RFE: issue (optional) warning when '==' is used on wrapper types

[dashorst]

I would like that Eclipse (similar to IDEA) would issue an optional warning when one uses the following construct:

Long x1 = 12345L;
Long x2 = 67890L;
if(x1 == x2)
    System.out.println("Equal");
else
    System.out.println("Not equal");

Because (as apparently we don't all know) Long is a wrapper class and using == only works for Long values between -127 and 127 due to caching (similarly for the other wrapper types).

This would at least uncover one such instance in our code base, of which I suspect to find a greater number alas.

[iloveeclipse]

What does javac report?
Note, static code analysis of semantically correct and compilable code is not necessarily what compiler is supposed to do.

[dashorst]

While I sympathise with the "don't bloat the compiler with too much work" (that would make it slower etc), I doubt that the ECJ "Method can be static" optional problem is something a compiler should do as well, but causes fewer programming errors.

In any case, javac (17) doesn't issue a warning for either using == on Longs nor Strings.

I don't propose to emit a warning always, but make it an option similar to x == x (which isn't flagged by javac either).

[iloveeclipse]

I don't propose to emit a warning always, but make it an option similar to x == x (which isn't flagged by javac either).

Sure. If you (or anyone else) is interested to provide this feature, feel free to contribute.

[dashorst]

Well, here you go!

[dashorst]

The PR builds correctly, I'd love some feedback on whether it is complete, hits the correct spots and suitability for adoption.