Looking at it in detail, this is not so important since detection of table size in ods files has not changed very much. It always has been a pain: the xml nodes duplicate the last unused column by "number-columns-repeated" attributes up to the specified maximum column count, and the same with the rows. Since fpspreadsheet creates a cell record for every table cell it is important that these superfluous cells are not considered. Now I have an idea how to approach this issue differently. Please give me some time.