Q_ASSERT(res == local) in QDateTime::currentDateTime() on ios

QOlzhas

Hi there. I'm trying debug my ios app. I use QDateTime::currentDateTime() and get Q_ASSERT(res == local) which occur in QLocalTime func. I have no idea what's wrong, how can two pointers can be same?

Axel Spoerl

@QOlzhas
Can you pls post a stack trace?

QOlzhas

@Axel-Spoerl, sure

QOlzhas

@Axel-Spoerl. I'm a little confused, how can res and locale be the same thing? Are two pointers being compared that cannot be equal, because the addresses are always different, or do I not understand something? The contents of these two pointers are absolutely identical.

Axel Spoerl

@QOlzhas
That's a bug. Can you file a bug report under https://bugreports.qt.io ?
Thanks!

J.Hilk

@QOlzhas out of curiosity, what do you have in line 58 of your main.cpp?

QOlzhas

@J-Hilk, nothing criminal

57 QApplication app (argc, argv);
58 if (!AppManager::init (&app))
59     return 0;

AppManager is an application core class, which has logger:


DapLogger::DapLogger(QObject *parent, QString appType, size_t prefix_width)
    : QObject(parent)
    , m_day(QDateTime::currentDateTime().toString("dd"))

In the logger constructor calls QDateTime to get current time, where app assert on res == local.

QOlzhas

@Axel-Spoerl, I did it. https://bugreports.qt.io/browse/QTBUG-122960

J.Hilk

@QOlzhas oh my, thats a rookie mistake in the Qt code,

it is

if(tm *res = localtime_r(&utc, local))

but it should be

if(tm *res = localtime_r(&utc, local); res)

someone disabled/ignored warnings during compiling ...

JonB

@J-Hilk said in Q_ASSERT(res == local) in QDateTime::currentDateTime() on ios:

if(tm *res = localtime_r(&utc, local); tm)

I don't use this kind of C++ construct, so excuse me if I am off base, but isn't tm a type here? Did you mean res by any chance?

And isn't the result of if(tm *res = localtime_r(&utc, local)) already res (even if it lacks an extra pair of of () to keep gcc happy, or test != nullptr if you prefer), unless this new-fangled inline declaration of a variable changes that behaviour?

J.Hilk

@JonB said in Q_ASSERT(res == local) in QDateTime::currentDateTime() on ios:

I don't use this kind of C++ construct, so excuse me if I am off base, but isn't tm a type here? Did you mean res by any chance?

you're correct, I changed it.

the result is res, but its not checked. Well I think this may be compiler dependent.

I'm not sure anymore, you made me question myself T_T

JonB

@J-Hilk said in Q_ASSERT(res == local) in QDateTime::currentDateTime() on ios:

the result is res, but its not checked. Well I think this may be compiler dependent.

I don't know what you mean by this. If you do an assignment in a conditional the result of the condition is the result of the assignment. That is C or C++. The only "wriggle" if that compilers like to warn you that if (a = b) might be a mistype for if (a == b), so they variously want you to either put in extra parentheses if ((a = b)) (gcc) or an explicit test if ((a = b) != nullptr) (MSVC?). Your new-fangled if (a = b; a) is just an alternative way of writing the same thing it seems to me?