This approach is not a good idea. It is possible that a runtime
environment is configured to have multiple instances of Fop being
instantiated by FopFactory with different rendering configurations (e.g.
renderer resolution values).
Its disappointing, but I've just noticed a bug in AFPPageOverlayElement,
an AFPPaintingState shouldn't just be instantiated in there like that.
The processNode implementation will not accurately calculate and plot
the page overlay position if the document resolution is different from
the detault value of 240dpi. This calculation should be carried out
much later at rendering time, *not* in here at configuration time - its
very hacky and you'll need to refactor this. I seem to remember that
Chris worked on this new feature so you may want to converse with him
about its implementation.
Venkat Reddy wrote:
AFPPaintingState is being used in three different places altogether in
FopTrunk source. The default constructor is being used in the
following three classes...
There is a variable 'resolution' is being initialized for each
instance, this resolution parameter can be set using the 'fop.xconf'
for a particular render...
Ex:- AFPRenderer configuration below
The bit depth and type of images produced
(this is the default setting)
<images mode="b+w" bits-per-pixel="8"/>
The above <renderer-resolution> is being hardcoded as '240dpi' in
AFPRendererConfigurator.java, which initiates the renderer resolution
based on the configuration set in 'fop.xconf'. In order to resolve
this problem, I will be changing the AFPPaintingState as singleton, so
that all the above classes will get the instance using 'getInstance()'
method instead of default constructor. This will resolve the
<renderer-resolution> problem as well, by a simple change in
AFPRendererConfigurator (instead of hardcoded value 240, assigning the
value from the configuration object).
Please review the above changes and tell me, if I am doing anything