Re: [REVIEW-3.5] crasher fdo#45987

On 18/05/12 10:31, Noel Power wrote:
> On 18/05/12 10:26, Noel Power wrote:
> > Hi
> > somewhat of a containing fix ( I think there is other rework needed, 
> > see below ) that avoids the mentioned crasher, please consider 
> > cherry-picking to 3.5
> >
> > Always ( afaict ) the code expects the index of the entry in the 
> > (maTabs) vector to correspond a tab of the same index. However the 
> > DeleteTab routine patched above will erase the entry for the tab but 
> > if that tab isn't the last tab but instead some random tab in the 
> > middle won't the order of the tabs be screwed ? We could check and 
> > only delete an entry if it is the last entry ( and otherwise make a 
> > null entry for the deleted tab ) but then we definitely would expect 
> > the code should be ready to deal with such a 'hole' (representing a 
> > sheet that no longer exists ) in the vector, that doesn't appear to 
> > be the case. Is this how it should work ? I could rework it like that 
> > if that is the intention, is it ?
> and I forgot to mention the commit ( or actually 2 commit ids as it 
> appears I introduced a wae that was fixed by sb )
>
> http://cgit.freedesktop.org/libreoffice/core/commit/?id=8b1d29bc9b00bc2730738a990023a65ab6e0219b 
> &
> http://cgit.freedesktop.org/libreoffice/core/commit/?id=abb26f51eea0399754cc8f5b7d7a7d648d68f630 

I took it that it should work how I outlined above and committed a 
further fix which should safeguard against illegal access, please 
additionally consider
http://cgit.freedesktop.org/libreoffice/core/commit/?id=8352eb5a1af1eb44550a9d60d31e6c2fb2dc43b9

thanks,

Noel