Both in revert_multirowOffset and revert_multirow, we do not really
do a proper reversion. The rowspan is set to two and the width to
2cm, because the information required to set both properly is not
readily available. So I wonder a bit what we are doing here and what
our policy is. The information could be gathered, though that would
take a bit of work.
It was a lot work! I've put it in now in r36220 and 36221.
And if we are just going to guess such things, why not just revert it
to normal cells and forget the multirow? Is the idea that, if we put
\multirow{} in ERT, then it will be easier for users to adjust it
themselves?
Yes. Because people can then easily change the width to the desired one. The other things to be done
like unsetting the table lines, preamble and TeX code stuff is already done. They also see directly
in the output that the table has a multirow and can fine-tune it if necessary.
But now it should work correctly out of the box and the user doesn't need to
fine-tune anything.
regards Uwe
