What I write next should be consumed with caution and the fact in mind that English is a foreign language to me, because I have big trouble translating 4.2.4. 4.2.4 defines "preceding" WRT to the area tree, but I really can't parse that section. OTOH, 7.15.12 talks about flow objects, not areas, even though we're in the area generation stage. I didn't find a definition for "preceding" for the FO tree, BUT! XPath defines the "preceding" axis so that the space after "Start" (in your example) is directly preceding the first space inside the inline. So if you ask me, the current behaviour of FOP is correct. D'oh! :-)
On 06.10.2005 08:22:32 Manuel Mall wrote: > Not sure if this is another of those areas in the spec which is cause > for much confusion but I noticed that FOP trunk collapses white space > across fo:inlines. For example (I use a . to represent a white space > character): > Start.<fo:inline>.Text.</fo:inline>.End > is rendered as: > Start.Text.End > > However, in 7.15.12 the white-space-collapse property definitions says > for "true" (the default value) and I am rephrasing here(!): A space > character is dropped if the immediately preceding flow object is a > space character. I would read this as to mean NOT to collapse across > the boundaries of a fo:inline (FWIW - neither RenderX nor AntennaHouse > seem to collapse across fo:inline boundaries) although it hinges on the > interpretation of the term "preceding" in the context of a tree > structure. > > Manuel Jeremias Maerki
