lenarb / crml-compiler Goto Github PK

An issues to keep track of evaluate and categories, will update with progress

A placeholder to keep track of all the current issues with sets, current problem with mapping all operators on sets and propagating sets in all expression evaluation.

An issue was opened here: OpenModelica/OpenModelica#12245

@lenaRB : Please find here #57 new tests for evaluating categories.

There is no keyword new in BooleanConstructors.crml
It would make it more readable and easy to parse if conversions used "new", but we can discuss it of course.

Update/add some examples to test the new time variable once it is added to the compiler

model OperatorExample2 is {
	// Definition of disjunction of two Booleans.
	Operator [ Boolean ] Boolean b1 or Boolean b2 = not (not b1 and not b2);
	// Example of function call
	Boolean b1 is external;
	Boolean b2 is external;
	Boolean b is b1 or b2;
};

but or is also defined in the specification as a built in operator, so it does not parse correctly, plus operator names should be in '', so the correct syntax is 'or' for a user defined operator

Update/add some examples to test external variables.

Are both of them useful?

@lenaRB : The test report says that the "CheckOver_no_ext" example is successful but the translation of "check over" operator seems to me as still not finished. Is that normal?
I don't know if it requires a lot of effort from but it would be better for the user to stop the compilation in that case and say that such instructions are not supported yet.

Create a folder to host CRML language specification as pdf (since html conversion is still under progress) as well as other publications made on CRML that are accessible online.

Status reported in test_report.html is not correct for some of the tests.
See in particular CheckOver example below that is mentioned as successful although no verification model has been provided (and hence no simulation could be verified).

update definition of 'check count' operators when traffic-light example will be supported by the compiler.

@audrey-jardin Should the variables here be defined as Booleans or am I reading the syntax wrong?

@adrpo: (not urgent at all) OMEdit icons "open a file" and "new file" should be updated to support opening and creation of files with .crml extension

@lenaRB : please find here #59 first reference results for some ETL operators. More will come asap.

Modelica code generated for the spec-doc-examples\BooleanAccumulation.crml file needs to be fixed.
BooleanAccumulation.zip

Ideally we would find a way of running the docker image from JUnit, but there needs to be a way that does not reload it everytime otherwise it is ridiculously inefficient

Otherwise some script that runs all the simulated models

One issue to consider is that a lot of them rely on external variables, so we might need to connect them in some batch way.

When loading CRML.mo library, error message appears and nothing happens (probably due to dependencies to other libraries such as Modelica 3.2.2 and Modelica_StateGraph.

@adrpo : could be useful to automatically load CRMLtoModelica when the user translates crml file in OMEdit as the generated file requested it to be simulated.

why is 'new' not used when creating a new period? I am still a bit fuzzy on when we use new and when it is implicit. Is it when we use it for type conversion? Like new Clock my_boolean?

A place holder to keep track of all the clock related issues

See the generated Modelica code:
BooleanPrimitives.zip

and the error message from OMEdit:

only CRML code is released yet

RealExp.crml - there is exp which should be probably defined as something concrete

@adrpo : on Windows machine, option "translate as" is working but not option "translate" probably due to path mixing.

@lenaRB : new ticket just to keep it in mind

For some tests, especially ones without external variables it would be easiest to have the expected value as some kind of annotation instead of a whole reference file with results, because we only care about the value of the requirement for example.

Then we should just have an annotation for that in the relevant tests? var = b, expectedResult = true
@adrpo, @audrey-jardin what do you think?