Bug 1907031

Summary: FTBFS: kdelibs on Rawhide (gcc 11)
Product: [Fedora] Fedora Reporter: Richard Shaw <hobbes1069>
Component: kdelibsAssignee: Kevin Kofler <kevin>
Status: CLOSED RAWHIDE QA Contact: Fedora Extras Quality Assurance <extras-qa>
Severity: unspecified Docs Contact:
Priority: unspecified    
Version: rawhideCC: jgrulich, jreznik, kde-sig, kevin, law, me, rdieter, smparrish, than
Target Milestone: ---   
Target Release: ---   
Hardware: Unspecified   
OS: Unspecified   
Whiteboard:
Fixed In Version: Doc Type: If docs needed, set a value
Doc Text:
Story Points: ---
Clone Of: Environment:
Last Closed: 2020-12-13 15:10:38 UTC Type: Bug
Regression: --- Mount Type: ---
Documentation: --- CRM:
Verified Versions: Category: ---
oVirt Team: --- RHEL 7.3 requirements from Atomic Host:
Cloudforms Team: --- Target Upstream Version:
Embargoed:

Description Richard Shaw 2020-12-12 13:14:33 UTC 
I am testing rebuilds of all packages dependent on OpenEXR as the latest version has move from autotools to CMake and now has ilmbase built in.

https://copr.fedorainfracloud.org/coprs/hobbes1069/openexr/build/1829270/

kdelibs appears to be FTBFS probably due to gcc 11 which finds new errors?


/builddir/build/BUILD/kdelibs-4.14.38/kdecore/localization/klocale_kde.cpp: In member function 'virtual bool KLocalePrivate::use12Clock() const':
/builddir/build/BUILD/kdelibs-4.14.38/kdecore/localization/klocale_kde.cpp:2441:59: error: ordered comparison of pointer with integer zero ('const void*' and 'int')
 2441 |     if ((timeFormat().contains(QString::fromLatin1("%I")) > 0) ||
      |          ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~
/builddir/build/BUILD/kdelibs-4.14.38/kdecore/localization/klocale_kde.cpp:2442:59: error: ordered comparison of pointer with integer zero ('const void*' and 'int')
 2442 |         (timeFormat().contains(QString::fromLatin1("%l")) > 0)) {
      |          ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^~~
/builddir/build/BUILD/kdelibs-4.14.38/kdecore/localization/klocale_kde.cpp:2447:1: warning: control reaches end of non-void function [-Wreturn-type]
 2447 | }
      | ^
Comment 1 Kevin Kofler 2020-12-13 00:21:07 UTC 
This code is a leftover from Qt 3 times: In Qt 3, QString::contains returned an int, the number of times the character occurs in the string. In Qt 4, QString::contains was changed to return a bool instead (which is still the case in Qt 5).

The fix is to remove the "> 0" in both lines.

This bug is still present in the copy of the code in KF5 kdelibs4support:
https://invent.kde.org/frameworks/kdelibs4support/-/blob/ff85d52bb12300cceb30492b2aff3e4692644e1c/src/kdecore/klocale_kde.cpp#L2125
Comment 2 Jeff Law 2020-12-13 15:10:38 UTC 
This should be fixed now as well as a minor const-correctness issue.
Comment 3 Kevin Kofler 2020-12-13 15:21:18 UTC 
Well, the != 0 checks don't really make sense semantically because contains actually returns a boolean. What happens there is that the Qt methods actually return a magic internal class that implicitly converts to both bool and void *, exactly so that legacy code that has explicit != 0 checks still works (and with GCC < 11, one got away even with > 0 checks), but it's still legacy. :-)
Comment 4 Jeff Law 2020-12-13 15:24:30 UTC 
That's the most direct fix without changing the meaning of the existing code.  Someone with a better understanding of the QT interfaces may well have a better fix.