Re: Cppcheck: false positive in pptx-text.cxx ?

On 22/01/2013 16:37, Markus Mohrhard wrote:
> > I'll do this one. You can see then how it works. You should maybe read
> > http://www.boost.org/doc/libs/1_52_0/libs/smart_ptr/shared_ptr.htm or
> > some other resource about shared_ptr.
> See http://cgit.freedesktop.org/libreoffice/core/commit/?id=ec8ff9ff68622a510319c404f4745a336e65d314
> for an example. Please note that I did some additional clean-up. The
> important parts are that in this case operator= and the copy c'tor are
> no longer needed.
>
> The remaining parts are mainly stylistic and moving the Impl class
> from the header file into the source file.
>
> If you try to change additional places please inspect them very
> carefully. As I and Eike already mentioned, shared_ptr comes with a
> performance penalty that may or may not be a problem.
To be clear, I thought that ref count had been put in place long time 
ago when there wasn't efficient libraries/version of libraries. So when 
you first answered by quoting boost::shared_ptr, I thought it could be a 
100% better solution than ref count.

Now with this performance penalty, I don't know what to think:
- don't touch anything and so keep ref counting
- find a specific solution for each case using than ref counting
- use boost::shared_ptr (but for which case which wouldn't generate perf 
penalty?)

I must recognize, I had naively thought that ref count was a part where 
we could make some bug proof/optimization/cleaning LO

Julien