Empty rows should not be rendered

[ericscheid]

Consider this table, where very long text data has been split over multiple lines for editing reasons, and all the cells of the row are rowspanned into the above row..

| Race     | Multi-class options                                         |
|:---------|:------------------------------------------------------------|
| Dwarf    | fighter/cleric, fighter/thief                               |
| Elf      | fighter/mage, fighter/thief, mage/thief,                    |
|         ^| fighter/mage/thief                                         ^|
| Gnome    | fighter/thief, fighter/cleric, fighter/illusionist,         | 
|         ^| thief/cleric, thief/illusionist, cleric/illusionist        ^|
| Half-elf | fighter/priest, fighter/mage, fighter/thief, ranger/priest, |
|         ^| mage/priest, mage/thief,                                   ^|
|         ^| fighter/mage/priest, fighter/mage/thief                    ^|
| Halfling | fighter/thief                                               |
| Human    | none                                                        |

Unfortunately, the markdown renders this with empty <tr></tr>...

This is a bug, and it messes up the even/odd row striping:

[calculuschild]

Hmm interesting find. Good eye.

[G-Ambatte]

To further muddy the waters, if the <tr></tr> is removed completely, then the preceding rowspan(s) must also be decremented, or else the entire table format will break in new and interesting ways.

As I understand it, this is not actually the intended use of rowspan. I can see how it might be clearer in the editor and thus more convenient - but as per the current implementation, it doesn't actually work.

WORKAROUNDS:

• If the user doesn't attempt to misuse rowspan in this manner and keeps the content for a single row in a single row, then it works fine.
• If the user ensures that they always span an odd number of rows (for the entire row), then it works fine.

It's worth noting that, despite the different approaches, the two tables above appear to be visually identical.

[ericscheid]

Oh, good point on decrementing rowspans.

[dbolack-ab]

Would the simplest fix be to strip empty pairs in html.renderer()?

Or must this be fixed in marked-extended-tables ?

[calculuschild]

Fixes to table generation should go in marked-extended-tables