On Mon, 25 Feb 2008, Paul Vigay wrote: > I'm not sure if there's an official (W3C?) recommendation, but I'd > have said that absolute widths should precedence over relative widths.
The various HTML specifications over the years have failed to provide sufficient detail on the effects of combining multiple units in tables or describing the behaviour when the units add up to strange amounts (for example, more than 100% on one row). As a result, browsers seem to have organically evolved a "what users expect" mechanism and if you spend a while trying some tests out with MSIE, Firefox, Opera, Safari or the like, you'll discover that the behaviour is exceptionally complex. Browse attempted to walk the line beween Navigator and MSIE behaviour back in its day. It was reasonably successful at matching the widthing algorithms in common cases, but not edge cases. Have a look at the Browse table widthing source, in particular tables_width_table from line 554, for an idea of the full horror. http://preview.tinyurl.com/yrctao CSS adds an additional layer of complexity by being able to specify its own width constraints in various different units. Although their interaction within CSS is better described by the specification than HTML widths in tables, the interaction between CSS widths and HTML width is not so well covered. -- TTFN, Andrew Hodgkinson Find some electronic music at: Rowing in Cambridge, UK? See: http://pond.org.uk/music.html http://nines.rowing.org.uk/
