If I create a table with rows with different number of columns, distribute columns evenly does not distribute columns based on the existing number of columns of the selected row but instead based on some “hidden” number of columns based on how the columns of ALL rows were created. How do I distribute columns evenly based on the existing number of selected columns?
As far as I can see it evenly distributes the columns based on the row with most number of cells. If you don’t want that then select the row with the most number of cells and manually set each width, starting from the left, according to your original expectation of widths for the other columns.
If you need to split cells it would be best to do it after everything else is set and the table laid out as you want
A table is a set of cells on a rectangular grid of n columns and m rows. You can deviate from this homogeneous schema only by merging cells but the underlying grid remains a m×n grid. Distribute Xxx Evenly works n this grid.
If your “table” needs a lay out where cell limits do not line up, then your data does not logically belong in a single table. You should then create several tables with different geometries.
To give the illusion of a single table, suppress the spacing between the table. But think twice about the way you present data. If column/row headings are meaningless due to the irregular shape, your data is not “tabular” and should be shown differently.
For better advice, provide an example.