replace deprecated Constructor based initialization of primitive type…

[eklaDFF]

…'s Wrapper Class with WrapperClass.valueOf(value)

[cyrille-artho]

I thought that the first half of the test still works? If we can keep the part with the explicit cast, that would be preferable.

[eklaDFF]

Explicit means Integer in = new Integer () ?

[cyrille-artho]

If the part above works, we can start comment out this part.
I think it is nicer to use /* here, akin to the commented-out part below. Definitely not // in addition to /*.
With /*.../*, we clearly mark two blocks of code as being parts of the test that do not (or do no longer) work. With //, we lose that information.

[eklaDFF]

Actually, none of the Assertion works.

Pavel taught me that Local Variable and Local Attribute are different entities in JPF.
Constructor type initialisation passes the value of Local Variables to Local Attributes but in case of new method it doesn't.

[eklaDFF]

There are two assert statements and none of them work. Pavel suggested that this might be due to Local Variables and Local Attributes miscommunication in new method.

…

On Fri, 2 Aug, 2024, 12:27 pm cyrille-artho, ***@***.***> wrote: ***@***.**** requested changes on this pull request. ------------------------------ In src/tests/gov/nasa/jpf/test/mc/basic/AttrsTest.java <#488 (comment)> : > + + + + + // Constructor based initialization is deprecated in Java 17. So we have to replace with "WrapperClass.valueOf(...)" based initialization. + // This below test has been commented out because we have an error when we replace the old constructor type initialization of Wrapper Class of primitive type int with "WrapperClass.valueOf(..)." + // Example : { Integer o = new Integer(v) } works but { Integer o = Integer.valueOf(v) } don't. + +// @SuppressWarnings("deprecation") +// @test public void testInteger() { +// if (verifyNoPropertyViolation()) { +// int v = 42; +// Verify.setLocalAttribute("v", 4200); +// +// // explicit +// Integer o = Integer.valueOf(v); I thought that the first half of the test still works? If we can keep the part with the explicit cast, that would be preferable. ------------------------------ In src/tests/gov/nasa/jpf/test/mc/basic/AttrsTest.java <#488 (comment)> : > + // This below test has been commented out because we have an error when we replace the old constructor type initialization of Wrapper Class of primitive type int with "WrapperClass.valueOf(..)." + // Example : { Integer o = new Integer(v) } works but { Integer o = Integer.valueOf(v) } don't. + +// @SuppressWarnings("deprecation") +// @test public void testInteger() { +// if (verifyNoPropertyViolation()) { +// int v = 42; +// Verify.setLocalAttribute("v", 4200); +// +// // explicit +// Integer o = Integer.valueOf(v); +// int j = o.intValue(); +// int attr = Verify.getLocalAttribute("j"); +// assert attr == 4200; +// +// // semi autoboxed If the part above works, we can start comment out this part. I think it is nicer to use /* here, akin to the commented-out part below. Definitely not // in addition to /*. With /*.../*, we clearly mark two blocks of code as being parts of the test that do not (or do no longer) work. With //, we lose that information. — Reply to this email directly, view it on GitHub <#488 (review)>, or unsubscribe <https://github.com/notifications/unsubscribe-auth/ASVOVF7M6N4U6NFK6NUBUWDZPMUVXAVCNFSM6AAAAABL3GS4H6VHI2DSMVQWIX3LMV43YUDVNRWFEZLROVSXG5CSMV3GSZLXHMZDEMJUHAYDCMJRHE> . You are receiving this because you authored the thread.Message ID: ***@***.***>

[cyrille-artho]

I see. In this case, please modify this PR to first retain the problematic test as is (with no code changes, keeping the code that produces compiler warnings).
I think it might be better to suppress the remaining compiler warnings in that problematic test in a second PR, to avoid commenting out or removing a lot of code.

[eklaDFF]

Made those changes, please review

[cyrille-artho]

Great, thanks!