About Revert "Check pointer before deferenging"

Stephan Bergmann sbergman at redhat.com
Mon Jun 10 07:22:31 PDT 2013

On 06/09/2013 08:32 AM, Matteo Casalin wrote:
> Hi Caolán,
>      first of all thanks for the rationale in your revert (commit
> 8a3f61c42b7dfd7221bd7bc9d9a5a70c052a0bf5), I understand it is not
> mandatory but it really helps.
> I introduced those checks while visually scrolling through code for a
> previous commit. Is there a rule of thumb on whether these checks are
> required or not (and where), or should I just rely on reports from code
> analysis tools?

First of all, introducing null pointer checks without understanding the 
code is never a good idea.  If you do not understand the code in 
question, you neither know (a) whether the pointer is intended not to be 
null here and can only be so due to a programming error somewhere else 
(so the fix should also go somewhere else), nor (b) what to do in the 
else branch.

And in the special case of a pointer obtained from plain new, it can 
never be null anyway, as already explained by Caolán.


More information about the LibreOffice mailing list