Tree transformation constructs the result tree. In XSL, this tree is called the element and attribute tree, with objects primarily in the "formatting object" namespace. In this tree, a formatting object is represented as an XML element, with the properties represented by a set of XML attribute-value pairs. The content of the formatting object is the content of the XML element. Tree transformation is defined in the XSLT Recommendation . A diagram depicting this conceptual process is shown below.

The XSL stylesheet is used in tree transformation. A stylesheet contains a set of tree construction rules. The tree construction rules have two parts: a pattern that is matched against elements in the source tree and a template that constructs a portion of the result tree. This allows a stylesheet to be applicable to a wide class of documents that have similar source tree structures.

In some implementations of XSL/XSLT, the result of tree construction can be output as an XML document. This would allow an XML document which contains formatting objects and formatting properties to be output. This capability is neither necessary for an XSL processor nor is it encouraged. There are, however, cases where this is important, such as a server preparing input for a known client; for example, the way that a WAP (http://www.wapforum.org/faqs/index.htm) server prepares specialized input for a WAP capable hand held device. To preserve accessibility, designers of Web systems should not develop architectures that require (or use) the transmission of documents containing formatting objects and properties unless either the transmitter knows that the client can accept formatting objects and properties or the transmitted document contains a reference to the source document(s) used in the construction of the document with the formatting objects and properties.

Formatting interprets the result tree in its formatting object tree form to produce the presentation intended by the designer of the stylesheet from which the XML element and attribute tree in the "fo" namespace was constructed.

The vocabulary of formatting objects supported by XSL - the set of fo: element types - represents the set of typographic abstractions available to the designer. Semantically, each formatting object represents a specification for a part of the pagination, layout, and styling information that will be applied to the content of that formatting object as a result of formatting the whole result tree. Each formatting object class represents a particular kind of formatting behavior. For example, the block formatting object class represents the breaking of the content of a paragraph into lines. Other parts of the specification may come from other formatting objects; for example, the formatting of a paragraph (block formatting object) depends on both the specification of properties on the block formatting object and the specification of the layout structure into which the block is placed by the formatter.

The properties associated with an instance of a formatting object control the formatting of that object. Some of the properties, for example "color", directly specify the formatted result. Other properties, for example 'space-before', only constrain the set of possible formatted results without specifying any particular formatted result. The formatter may make choices among other possible considerations such as esthetics.

Formatting consists of the generation of a tree of geometric areas, called the area tree. The geometric areas are positioned on a sequence of one or more pages (a browser typically uses a single page). Each geometric area has a position on the page, a specification of what to display in that area and may have a background, padding, and borders. For example, formatting a single character generates an area sufficiently large enough to hold the glyph that is used to present the character visually and the glyph is what is displayed in this area. These areas may be nested. For example, the glyph may be positioned within a line, within a block, within a page.

Rendering takes the area tree, the abstract model of the presentation (in terms of pages and their collections of areas), and causes a presentation to appear on the relevant medium, such as a browser window on a computer display screen or sheets of paper. The semantics of rendering are not described in detail in this specification.

The first step in formatting is to "objectify" the element and attribute tree obtained via an XSLT transformation. Objectifying the tree basically consists of turning the elements in the tree into formatting object nodes and the attributes into property specifications. The result of this step is the formatting object tree.

As part of the step of objectifying, the characters that occur in the result tree are replaced by fo:character nodes. Characters in text nodes which consist solely of white space characters and which are children of elements whose corresponding formatting objects do not permit fo:character nodes as children are ignored. Other characters within elements whose corresponding formatting objects do not permit fo:character nodes as children are errors.

The content of the fo:instream-foreign-object is not objectified; instead the object representing the fo:instream-foreign-object element points to the appropriate node in the element and attribute tree. Similarly any non-XSL namespace child element of fo:declarations is not objectified; instead the object representing the fo:declarations element points to the appropriate node in the element and attribute tree.

The second phase in formatting is to refine the formatting object tree to produce the refined formatting object tree. The refinement process handles the mapping from properties to traits. This consists of: (1) shorthand expansion into individual properties, (2) mapping of corresponding properties, (3) determining computed values (may include expression evaluation), (4) handling white-space-treatment and linefeed-treatment property effects, and (5) inheritance. Details on refinement are found in .

The refinement step is depicted in the diagram below.

The third step in formatting is the construction of the area tree. The area tree is generated as described in the semantics of each formatting object. The traits applicable to each formatting object class control how the areas are generated. Although every formatting property may be specified on every formatting object, for each formatting object class, only a subset of the formatting properties are used to determine the traits for objects of that class.

Area generation is depicted in the diagram below.

Doing both scrollable document windows and pagination introduces new complexities to the styling (and pagination) of XML content. Because pagination introduces arbitrary boundaries (pages or regions on pages) on the content, concepts such as the control of spacing at page, region, and block boundaries become extremely important. There are also concepts related to adjusting the spaces between lines (to adjust the page vertically) and between words and letters (to justify the lines of text). These do not always arise with simple scrollable document windows, such as those found in today's browsers. However, there is a correspondence between a page with multiple regions, such as a body, header, footer, and left and right sidebars, and a Web presentation using "frames". The distribution of content into the regions is basically the same in both cases, and XSL handles both cases in an analogous fashion.

XSL was developed to give designers control over the features needed when documents are paginated as well as to provide an equivalent "frame" based structure for browsing on the Web. To achieve this control, XSL has extended the set of formatting objects and formatting properties. In addition, the selection of XML source components that can be styled (elements, attributes, text nodes, comments, and processing instructions) is based on XSLT and XPath , thus providing the user with an extremely powerful selection mechanism.

The design of the formatting objects and properties extensions was first inspired by DSSSL. The actual extensions, however, do not always look like the DSSSL constructs on which they were based. To either conform more closely with the CSS2 specification or to handle cases more simply than in DSSSL, some extensions have diverged from DSSSL.

There are several ways in which extensions were made. In some cases, it sufficed to add new values, as in the case of those added to reflect a variety of writing-modes, such as top-to-bottom and bottom-to-top, rather than just left-to-right and right-to-left.

In other cases, common properties that are expressed in CSS2 as one property with multiple simultaneous values, are split into several new properties to provide independent control over independent aspects of the property. For example, the "white-space" property was split into four properties: a "white-space-treatment" property that controls how white space is processed, a "linefeed-treatment" property that controls how line feeds are processed, a "white-space-collapse" property that controls how multiple consecutive spaces are collapsed, and a "wrap-option" property that controls whether lines are automatically wrapped when they encounter a boundary, such as the edge of a column. The effect of splitting a property into two or more (sub-)properties is to make the equivalent existing CSS2 property a "shorthand" for the set of sub-properties it subsumes.

In still other cases, it was necessary to create new properties. For example, there are a number of new properties that control how hyphenation is done. These include identifying the script and country the text is from as well as such properties as "hyphenation-character" (which varies from script to script).

Some of the formatting objects and many of the properties in XSL come from the CSS2 specification, ensuring compatibility between the two.

There are four classes of XSL properties that can be identified as:

As mentioned above, XSL uses XSLT and XPath for tree construction and pattern selection, thus providing a high degree of control over how portions of the source content are presented, and what properties are associated with those content portions, even where mixed namespaces are involved.

For example, the patterns of XPath allow the selection of a portion of a string or the Nth text node in a paragraph. This allows users to have a rule that makes all third paragraphs in procedural steps appear in bold, for instance. In addition, properties can be associated with a content portion based on the numeric value of that content portion or attributes on the containing element. This allows one to have a style rule that makes negative values appear in "red" and positive values appear in "black". Also, text can be generated depending on a particular context in the source tree, or portions of the source tree may be presented multiple times with different styles.

There is a set of formatting objects in XSL to describe both the layout structure of a page or "frame" (how big is the body; are there multiple columns; are there headers, footers, or sidebars; how big are these) and the rules by which the XML source content is placed into these "containers".

The layout structure is defined in terms of one or more instances of a "simple-page-master" formatting object. This formatting object allows one to define independently filled regions for the body (with multiple columns), a header, a footer, and sidebars on a page. These simple-page-masters can be used in page sequences that specify in which order the various simple-page-masters shall be used. The page sequence also specifies how styled content is to fill those pages. This model allows one to specify a sequence of simple-page-masters for a book chapter where the page instances are automatically generated by the formatter or an explicit sequence of pages such as used in a magazine layout. Styled content is assigned to the various regions on a page by associating the name of the region with names attached to styled content in the result tree.

In addition to these layout formatting objects and properties, there are properties designed to provide the level of control over formatting that is typical of paginated documents. This includes control over hyphenation, and expanding the control over text that is kept with other text in the same line, column, or on the same page.

The extension of the properties and formatting objects, particularly in the area on control over the spacing of blocks, lines, and page regions and within lines, necessitated an extension of the CSS2 box formatting model. This extended model is described in of this specification. The CSS2 box model is a subset of this model. See the mapping of the CSS2 box model terminology to the XSL Area Model terminology in . The area model provides a vocabulary for describing the relationships and space-adjustment between letters, words, lines, and blocks.

There are some scripts, in particular in the Far East, that are typically set with words proceeding from top-to-bottom and lines proceeding either from right-to-left (most common) or from left-to-right. Other directions are also used. Properties expressed in terms of a fixed, absolute frame of reference (using top, bottom, left, and right) and which apply only to a notion of words proceeding from left to right or right to left do not generalize well to text written in those scripts.

For this reason XSL (and before it DSSSL) uses a relative frame of reference for the formatting object and property descriptions. Just as the CSS2 frame of reference has four directions (top, bottom, left and right), so does the XSL relative frame of reference have four directions (before, after, start, and end), but these are relative to the "writing-mode". The "writing-mode" property is a way of controlling the directions needed by a formatter to correctly place glyphs, words, lines, blocks, etc. on the page or screen. The "writing-mode" expresses the basic directions noted above. There are writing-modes for "left-to-right - top-to-bottom" (denoted as "lr-tb"), "right-to-left - top-to-bottom" (denoted as "rl-tb"), "top-to-bottom - right-to-left" (denoted as "tb-rl") and more. See for the description of the "writing-mode" property. Typically, the writing-mode value specifies two directions: the first is the inline-progression-direction which determines the direction in which words will be placed and the second is the block-progression-direction which determines the direction in which blocks (and lines) are placed one after another. In addition, the inline-progression-direction for a sequence of characters may be implicitly determined using bidirectional character types for those characters from the Unicode Character Database for those characters and the Unicode bidirectional (BIDI) algorithm .

Besides the directions that are explicit in the name of the value of the "writing-mode" property, the writing-mode determines other directions needed by the formatter, such as the shift-direction (used for subscripts and superscripts), etc.

Because XML, unlike HTML, has no built-in semantics, there is no built-in notion of a hypertext link. In this context, "link" refers to "hypertext link" as defined in http://www.w3.org/TR/html401/struct/links.html#h-12.1 as well as some of the aspects of "link" as defined in http://www.w3.org/TR/xlink/#intro, where "link is a relationship between two or more resources or portions of resources, made explicit by an XLink linking element". Therefore, XSL has a formatting object that expresses the dual semantics of formatting the content of the link reference and the semantics of following the link.

XSL provides a few mechanisms for changing the presentation of a link target that is being visited. One of these mechanisms permits indicating the link target as such; another allows for control over the placement of the link target in the viewing area; still another permits some degree of control over the way the link target is displayed in relationship to the originating link anchor.

XSL also provides a general mechanism for changing the way elements are formatted depending on their active state. This is particularly useful in relation to links, to indicate whether a given link reference has already been visited, or to apply a given style depending on whether the mouse, for instance, is hovering over the link reference or not.

There are two types of areas: block-areas and inline-areas. These differ according to how they are typically stacked by the formatter. An area can have block-area children or inline-area children as determined by the generating formatting object, but a given area's children must all be of one type. Although block-areas and inline-areas are typically stacked, some areas can be explicitly positioned.

A line-area is a special kind of block-area whose children are all inline-areas. A glyph-area is a special kind of inline-area which has no child areas, and has a single glyph image as its content.

Typical examples of areas are: a paragraph rendered by using an fo:block formatting object, which generates block-areas, and a character rendered by using an fo:character formatting object, which generates an inline-area (in fact, a glyph-area).

Associated with any area are two directions, which are derived from the generating formatting object's writing-mode and reference-orientation properties: the block-progression-direction is the direction for stacking block-area descendants of the area, and the inline-progression-direction is the direction for stacking inline-area descendants of the area. Another trait, the shift-direction, is present on inline-areas and refers to the direction in which baseline shifts are applied. Also the glyph-orientation defines the orientation of glyph-images in the rendered result.

If the reference-orientation for an area is 0, then the top, bottom, left, and right edges of the content are parallel to those of the area's parent and consistent with them. Otherwise the edges are rotated from those of the area's parent as described in . The inline-progression-direction and block-progression-direction are determined by the location of these edges as described in .

The Boolean trait is-reference-area determines whether or not an area establishes a coordinate system for specifying indents. An area for which this trait is true is called a reference-area. Only a reference-area may have a block-progression-direction which is different from that of its parent. A reference-area may be either a block-area or an inline-area.

The Boolean trait is-viewport-area determines whether or not an area establishes an opening through which its descendant areas can be viewed, and can be used to present clipped or scrolled material; for example, in printing applications where bleed and trim is desired. An area for which this trait is true is called a viewport-area.

A common construct is a viewport/reference pair. This is a viewport-area V and a block-area reference-area R, where R is the sole child of V and where the start-edge and end-edge of the content-rectangle of R are parallel to the start-edge and end-edge of the content-rectangle of V.

Each area has the traits top-position, bottom-position, left-position, and right-position which represent the distance from the edges of its content-rectangle to the like-named edges of the nearest ancestor reference-area (or the page-viewport-area in the case of areas generated by descendants of formatting objects whose absolute-position is fixed); the left-offset and top-offset determine the amount by which a relatively-positioned area is shifted for rendering. These traits receive their values during the formatting process, or in the case of absolutely positioned areas, during refinement.

The block-progression-dimension and inline-progression-dimension of an area represent the extent of the content-rectangle of that area in each of the two relative dimensions.

Other traits include:

Unless otherwise specified, the traits of a formatting object are present on each of its generated areas, and with the same value. (However, see sections and .)

As described above, the content-rectangle is the rectangle bounding the inside of the padding and is used to describe the constraints on the positions of descendant areas. It is possible that marks from descendant glyphs or other areas may appear outside the content-rectangle.

Related to this is the allocation-rectangle of an area, which is used to describe the constraints on the position of the area within its parent area. For an inline-area this is either the normal-allocation-rectangle or the large-allocation-rectangle. The normal-allocation-rectangle extends to the content-rectangle in the block-progression-direction and to the border-rectangle in the inline-progression-direction. The large-allocation-rectangle is the border-rectangle. Unless otherwise specified, the allocation-rectangle for an area is the normal-allocation-rectangle.

For a block-area, the allocation-rectangle extends to the border-rectangle in the block-progression-direction and outside the content-rectangle in the inline-progression-direction by an amount equal to the end-indent, and in the opposite direction by an amount equal to the start-indent.

The edges of a rectangle are designated as follows:

For purposes of this definition, the content-rectangle of an area uses the inline-progression-direction and block-progression-direction of that area; but the border-rectangle, padding-rectangle, and allocation-rectangle use the directions of its parent area. Thus the edges designated for the content-rectangle may not correspond to the same-named edges on the padding-, border-, and allocation-rectangles. This is important in the case of nested block-areas with different writing-modes.

The following diagram shows the correspondence between the various edge names for a mixed writing-mode example:

In the area tree, the set of areas with a given parent is ordered. The terms initial, final, preceding, and following refer to this ordering.

In any ordered tree, this sibling order extends to an ordering of the entire tree in at least two ways.

"Preceding" and "following", when applied to non-siblings, will depend on the extension order used, which must be specified. However, in either of these given orders, the leaves of the tree (nodes without children) are unambiguously ordered.

This section defines the notion of block-stacking constraints and inline-stacking constraints involving areas. These are defined as ordered relations, i.e., if A and B have a stacking constraint it does not necessarily mean that B and A have a stacking constraint. These definitions are recursive in nature and some cases may depend upon simpler cases of the same definition. This is not circularity but rather a consequence of recursion. The intention of the definitions is to identify areas at any level of the tree which have only space between them.

The area-class trait is an enumerated value which is xsl-normal for an area which is stacked with other areas in sequence. A normal area is an area for which this trait is xsl-normal. A page-level-out-of-line area is an area with area-class xsl-footnote, xsl-before-float, or xsl-fixed; placement of these areas is controlled by the fo:page-sequence ancestor of its generating formatting object. A reference-level-out-of-line area is an area with area-class xsl-side-float or xsl-absolute; placement of these areas is controlled by the formatting object generating the relevant reference-area. An anchor area is an area with area-class xsl-anchor; placement of these areas is arbitrary and does not affect stacking. Areas with area-class equal to one of xsl-normal, xsl-footnote, or xsl-before-float are defined to be stackable, indicating that they are supposed to be properly stacked.

Block-stacking constraints

If P is a block-area, then there is a fence preceding P if P is a reference-area or if the border-before-width or padding-before-width of P are non-zero. Similarly, there is a fence following P if P is a reference-area or if the border-after-width or padding-after-width of P are non-zero.

If A and B are stackable areas, and S is a sequence of space-specifiers (see ), it is defined that A and B have block-stacking constraint S if any of the following conditions holds:

When A and B have a block-stacking constraint, the adjacent edges of A and B are an ordered pair recursively defined as:

Example. In this diagram each node represents a block-area. Assume that all padding and border widths are zero, and none of the areas are reference-areas. Then P and A have a block-stacking constraint, as do A and B, A and C, B and C, C and D, D and B, B and E, D and E, and E and P; these are the only pairs in the diagram having block-stacking constraints. If B had non-zero padding-after, then D and E would not have any block-stacking constraint (though B and E would continue to have a block-stacking constraint).

Inline-stacking constraints.

This section will recursively define the inline-stacking constraints between two areas (either two inline-areas or one inline-area and one line-area), together with the notion of fence preceding and fence following; these definitions are interwoven with one another. This parallels the definition for block-stacking constraints, but with the additional complication that we may have a stacking constraint between inline-areas which are stacked in opposite inline-progression-directions. (This is not an issue for block-stacking constraints because a block-area which is not a reference-area may not have a block-progression-direction different from that of its parent.)

If P and Q have an inline-stacking constraint, then P has a fence preceding Q if P is a reference-area or has non-zero border-width or padding-width at the first adjacent edge of P and Q. Similarly, Q has a fence following P if Q is a reference-area or has non-zero border-width or padding-width at the second adjacent edge of P and Q.

If A and B are normal areas, and S is a sequence of space-specifiers, it is defined that A and B have inline-stacking constraint S if any of the following conditions holds:

When A and B have an inline-stacking constraint, the adjacent edges of A and B are an ordered pair defined as:

Two areas are adjacent if they have a block-stacking constraint or an inline-stacking constraint. It follows from the definitions that areas of the same type (inline or block) can be adjacent only if all their non-common ancestors are also of the same type (up to but not including their nearest common ancestor). Thus, for example, two inline-areas which reside in different line-areas are never adjacent.

An area A begins an area P if A is a descendant of P and P and A have either a block-stacking constraint or an inline-stacking constraint. In this case the second of the adjacent edges of P and A is defined to be a leading edge in P. A space-specifier which applies to the leading edge is also defined to begin P.

Similarly, An area A ends an area P if A is a descendant of P and A and P have either a block-stacking constraint or an inline-stacking constraint. In this case the first of the adjacent edges of A and P is defined to be a trailing edge in P. A space-specifier which applies to the trailing edge is also defined to end P.

Each script has its preferred "baseline" for aligning glyphs from that script. Western scripts typically use an "alphabetic" baseline that touches at or near the bottom of capital letters. Further, for each font there is a preferred way of aligning embedded glyphs from different scripts, e.g., for a Western font there are separate baselines for aligning embedded ideographic or Indic glyphs.

Each block-area and inline-area has a dominant-baseline-identifier trait whose value is a baseline identifier corresponding to the type of alignment expected for inline-area descendants of that area, and each inline-area has an alignment-baseline which specifies how the area is aligned to its parent. These traits are interpreted as described in section .

For each font, an actual-baseline-table maps these identifiers to points on the start-edge of the area. By abuse of terminology, the line in the inline-progression-direction through the point corresponding to the dominant-baseline-identifier is called the "dominant baseline."

The resolved space-specifier of a given space-specifier S is computed as follows. Consider the maximal inline-stacking constraint or block-stacking constraint S'' containing the space-specifier S as an element of the sequence (S'' is a sequence of space-specifiers; see ). Define S' to be a subsequence of S'' as follows:

The resolved space-specifier of S is a non-conditional, forcing space-specifier computed in terms of the sequence S'.

Example. Suppose the sequence of space values occurring at the beginning of a reference-area is: first, a space with value 10 points (that is minimum, optimum, and maximum all equal to 10 points) and conditionality discard; second, a space with value 4 points and conditionality retain; and third, a space with value 5 points and conditionality discard; all three spaces having precedence zero. Then the first (10 point) space is suppressed under rule 1, and the second (4 point) space is suppressed under rule 3. The resolved value of the third space is a non-conditional 5 points, even though it originally came from a conditional space.

The padding of a block-area does not interact with any space-specifier (except that by definition, the presence of padding at the before- or after-edge prevents areas on either side of it from having a stacking constraint.)

The border or padding at the before-edge or after-edge of a block-area B may be specified as conditional. If so, then it is set to zero if its associated edge is a leading edge in a reference-area, and the is-first trait of B is false, or if its associated edge is a trailing edge in a reference-area, and the is-last trait of B is false. In either of these cases, the border or padding is taken to be zero for purposes of the stacking constraint definitions.

The border or padding at the start-edge or end-edge of an inline-area I may be specified as conditional. If so, then it is set to zero if its associated edge is a leading edge in a line-area, and the is-first trait of I is false, or if its associated edge is a trailing edge in a line-area, and the is-last trait of I is false. In either of these cases, the border or padding is taken to be zero for purposes of the stacking constraint definitions.

When an area P is generated by a formatting object whose block-progression-dimension is "auto", then the constraints involving the before-edge and after-edge of the content-rectangle of P, together with the constraints between the various descendants of P, result in a constraint on the actual value of the block-progression-dimension. If the block-progression-dimension is instead specified as a length, then this might result in an overconstrained area tree, for example an incompletely-filled fo:block with a specified size. In that case some constraints between P and its descendants should be relaxed; those that are eligible for this treatment are said to be subject to overconstrainment relaxing, and treated as in the previous section.

Block-area children of an area are typically stacked in the block-progression-direction within their parent area, and this is the default method of positioning block-areas. However, formatting objects are free to specify other methods of positioning child areas of areas which they generate, for example list-items or tables.

For a parent area P whose children are block-areas, P is defined to be properly stacked if all of the following conditions hold:

Intrusion adjustments (both start- and end-) are defined to account for the indentation that occurs as the result of side floats.

If A and B are areas which have the same nearest reference area ancestor, then A and B are defined to be inline-overlapping if there is some line parallel to the inline-progression-direction, which intersects both the allocation-rectangle of A and the allocation-rectangle of B.

If A is an area of class xsl-side-float with float="start", and B is a block-area, and A and B have the same nearest reference area ancestor, then A is defined to encroach upon B if A and B are inline-overlapping and the start-indent of B is less than the sum of the start-indent of A and the inline-progression-dimension of A. The start-encroachment of A on B is then defined to be amount by which the start-indent of B is less than the sum of the start-indent of A and the inline-progression-dimension of A.

If A is an area of class xsl-side-float with float="end", and B is a block-area, and A and B have the same nearest reference area ancestor, then A is defined to encroach upon B if A and B are inline-overlapping and the end-indent of B is less than the sum of the end-indent of A and the inline-progression-dimension of A. The end-encroachment of A on B is then defined to be amount by which the end-indent of B is less than the sum of the end-indent of A and the inline-progression-dimension of A.

If B is a block-area which is not a line-area, then its local-start-intrusion-adjustment is computed as the maximum of the following lengths:

The start-intrusion-adjustment of a block-area B is then defined to be the maximum of the local-start-intrusion-adjustments of the normal block-areas generated and returned by the generating formatting object of B.

If L is a line-area, then its start-intrusion-adjustment is computed as the maximum of the following lengths:

The end-intrusion-adjustment for a block-area is computed in a precisely analogous manner. That is:

If B is a block-area which is not a line-area, then its local-end-intrusion-adjustment is computed as the maximum of the following lengths:

The end-intrusion-adjustment of a block-area B is then defined to be the maximum of the local-end-intrusion-adjustments of the normal block-areas generated and returned by the generating formatting object of B.

If L is a line-area, then its end-intrusion-adjustment is computed as the maximum of the following lengths:

Inline-area children of an area are typically stacked in the inline-progression-direction within their parent area, and this is the default method of positioning inline-areas.

Inline-areas are stacked relative to the dominant baseline, as defined above ().

For a parent area P whose children are inline-areas, P is defined to be properly stacked if all of the following conditions hold:

The most common inline-area is a glyph-area, which contains the representation for a character (or characters) in a particular font.

A glyph-area has an associated nominal-font, determined by the area's typographic traits, which apply to its character data, and a glyph-orientation determined by its writing-mode and reference-orientation, which determine the orientation of the glyph when it is rendered.

The alignment-point and dominant-baseline-identifier of a glyph-area are assigned according to the writing-system in use (e.g., the glyph baseline in Western languages), and are used to control placement of inline-areas descendants of a line-area. The formatter may generate inline-areas with different inline-progression-directions from their parent to accommodate correct inline-area stacking in the case of mixed writing systems.

A glyph-area has no children. Its block-progression-dimension and actual-baseline-table are the same for all glyphs in a font. Conforming implementations may choose to compute the block-progression-dimension for a glyph area based on the actual glyph size rather than using a common size for all glyphs in a font.

A subset S of the areas returned to a formatting object is called properly ordered if the areas in that subset have the same order as their generating formatting objects. Specifically, if A₁ and A₂ are areas in S, returned by child formatting objects F₁ and F₂ where F₁ precedes F₂, then A₁ must precede A₂ in the pre-order traversal order of the area tree. If F₁ equals F₂ and A₁ is returned prior to A₂, then A₁ must precede A₂ in the pre-order-traversal of the area tree.

For each formatting object F and each area-class C, the subset consisting of the areas returned to F with area-class C must be properly ordered, except where otherwise specified.

This section describes the ordering constraints that apply to formatting an fo:block or similar block-level object.

A block-level formatting object F which constructs lines does so by constructing block-areas which it returns to its parent formatting object, and placing normal areas and/or anchor areas returned to F by its child formatting objects as children of those block-areas or of line-areas which it constructs as children of those block-areas.

For each such formatting object F, it must be possible to form an ordered partition P consisting of ordered subsets S₁, S₂, ..., S_n of the normal areas and anchor areas returned by the child formatting objects, such that the following are all satisfied:

Substitutions that replace a sequence of glyph-areas with a single glyph-area should only occur when the margin, border, and padding in the inline-progression-direction (start- and end-), baseline-shift, and letter-spacing values are zero, treat-as-word-space is false, and the values of all other relevant traits match (i.e., alignment-adjust, alignment-baseline, color trait, background traits, dominant-baseline-identifier, font traits, text-depth, text-altitude, glyph-orientation-horizontal, glyph-orientation-vertical, line-height, line-height-shift-adjustment, text-decoration, text-shadow).

This section describes the ordering constraints that apply to formatting an fo:inline or similar inline-level object.

An inline-level formatting object F which constructs one or more inline-areas does so by placing normal inline-areas and/or anchor inline-areas returned to F by its child formatting objects as children of inline-areas which it generates.

For each such formatting object F, it must be possible to form an ordered partition P consisting of ordered subsets S₁, S₂, ..., S_n of the normal and/or anchor inline-areas returned by the child formatting objects, such that the following are all satisfied:

Each area is rendered in a particular location. Formatting object semantics describe the location of intrinsic marks relative to the object's location, i.e., the left, right, top, and bottom edges of its content-rectangle. This section describes how the area's location is determined, which determines the location of its intrinsic m