4708 warnings...

[Benjamin Meyer]

On May 5, 2009, at 8:46 AM, Thomas McGuire wrote:

> Hi,
>
> On Tuesday 05 May 2009 01:57:10 Maksim Orlovich wrote:
>>> If anyone wanted to dramatically reduce the warning count, she would
>>> start fixing those with the highest duplicates :)
>>
>> You will need to fix gcc to fix the first 4, since the code is doing
>> absolutely nothing wrong.
>
> Of course KHTML/KJS does nothing wrong, but that is not the point.
> The point is to get rid of these warning to see the warnings that  
> actually
> matter, so that we can see real errors in the code. I would like to  
> see these
> warnings fixed to reduce the noise when compiling.
>
> Actually, the virtual-overload warnings _did_ find a real bug in  
> KMail. For
> most cases those are false positives, but those can be suppressed with
> "using".
> Other warnings can also be fixed easily.
>
> Regards,
> Thomas

In a similar vain to warnings is console debug.  A year or so ago I  
was trying out a version of Amarok and when running it from the  
command line it spit out so much to the console that it was just  
embarrassing.  The worst part was that when going back over and  
reviewing what it had outputed there were several QObject::connect()  
statements about incorrect connects.  There was so much "debug" that  
the real errors were getting lost!  It is really a shame that KDE apps  
dump so very much pointless stuff to the console when in release  
mode.  When an end user runs your app there should be no output to  
stderr by default unless it is a specific message to the user or an  
actual error.  Just like with warnings when you have thousands of them  
you miss the real errors.

Building without compile warnings and running in release without  
filling the console are signs of professionalism.  When I see  
applications that can't even do that I really wonder what problems are  
hiding just under the covers.

-Benjamin Meyer