new "condition is always true/false" FPs

[Martin]

Hi,

with
98b2fd8151897274bfacf5cf151068536c9be3c5
there are a number of new FPs:

return static_cast<pa_usec_t>(-1);
(style) Condition 'static_cast<pa_usec_t>(-1)' is always true

return (HttpRequestType)0;
(style) Condition '(HttpRequestType)0' is always false

return std::shared_ptr<SoftwareUpdateObject>(nullptr);
(style) Condition 'std::shared_ptr<SoftwareUpdateObject>(nullptr)' is always false

return std::unique_ptr<Value>(nullptr);
(style) Condition 'std::unique_ptr<Value>(nullptr)' is always false

return reinterpret_cast<unsigned char const*>(p);
(style) Condition 'reinterpret_cast<unsigned char const*>(p)' is always false

return (char *) -1;
(style) Condition '(char*)-1' is always true

Could you please check?

Thanks,
Martin

[CHR]

Thanks, should be fixed by https://github.com/danmar/cppcheck/pull/4419

[Martin]

Regarding that pull request, why is

"[test.cpp:7]: (style) Condition '(int)0' is always false\n"

(line 4333 in testcondítion.cpp)
a correct finding? There's no condition in

return (int)0;

Same for

return static_cast<int>(1)

[CHR]

Because those statements are equivalent to return false/true;respectively (if the return type is bool)?

[Martin]

But false/true aren't conditions, but boolean values? What's wrong with "return true"?

[CHR]

return true; is readable, return static_cast<int>(1); not so much.
Maybe we should move the discussion to the PR. I don't know if it will be merged as-is.

[Martin]

The PR is merged now, I'd still argue though that if you actually want have warnings like (with the current head)

(style) Condition '(Return_t)0' is always false

for

return (Return_t)0;

then at least the wording of the message should be changed. "condition" and "boolean value" cannot be used completely interchangeably.