On 10/2/18 8:15 AM, Jean-Marc Lasgouttes wrote:
> Le 02/10/2018 à 06:24, Richard Kimberly Heck a écrit :
>> There are some known issues about math insets and buffer_ pointers we've
>> never sorted out. This OUGHT to be set somewhere near creation of the
>> inset. The message has the status of a warning. But this problem can
>> lead to a crash: See Inset::buffer(). So maybe it would be worth
>> committing your patch, but with a corresponding warning such as:
>> LYXERR0("Unassigned buffer_ detected in
>> InsetMathGrid::doDispatch.");
>> It has proven very hard to fix these kinds of bugs properly.
>
> I think I know what happens: copying a cell as is done in copyRow
> leads to cloning the insets. However, inset cloning does not keep the
> buffer_ value, which makes sense.
>
> The point where we know that (1) insets have been copied and (2) that
> they go in a known buffer is in copyRow() as Scott proposed.
>
> If we want to be a bit cleaner, we can redefine MathData::setBuffer so
> that it sets buffer to its contents (and use that in
> InsetNest::setBuffer).
Seems sensible.
RKH
