[XSL-LIST Mailing List Archive Home] [By Thread] [By Date]

Re: [xsl] xsl 2.0?

Subject: Re: [xsl] xsl 2.0?
From: Toshihiko Makita <tmakita@xxxxxxxxxxxxx>
Date: Mon, 04 Nov 2013 19:23:41 +0900

Dear Liam Quin,

>It is not really technical but social - CSS is more widely used, and it
>is easier to find designers who can work in terms of CSS.

It sounds to me that it is hard to judge it to be a rational reason that people should  move from XSL-FO to CSS.

I want point out the fact that it will not become the benefit using CSS rather than XSL-FO even if it is widely used than XSL-FO. It is the maintenance & debugging cost.

Given an last publication A (such as PDF) and there are 500 styles are needed to make this publication. If we adopt XSL-FO, these styles are all embed as property of XSL formatting objects. 

[Example: DITA authoring snippet]
<!-- This is only a example. DITA has <b>,<u>,<i> elements. -->
<p outputclass="b u i">

[Converted XSL-FO]
<!-- font-family & font size are the style of "p". The others are obtained from @outputclass -->
<fo:block font-family="Arial" font-size="10pt" font-weight="bold" text-decoration="underline" font-style="italic">

In the case of CSS all of the styles are written in the another file like following.

[Converted XHTML]
<link rel="stylesheet" type="text/css" media="print" href="style_print.css" />

<div class="p b u i">


If we find unexpected formatting result in the PDF, we can examine XSL-FO file formatting objects and properties for it.

However in the case of CSS case we must examine all of the CSS styelsheet file attached and find the selector that are applied to the relevant formatting result. 

Next example shows more complex case:

[Example: DITA authoring snippet]
<!-- We want to apply color red for first paragraph. Also we want apply blue for last paragraph. -->
  <p>Paragraph 1</p>
  <p>Paragraph 2</p>
  <p>Paragraph 3</p>

[XSL-FO stylehesst]
<xsl:template match="body">
  <fo:block use-attribute-sets="body" >
    <xsl:apply-templates select="*"/>
<xsl:template match="p">
  <fo:block use-attribute-sets="p">
      <xsl:when test="position() eq 1">
        <xsl:attribute name="color" select="'red'"/>
      <xsl:when test="position() eq last()">
        <xsl:attribute name="color" select="'blue'"/>

[Result XSL-FO]
<fo:block ...>
  <fo:block font-family="Arial" font-size="10pt" color="red">Paragraph 1</fo:block>
  <fo:block font-family="Arial" font-size="10pt">Paragraph 2</fo:block>
  <fo:block font-family="Arial" font-size="10pt" color="blue">Paragraph 3</fo:block>

[CSS stylehesst]
<xsl:template match="body">
  <div class="body" >
    <xsl:apply-templates select="*"/>

<xsl:template match="p">
  <div class="p">

[XHTML result]
<div class="body">
  <div class="p">Paragraph 1</div>
  <div class="p">Paragraph 2</div>
  <div class="p">Paragraph 3</div>

[Attached CSS stylesheets]
// CSS programmer will prefer to use below selector than adding new class value.

In above example it is clear from XSL-FO that the aimed style has been applied before rendition (formatting) phase. In the CSS case the style is dynamically applied at rendering (formatting) phase. It needs more effort to debug the formatting results in CSS than XSL-FO.

The effort for maintaining and debugging for making publication does not become decreased by adopting of the CSS. Rather it will be increased by its architecture than XSL-FO especially in debugging.

I heard that production level CSS stylesheet become over 3000 steps. If the final publication is not so complex, CSS can be a good solution. However maintaining and debugging the styles for publication by CSS is not so easy than XSL-FO.

If my opinion has a misunderstanding, please indicate.


 Toshihiko Makita
 Development Group. Antenna House, Inc. Ina Branch
 E-Mail tmakita@xxxxxxxxxxxxx
 8077-1 Horikita Minamiminowa Vil. Kamiina Co.
 Nagano Pref. 399-4511 Japan
 Tel +81-265-76-9300 Fax +81-265-78-1668
 Web site:

Current Thread