[prev in list] [next in list] [prev in thread] [next in thread] 

List:       fop-cvs
Subject:    =?iso-8859-1?q?=5BXML_Graphics_-_FOP_Wiki=5D_Updated=3A__PageLayout?=
From:       fop-cvs () xml ! apache ! org
Date:       2005-02-25 21:50:17
Message-ID: 20050225215017.50779.19002 () minotaur ! apache ! org
[Download RAW message or body]

   Date: 2005-02-25T13:50:17
   Editor: SimonPepping
   Wiki: XML Graphics - FOP Wiki
   Page: PageLayout
   URL: http://wiki.apache.org/xmlgraphics-fop/PageLayout

   no comment

Change Log:

------------------------------------------------------------------------------
@@ -1,5 +1,7 @@
 = Page layout management =
 
+== Returning, collecting and managing the possible pagebreaks ==
+
 This is a plan to achieve page layout management in FOP. With page
 layout management I mean that the application is able to apply an
 algorithm that makes the best choice from a number of layout
@@ -84,6 +86,26 @@
 the caller if the BP is part of the current page. It can do so due to
 its ordered list of BPs.
 
+Luca Furini
+[http://nagoya.apache.org/eyebrowse/ReadMsg?listName=fop-dev@xml.apache.org&msgNo=10549]
 +pointed to the situation where an FONode has keep-together, which
+influences the BPs of its whole subtree. This presents some difficulty
+for the above strategy. The property must be propagated down the
+subtree of LMs, which must signal it in their BPs so that the PageLM
+can take it into account.
+
+Finn Bock
+[http://nagoya.apache.org/eyebrowse/ReadMsg?listName=fop-dev@xml.apache.org&msgNo=10550]
 +favours returning the BPs up the stack of LMs up to the LM that does
+the page breaking, analogous to the LineLM in paragraph layout. It
+makes it easier to handle the above case. But it is rather expensive
+in processing, also because each LM creates a new BP which wraps the
+received BP and which is returned to its parentLM. 
+
+Page breaking was discussed in this email thread:
+[http://nagoya.apache.org/eyebrowse/BrowseList?listName=fop-dev@xml.apache.org&by=thread&from=984205].
 +The above two messages are part of it.
+
 == Current situation ==
 
 LineLM.getNextBreakPoss returns a BP for each line.
@@ -99,5 +121,157 @@
 child BP to return the areas up to that child BP. A BP has a direct
 correspondence to a child BP through its member leafPos, which is an
 index into the list of child BPs.
+
+== Page breaking strategies ==
+
+The current BP system uses a first-fit strategy. As soon as the page
+height is exceeded, the page break is lain at the preceding BP. This
+strategy is simple, but does not provide choice. Therefore it is hard,
+although not impossible to take keeps into account.
+
+The next simple strategy is a best-fit strategy. One collects at least
+one BP that no longer fits on the page. There may be several possible
+page breaks due to stretchability/shrinkability on the page. Then one
+weighs the possible page breaking points with a suitable merits and
+demerits system, and selects the best page break. This method requires
+a better overview over the available break points and
+stretchability/shrinkability on the page. This strategy is used by TeX
+in its vertical list.
+
+The next possible strategy uses look ahead. There is a sliding window
+of ''N'' pages. The best page breaks are calculated over all pages in the
+window, but only the page break of the first page is used. Then the
+window is moved forward one page, and the effort is repeated. This
+strategy may result in a better placement of floating elements.
+
+In a variant of the best-fit strategy and of the look-ahead strategy,
+one may use the page breaks of ''M'' pages. This is useful for balancing
+facing pages (''M''=2, even page/odd page).
+
+A total-fit strategy, as used in paragraph breaking, can only be
+applied to a complete page sequence. This will almost always to
+expensive, both in computing effort and in memory requirements.
+
+== Layout around a page break ==
+
+The layout of a part of the page depends on whether the part is layed
+out at a page break or not. For example, the resolution of space
+specifiers is different when they occur at the edge of a reference
+area or not. When a page break occurs in a table, a footer is
+added. The borders of the footer interact with those of the last rows
+and cells.
+
+A BP should always represent the situation which would occur if it
+were the selected page break. The following BP makes the calculation
+for its predecessor partly undone, because it represents itself as the
+selected page break.
+
+When a page break has been selected, the calculations for the BP after
+the selected page break must be revised. They must now take elements
+at the start of a new page into account. The resolution of the space
+specifiers before the first block is different because it now occurs
+at the before edge of a reference area. When a page break occurs in a
+table, a header is added. The borders of the header interact with
+those of the first rows and cells.
+
+The paragraph layout mechanism of Knuth has elements, penalties, which
+are taken into account when they occur at a line break, but are
+ignored when they do not occur at a line break. See especially the use
+of a penalty to represent a possible hyphenation point. Other
+elements, glue items, are ignored when they occur at a line break, but
+are taken into account when they do not occur at a line break.
+
+It may be useful to do the same for page breaks (see Finn Bock's
+ideas). But the situation both at the end and at the start of a page
+is more complicated. At the end elements are not only removed, they
+may also be added. Similarly at the start of a page.
+
+== Expressing layout around a pagebreak as Knuth elements ==
+
+We consider the following layout situations.
+
+=== Space specifiers ===
+
+When the space specifiers resolve to zero around a page break, we are
+in the same situation as that of a word space in line breaking. It is
+represented by the sequence `box - glue - box`.
+
+When the space specifiers do not resolve to zero around a page break,
+we are in the same situation as that of a word space in line breaking
+in the case of centered lines. It is represented by the sequence 
+{{{
+box - infinite penalty - glue(ha) - zero penalty - glue(hn-ha-hb) - zero width box - \
infinite penalty - glue(hb) - box +}}}
+where ha is the bpd of
+the space-after before the page break, hb is the bpd of the
+space-before after the page-break, hw is the space when there is no
+page break.
+
+=== Possible page break between content elements ===
+
+Here the most general situation is that when the content is different
+with and without page break:
+ * content Cn when there is no page break,
+ * content Ca at the end of the page before the page break,
+ * content Cb at the start of the page after the page break.
+
+An example of this situation is a page break between table rows:
+
+{{{
+no page break:    page break:
+
+---------         ---------
+  row 1             row 1
+---------         ---------
+ border n          border a
+---------         ---------
+  row 2            footer
+---------         ---------
+                  page break
+                  ---------
+                   header
+                  ---------
+                   border b
+                  ---------
+                    row 2
+                  ---------
+}}}
+
+This situation cannot be dealt with using Knuth's box/glue/penalty
+model. We introduce two new type of elements, which are a kind of
+penalties:
+
+  1. A penalty with height hn, which is the height of Cn, and with height ha, which \
is the height of Ca. +  1. An infinite penalty with height hn', which is the height \
of Cn', and with height hb, which is the height of Cb, and with a flag value of \
start-page. Here Cn' is the part of Cn which is not represented in the penalty of \
type 1; normally it is empty. +
+Both penalty types differ from a normal penalty in that they also
+insert a height when there is no page break. For type 2 that is a
+rather theoretical possibility. Penalty type 2 differs from a normal
+penalty in that it is not discarded when it occurs at the start of a
+new page. Note that in that case it is not the page break itself; the
+page break is at a preceding glue or penalty from which it is
+separated by glue or normal penalty items, which are discarded at the
+start of a page. Both types share some features of a box and of a
+penalty; they are a kind of conditional box.
+
+The above table rows are then represented as:
+{{{
+box(row 1) - penalty(hn, ha, 0, 1) - penalty(0, hb, inf, start-page) - box(row 2)
+}}}
+Without a page break this becomes:
+{{{
+box(row 1) - hn (border n) - box(row 2)
+}}}
+With a page break this becomes:
+{{{
+box(row 1) - ha (border a + footer)
+hb (header + border b) - box(row 2)
+}}}
+Here ha and hb include the border-after of the footer and the
+border-before of the header.
+
+Finn Bock launched the idea to use Knuth semantics to express page
+breaking.
+
 
 This plan was written by SimonPepping

---------------------------------------------------------------------
To unsubscribe, e-mail: fop-cvs-unsubscribe@xml.apache.org
For additional commands, e-mail: fop-cvs-help@xml.apache.org


[prev in list] [next in list] [prev in thread] [next in thread]