jtamate added inline comments. INLINE COMMENTS
> dfaure wrote in udsentry.cpp:454 > Hmm why can't this be the friend function directly? > > I don't like the added global functions in the public header... > Hmm why can't this be the friend function directly? Because the compiler (clang++ in this case) doesn't know which one to apply. Unless both definitions are equivalent (in source and binary) and the later can be removed. That is something I don't known. src/core/slavebase.cpp:728:14: error: use of overloaded operator '<<' is ambiguous (with operand types 'QDataStream' and 'const KIO::UDSEntry') KIO_DATA << entry; ~~~~~~~~ ^ ~~~~~ src/core/udsentry.h:315:40: note: candidate function friend QDataStream &operator<< (QDataStream &s, const KIO::UDSEntry &a); ^ src/core/udsentry.h:361:29: note: candidate function KIOCORE_EXPORT QDataStream &operator<< (QDataStream &s, const KIO::UDSEntry &a); ^ > I don't like the added global functions in the public header... Me neither, but otherwise clang++will not be able to find the function. src/core/udsentry.h:314:19: error: no function named 'save' with type 'void (QDataStream &, const KIO::UDSEntry &)' was found in the specified scope friend void ::save(QDataStream &, const KIO::UDSEntry &); ^ > dfaure wrote in udsentry.cpp:45 > 1ms is relative to a benchmark which isn't clear when reading this code in > other contexts. As someone said in another review, do we still need this > operator== anyway, given that you pass lambdas to find_if()? ok, ok, bye bye == REPOSITORY R241 KIO REVISION DETAIL https://phabricator.kde.org/D12696 To: jtamate, dfaure, #frameworks Cc: bruns, michaelh, ngraham