- From: Simon Sapin <simon.sapin@kozea.fr>
- Date: Sun, 17 Jun 2012 20:10:12 +0200
- To: Anton Prowse <prowse@moonhenge.net>
- CC: "www-style@w3.org" <www-style@w3.org>
Le 16/06/2012 20:50, Anton Prowse a écrit : > Supposedly, the 'position' property applies to all elements. I can't > find any evidence of relative positioning applying to internal table > elements in Gecko, though. (Not even to table cells; a relpos > table-cell doesn't seem to establish the containing block for its abspos > children.) "Applying" to all elements is needed so that 9.7 can turn eg. a table row into an abspos block. > I feel that it ought to be possible to relatively position these > elements, but perhaps with the restriction in CSS21 that 'top', > 'bottom', left' and 'right' don't have any effect (if that makes life > easier for implementers). > > If internal table elements can be relpos'd, then presumably z-index can > act on them, and hence they can establish stacking contexts. I don’t know if any combination of z-index can make sense given the complicated relationships of backgrounds inside a table. For example according to 17.5.1, the empty-cells property can make not only a cell transparent, but also row, row group, column and column group backgrounds. (By the way we did not care to implement this in WeasyPrint. We always make backgrounds stacked rectangles and never drill "holes" in them. This is probably non-conforming, but the use case is just too weak to bother, at least for us and for now.) Regards, -- Simon Sapin
Received on Sunday, 17 June 2012 18:10:42 UTC