Hi Andreas,
I'm guessing by "flaws and dirt" you are referring to the broken junit test in FOTreeTestSuite due
to fo:retrieve-table-marker being a child of fo:block in
test/fotree/testcases/table_retrieve-table-marker.fo.
Spec states: "An fo:retrieve-table-marker is only permitted as the descendant of an fo:table-header
or fo:table-footer or as a child of fo:table in a position where fo:table-header or fo:table-footer
is permitted."
It seems FOP is enforcing this. Would be good if you could fix this as I am ready to commit some of
my own work to trunk but ideally would like to have a clean junit run before doing so.
Adrian.
Andreas Delmelle wrote:
On May 1, 2008, at 23:34, [EMAIL PROTECTED] wrote:
Author: adelmelle
Date: Thu May 1 14:34:19 2008
New Revision: 652673
URL: http://svn.apache.org/viewvc?rev=652673&view=rev
Log:
First steps towards fo:retrieve-table-marker
FWIW: I decided to finally look into this, as I've been noticing
increasing demands for a port of 0.20.5's table-continuation extension.
I already noticed some flaws and dirt in what I committed earlier;
cleanup to follow as soon as SVN commit is restored.
After this commit, the node is completely implemented on the fo-tree
side, but still lacks a backing implementation in the layout engine. It
does not yet generate any LayoutManagers (see modification to
LayoutManagerMapping.java)
In theory, everything is in place for the RetrieveTableMarker to behave
in exactly the same way as a RetrieveMarker.
Currently, a RetrieveMarker is intercepted and resolved in
AbstractLayoutManager.createChildLMs().
I'm not sure yet if this suffices for a RetrieveTableMarker. Probably
not...
Resolution for regular RetrieveMarkers is ultimately triggered by
StaticContentLM.getNextKnuthElements(), which in turn is called when
finishing a page.
At that point, the areas for the region-body's content will all have
been added to the page, together with any associated fo:markers.
It seems that, at least for retrieve-position-within-table="page",
communication/interaction with the PageBreaker/PageSequenceLM is necessary.
Any thoughts?
Cheers
Andreas