kdelibs/kdeui

Dirk Mueller mueller at kde.org
Sun Apr 21 18:22:57 BST 2002


On Son, 21 Apr 2002, Lubos Lunak wrote:

>  No, it's not clean, otherwise there wouldn't be a valid warning. Ideally, 
> there should be of course 'using', but since it doesn't work with the 
> oficially recommended gcc version, overriding instead is a workaround, that 
> maybe doesn't look 100% perfect, but does the job and doesn't break anything.

Yes, it doesn't break, but it doesn't make the code more cleaner. 

> > Of course, but most of the time this hiding was intended (or we can't
> > change it, i.e. faults in Qt design).
>  Intended hiding of virtual methods, hmm? And faults can be fixed, or should 
> they stay there forever? They could have been fixed for Qt3.

As an example is the changed signatures between QMultiLineEdit and 
QTextEdit. Therefore any inclusion of QMultiLineEdit widget will cause these 
warnings, which are "intended". 

Anyway, I agree the warnings might be useful, but they're safe enough to be 
enabled by default. They're enabled in khtml, because that is code that is 
mostly built around its own, and doesn't depend on Qt or kdelibs inherited 
signature mismatches too much. 

> > by inlining a virtual method we can not change it anymore.
>  I said it can be non-inline, without any serious performance hit or whatever.

Yes, but that wasn't done, and this is the part that annoyed me. thats all. 

> > Yes, does that prevent people from committing broken code? no. When we
> > change the default to --enable-warnings, those people who care about the
> > 1% performance penality will immediately switch to --disable-warnings and
> > continue committing broken code.
>  But such people will be different from those who now don't see the warnings - 
> those who don't see the warnings now often don't know how to turn it on or 
> even that they can turn it on, while people who will use --disable-warnings 
> and commit code with warnings are ... well ... @#%#!

Yes, you get the problem :-) Its the same with the recommendation to use 
--enable-debug, and still, how many developers use --enable-debug ?


Dirk




More information about the kde-core-devel mailing list