Fix #13202 FP oppositeInnerCondition for string defines #6889
Conversation
lib/astutils.cpp
Outdated
| if (macro) { | ||
| if (tok1->isExpandedMacro() != tok2->isExpandedMacro()) | ||
| return false; | ||
| if (tok1->isExpandedMacro() && tok1->originalName() != tok2->originalName()) |
There was a problem hiding this comment.
if (MACRO == MACRO) should still result in a warning
There was a problem hiding this comment.
hmm I am not sure but I don't think originalName checks the macro name? If we check that the macro names in LHS and RHS are the same then I would agree with you.
There was a problem hiding this comment.
Should probably use getMacroName() instead, and add a test. originalName() seems to be for typedefs only.
There was a problem hiding this comment.
Should say "typdefs and other simplifications which aren't macro replacements". We should probably spell that out in a comment for originalName().
| @@ -7066,6 +7066,14 @@ class TestOther : public TestFixture { | |||
| " if ($a == $a) { }\n" | |||
| "}"); | |||
| ASSERT_EQUALS("", errout_str()); | |||
|
|
|||
| checkP("#define X 1\n" | |||
There was a problem hiding this comment.
I'm not sure what macroName there will be if there are nested macros. Can you try this:
#define X 1
#define Y X
void f() {
if (X == Y) {}
}
I think that ideally we don't warn about that because Y could possibly have different definitions..
There was a problem hiding this comment.
it strikes me that it might be within reach to identify if a macro can have different possible values in different configurations.. if the code is something like:
#ifdef __GNUC__
#define X 1
#else
#define X 2
#endif
I think it would not be impossible to mark "X" as a macro that has different values using the info from simplecpp that I think is available. But the info must be propagated to the checkers somehow..
There was a problem hiding this comment.
We also don't know if a macro was a simple define or a functional-style macro, which should be handled differently (see the failing test in windows.cpp).
Caused inconsistent results in #6889
Caused inconsistent results in danmar#6889
No description provided.