You find that surprising? You should be used to that by now. :-) I think the spec can be interpreted both ways. Have you checked what other implementations do? If you find only one behaviour, we've missed something. ;-)
Jeremias Maerki On 21.11.2007 12:19:56 Vincent Hennebert wrote: > Hi, > > Surprisingly enough I found something unclear in the specification > regarding breaks in tables: > In the separate border model there is a border-separation property that > specifies the distance between adjacent cells. The questions are: > - if there is a break in a table, between cells (regardless of whether > the table’s border is conditional, see below), should the last cells > on the page also have half of the border-separation after them? > Technically speaking there are no adjacent cell after them... > I’d say yes, and that’s how the current code behaves, because the spec > explicitly states that for cells the border has two components, but... > - if the corresponding border is conditional > (border-before/after-width.conditionality = discard), should the > border-separation also be discarded? The question holds for fo:table > as well as for fo:table-cell (when there is a break /inside/ it) > > I tend to think that the separation should be discarded if so is the > border, but I’d welcome confirmation on this topic. And that may lead to > funny things, for example if a two-column table has a conditional > border-after, so has the cell on the first column, but the cell on the > second column has a retained border. The bottom of the first cell will > align with the bottom of the region-body, but for the second cell there > will be /half/ of the border-separation between its border-after and the > bottom of the region-body. Follow me ;-) ?