Umple doesn't parse a guard with a string variable and equal operation

[vahdat-ab]

consider the following example"

class CommandControl{
    String cmdString = "";

    CommandControl{
        IDLE{
            execute -> CHECK_AVAILABILITY;
        }
        CHECK_AVAILABILITY{
        //The issue is with =="" and !="". When I remove !="" or =="" it works
        execute [cmdString!=""]-> COMMAND_VALIDITY;
        //  execute [cmdString==""] -> IDLE;

        //  execute [cmdString]-> COMMAND_VALIDITY;
        //  execute [cmdString] -> IDLE;
        }
        COMMAND_VALIDITY{
            [isCommandValid(cmdString)] -> EXECUTION;
        }
        EXECUTION{
        }
    }
}

It doesn't compile because of cmdString!="". When the !="" is removed it is comp[iled.

[vahdat-ab]

@opeyemiAdesina

[TimLethbridge]

Yes it should use equals() when the comparison is string or some other non-primitive data.

[vahdat-ab]

I think it must be parsed because if we give correct rules, there is no problem. Using equals() is just a programming level concept and normally people won't use it at the modeling level.
if we parse it correctly, the analysis phase can take care of correctness of the variable 9e.g. if it's been defined or it's an Integer or String). I think we should avoid bringing programming concepts to the modeling because they are hard to interpret.

[TimLethbridge]

Yes of course parse =. What I meant was convert it to generate .equals() internally in generated code. The user would never see that.

[tlaport4]

This issue seems to have already been fixed. The generated code uses .equals to compare Strings.

In this example, the conditions in the generated code look like this:

 if (!"".equals(getCmdString()))

and

 if ("".equals(getCmdString()))

Should I just add some tests for this specific situation?

[vahdat-ab]

Yes. Just add test cases and then we will close this issue.