On Thu, Mar 25, 2004 at 07:43:00PM +0000, Angus Leeming wrote:
> Furthermore, there is a leak in the code above. I think the minimal
> change is:
>
> if (from.getInset(i)) {
> // the inset is not in a paragraph anymore
> tmpinset = from.insetlist.release(i);
> + from.insetlist.erase(i);
> }
>
> - if (!to.insetAllowed(tmpinset->lyxCode()))
> + if (!to.insetAllowed(tmpinset->lyxCode())) {
> + delete tmpinset;
> return false;
> + }
> - to.insertInset(j, tmpinset, tmpfont);
> + if (tmpinset)
> + to.insertInset(j, tmpinset, tmpfont);
>
> Ie, I think it was wrong in THREE places!
Does it still work after this change?
[A simple solution would be a wrapper to an InsetBase * that provides
value-semantics pretty much the way MathAtom encapsulates a MathInset.
This might cost a few cycles now and then, but it would also simplify
the Paragraph c'c'tor and d'tor.]
Andre'
