--- Comment #7 from Andreas L. Delmelle <>  2009-06-26 
04:37:10 PST ---

Just noting this for general interest (has little or nothing to do with this
issue per se):
The previous comments suddenly reminded me that I have always considered the
current way that extension /properties/ are handled, as lacking in robustness.
We practically force potential implementors of extension properties to modify
FOP's codebase. For extension elements or attributes, the pattern is much more
open and generic. They can be implemented without necessarily having to modify
FOP and recompile. The same should become the case for extension properties,
eventually. We probably will want to take a look at offering something like an
'ExtensionPropertyMapping' (currently non-existent), so that implementors can
re-use existing PropertyMakers, define their own initial values/enums/keywords,
mark properties as inherited, and so on...

There are already a few 'native' extension properties, for which we define
symbolic literals in fo.Constants and which are also registered in
fo.FOPropertyMapping. I have never really been too happy with that practice.
Contributors are invited to follow that pattern, which will only lead to more

Tackling that issue, however, is not for the faint of heart. It would likely
also imply revisiting the way properties are attached to the FONodes and how
they are exposed to the outside (layoutengine & renderers), so would lead to
changes in a Lot of classes (capital 'L', if you catch the drift...)

Configure bugmail:
------- You are receiving this mail because: -------
You are the assignee for the bug.

Reply via email to