Pre and Post Conditions are suggested for "value != null" even when there is already a condition for !string.IsNullOrWhiteSpace(value)

[jmckone3]

While this doesn't create incorrect behavior, it does create a lot of noise and make the true warnings much less valuable.

[SergeyTeplyakov]

Thanks a lot.

Could you provide a version of Code Contract that you're using?

[SergeyTeplyakov]

I've tried with string.IsNullOrEmpty and got the same result. Could you please check with !string.IsNullOrEmpty as well? This can simplify investigation...

[jmckone3]

I am using the latest RC - DotNet.Contracts 1.10.20606.1

Yes. I also get the same issue with string.IsNullOrEmpy()

[yaakov-h]

What version of .NET Framework are you targetting?

This smells like #339 to me.

[jmckone3]

.Net Framework is 4.5.1.1

[yaakov-h]

How did you get a .NET 4.5.1.1 target framework? As far as I know, that doesn't exist.

[KAW24]

I can reproduce this using String.IsNullOrEmpty as follows:

public class Test
{
    private readonly string m_name;

    public Test(string name)
        : base()
    {
        Contract.Requires(!string.IsNullOrEmpty(name));
        m_name = name;
    }

    public string Name
    {
        get
        {
            Contract.Ensures(!string.IsNullOrEmpty(Contract.Result<string>()));
            // warning : CodeContracts: Consider adding the postcondition
            // Contract.Ensures(Contract.Result<System.String>() != null); to provide extra-
            // documentation to the library clients
            return m_name;
        }
    }

    [ContractInvariantMethod]
    private void ObjectInvariant() => Contract.Invariant(!string.IsNullOrEmpty(m_name));
}

I'm not sure this is related to #339 since there is no mention of [Pure] in the warning.

This was tested with VS 2015 Update 2 and Code Contracts 1.9.10714.2. The warning occurs for me when targeting both 4.6 and 4.6.1.

[SergeyTeplyakov]

It seems that cccheck can't infer relationship between nullness of the variable from the predicate !string.IsNullOrEmpty(s)...

Valid bug and it seems that this is not a regression. We'll see what we can do.

Up for grabs:)

[jmckone3]

Was using .NET Fx 4.5.1
Sorry about the typo.

Thanks for confirming the issue Sergey.