RE: FOP FAD design approaches

[Victor Mote]

Peter B. West wrote:

 It occurs to me that some of the implications of the FAD 
 approach have not been successfully communicated.  Part of 
 this may well be because of my own inadequate understanding 
 of the FOP process.  Before I continuing with this 
 discussion, I had better ensure that my understanding of one 
 important point is correct.  Why does FOP process in minimum 
 units of a page-sequence?

1. To process in units smaller than a page-sequence presents layout
dependency problems. An item on the last page of a page-sequence can
theoretically change the layout of something on the first page.
2. To process in units larger than page-sequence presents memory-usage
problems.

Except for things like unresolved page numbers in other page-sequences
(deemed to have an acceptable workaround), you can layout a page-sequence in
isolation from the other page-sequences.

BTW, although the main purpose of the FO Tree event-firing mechanism
(recently removed from HEAD) was to allow different layout engines to react
to page-sequence objects in different ways, one of the IMO advantageous
side-effects was that other high-level events could be fired as well,
allowing a layout engine to use something other than page-sequence as a
trigger. I thought at the time that this might be helpful to those wanting a
more eager layout strategy.

Victor Mote

FOP FAD design approaches

[Peter B. West]

Fop-devs,
It occurs to me that some of the implications of the FAD approach have 
not been successfully communicated.  Part of this may well be because of 
my own inadequate understanding of the FOP process.  Before I continuing 
with this discussion, I had better ensure that my understanding of one 
important point is correct.  Why does FOP process in minimum units of a 
page-sequence?

Peter
--
Peter B. West http://www.powerup.com.au/~pbwest/resume.html

Re: FOP FAD design approaches

[Glen Mazza]

Here is what I understand, others may have more to
add:

I see two ways to interpret your question:  (1) what
happens in the code that makes page-sequence be the
lowest granularity we have, and (2) why we do it that
way.  

1) When the page-sequence's children
(layout-master-set, declarations?, flow) have been
finished processing, endElement() in our SAXHandler
FOTreeBuilder is called. [1]  That method calls end()
of the PageSequenceObject [2], which in turn calls
endPageSequence() [3] of the FOInputHandler object,
which is either FOTreeHandler for PDF/PS, etc. or
MIFHandler or RTFHandler.

It is endPageSequence() which calls
formatPageSequence(pageSequence, areaTree)--which
starts the formatting process.

[1]
http://cvs.apache.org/viewcvs.cgi/xml-fop/src/java/org/apache/fop/fo/FOTreeBuilder.java?annotate=1.35#197

[2]
http://cvs.apache.org/viewcvs.cgi/xml-fop/src/java/org/apache/fop/fo/pagination/PageSequence.java?annotate=1.27#170

[3]
http://cvs.apache.org/viewcvs.cgi/xml-fop/src/java/org/apache/fop/fo/FOTreeHandler.java?annotate=1.25#199

2) Our processing is directly on top of SAX--we react
at the start and end of XML elements as they come in. 
Unlike FAD we don't cache into any intermediate data
structures, so I don't see how we could do a lower
granularity.  We could stop at the end of fo:flow, but
that is only one element away from fo:page-sequence's
object endElement() being called from above, so 
nothing measurable would be gained.  

I guess the answer to your question in (2) is (a) that
our process is driven by fo:objects, not the areas
that they generate, and there is no fo:object that
corresponds to 1 page, etc.--that's determined by
fo:flow as it is formatted and (b) there are some
programmatic conveniences to stopping at the
page-sequence level, as the layout-master-set can be
read in one time and is valid for all the pages
generated.

Glen

--- Peter B. West [EMAIL PROTECTED] wrote:
 Fop-devs,
 
 It occurs to me that some of the implications of the
 FAD approach have 
 not been successfully communicated.  Part of this
 may well be because of 
 my own inadequate understanding of the FOP process. 
 Before I continuing 
 with this discussion, I had better ensure that my
 understanding of one 
 important point is correct.  Why does FOP process in
 minimum units of a 
 page-sequence?
 
 Peter
 -- 
 Peter B. West
 http://www.powerup.com.au/~pbwest/resume.html