

Modelica.Utilities.Streams.print(String(x));
end test3;

[Martin Sjölund]

[MissingEquation.mo:5:3-5:9:writable] Error: Parse error: Found the start of a connect equation but expected an element (are you missing the equation keyword?)

But the others require backtracking and I think we should accept the error ANTLR chooses for us.

[Per Östlund]

Closing this since it doesn't seems like something that will ever be fixed. Syntax errors are hard to give good errors for since you don't know what the user intended. In this case it could be a missing equation or algorithm, but it could also be a missing type before x or something else entirely.