source: trunk/third/libxslt/doc/internals.html @ 18543

Revision 18543, 27.0 KB checked in by ghudson, 21 years ago (diff)
This commit was generated by cvs2svn to compensate for changes in r18542, which included commits to RCS files with non-trunk default branches.
Line 
1<!DOCTYPE HTML PUBLIC "-//W3C//DTD HTML 4.01 Transitional//EN" "http://www.w3.org/TR/1999/REC-html401-19991224/loose.dtd">
2<html>
3<head>
4<meta http-equiv="Content-Type" content="text/html; charset=ISO-8859-1">
5<style type="text/css"><!--
6TD {font-family: Verdana,Arial,Helvetica}
7BODY {font-family: Verdana,Arial,Helvetica; margin-top: 2em; margin-left: 0em; margin-right: 0em}
8H1 {font-family: Verdana,Arial,Helvetica}
9H2 {font-family: Verdana,Arial,Helvetica}
10H3 {font-family: Verdana,Arial,Helvetica}
11A:link, A:visited, A:active { text-decoration: underline }
12--></style>
13<title>Library internals</title>
14</head>
15<body bgcolor="#8b7765" text="#000000" link="#000000" vlink="#000000">
16<table border="0" width="100%" cellpadding="5" cellspacing="0" align="center"><tr>
17<td width="100">
18<a href="http://www.gnome.org/"><img src="gnome2.png" alt="Gnome2 Logo"></a><a href="http://www.redhat.com"><img src="redhat.gif" alt="Red Hat Logo"></a><div align="left"><a href="http://xmlsoft.org/XSLT/"><img src="Libxslt-Logo-180x168.gif" alt="Made with Libxslt Logo"></a></div>
19</td>
20<td><table border="0" width="90%" cellpadding="2" cellspacing="0" align="center" bgcolor="#000000"><tr><td><table width="100%" border="0" cellspacing="1" cellpadding="3" bgcolor="#fffacd"><tr><td align="center">
21<h1>The XSLT C library for Gnome</h1>
22<h2>Library internals</h2>
23</td></tr></table></td></tr></table></td>
24</tr></table>
25<table border="0" cellpadding="4" cellspacing="0" width="100%" align="center"><tr><td bgcolor="#8b7765"><table border="0" cellspacing="0" cellpadding="2" width="100%"><tr>
26<td valign="top" width="200" bgcolor="#8b7765"><table border="0" cellspacing="0" cellpadding="1" width="100%" bgcolor="#000000"><tr><td>
27<table width="100%" border="0" cellspacing="1" cellpadding="3">
28<tr><td colspan="1" bgcolor="#eecfa1" align="center"><center><b>Main Menu</b></center></td></tr>
29<tr><td bgcolor="#fffacd">
30<form action="search.php" enctype="application/x-www-form-urlencoded" method="GET">
31<input name="query" type="TEXT" size="20" value=""><input name="submit" type="submit" value="Search ...">
32</form>
33<ul>
34<li><a href="index.html">Home</a></li>
35<li><a href="intro.html">Introduction</a></li>
36<li><a href="docs.html">Documentation</a></li>
37<li><a href="bugs.html">Reporting bugs and getting help</a></li>
38<li><a href="help.html">How to help</a></li>
39<li><a href="downloads.html">Downloads</a></li>
40<li><a href="FAQ.html">FAQ</a></li>
41<li><a href="news.html">News</a></li>
42<li><a href="xsltproc2.html">The xsltproc tool</a></li>
43<li><a href="docbook.html">DocBook</a></li>
44<li><a href="API.html">The programming API</a></li>
45<li><a href="python.html">Python and bindings</a></li>
46<li><a href="internals.html">Library internals</a></li>
47<li><a href="extensions.html">Writing extensions</a></li>
48<li><a href="contribs.html">Contributions</a></li>
49<li>
50<a href="xslt.html">flat page</a>, <a href="site.xsl">stylesheet</a>
51</li>
52</ul>
53</td></tr>
54</table>
55<table width="100%" border="0" cellspacing="1" cellpadding="3">
56<tr><td colspan="1" bgcolor="#eecfa1" align="center"><center><b>Related links</b></center></td></tr>
57<tr><td bgcolor="#fffacd"><ul>
58<li><a href="tutorial/libxslttutorial.html">Tutorial</a></li>
59<li><a href="xsltproc.html">Man page for xsltproc</a></li>
60<li><a href="http://mail.gnome.org/archives/xslt/">Mail archive</a></li>
61<li><a href="http://xmlsoft.org/">XML libxml</a></li>
62<li><a href="http://phd.cs.unibo.it/gdome2/">DOM gdome2</a></li>
63<li><a href="ftp://xmlsoft.org/">FTP</a></li>
64<li><a href="http://www.zlatkovic.com/projects/libxml/">Windows binaries</a></li>
65<li><a href="http://garypennington.net/libxml2/">Solaris binaries</a></li>
66<li><a href="http://www.zveno.com/open_source/libxml2xslt.html">MacOsX binaries</a></li>
67<li><a href="http://sourceforge.net/projects/libxml2-pas/">Pascal bindings</a></li>
68<li><a href="http://bugzilla.gnome.org/buglist.cgi?product=libxslt">Bug Tracker</a></li>
69<li><a href="http://xsldbg.sourceforge.net/">Xsldbg Debugger</a></li>
70</ul></td></tr>
71</table>
72<table width="100%" border="0" cellspacing="1" cellpadding="3">
73<tr><td colspan="1" bgcolor="#eecfa1" align="center"><center><b>API Indexes</b></center></td></tr>
74<tr><td bgcolor="#fffacd"><ul>
75<li><a href="APIchunk0.html">Alphabetic</a></li>
76<li><a href="APIconstructors.html">Constructors</a></li>
77<li><a href="APIfunctions.html">Functions/Types</a></li>
78<li><a href="APIfiles.html">Modules</a></li>
79<li><a href="APIsymbols.html">Symbols</a></li>
80</ul></td></tr>
81</table>
82</td></tr></table></td>
83<td valign="top" bgcolor="#8b7765"><table border="0" cellspacing="0" cellpadding="1" width="100%"><tr><td><table border="0" cellspacing="0" cellpadding="1" width="100%" bgcolor="#000000"><tr><td><table border="0" cellpadding="3" cellspacing="1" width="100%"><tr><td bgcolor="#fffacd">
84<h3>Table  of contents</h3>
85<ul>
86<li><a href="internals.html#Introducti">Introduction</a></li>
87  <li><a href="internals.html#Basics">Basics</a></li>
88  <li><a href="internals.html#Keep">Keep it simple stupid</a></li>
89  <li><a href="internals.html#libxml">The libxml nodes</a></li>
90  <li><a href="internals.html#XSLT">The XSLT processing steps</a></li>
91  <li><a href="internals.html#XSLT1">The XSLT stylesheet compilation</a></li>
92  <li><a href="internals.html#XSLT2">The XSLT template compilation</a></li>
93  <li><a href="internals.html#processing">The processing itself</a></li>
94  <li><a href="internals.html#XPath">XPath expressions compilation</a></li>
95  <li><a href="internals.html#XPath1">XPath interpretation</a></li>
96  <li><a href="internals.html#Descriptio">Description of XPath
97  Objects</a></li>
98  <li><a href="internals.html#XPath3">XPath functions</a></li>
99  <li><a href="internals.html#stack">The variables stack frame</a></li>
100  <li><a href="internals.html#Extension">Extension support</a></li>
101  <li><a href="internals.html#Futher">Further reading</a></li>
102  <li><a href="internals.html#TODOs">TODOs</a></li>
103</ul>
104<h3><a name="Introducti2">Introduction</a></h3>
105<p>This document describes the processing of <a href="http://xmlsoft.org/XSLT/">libxslt</a>, the <a href="http://www.w3.org/TR/xslt">XSLT</a> C library developed for the <a href="http://www.gnome.org/">Gnome</a> project.</p>
106<p>Note: this documentation is by definition incomplete and I am not good at
107spelling, grammar, so patches and suggestions are <a href="mailto:veillard@redhat.com">really welcome</a>.</p>
108<h3><a name="Basics1">Basics</a></h3>
109<p>XSLT is a transformation language. It takes an input document and a
110stylesheet document and generates an output document:</p>
111<p align="center"><img src="processing.gif" alt="the XSLT processing model"></p>
112<p>Libxslt is written in C. It relies on <a href="http://www.xmlsoft.org/">libxml</a>, the XML C library for Gnome, for
113the following operations:</p>
114<ul>
115<li>parsing files</li>
116  <li>building the in-memory DOM structure associated with the documents
117    handled</li>
118  <li>the XPath implementation</li>
119  <li>serializing back the result document to XML and HTML. (Text is handled
120    directly.)</li>
121</ul>
122<h3><a name="Keep1">Keep it simple stupid</a></h3>
123<p>Libxslt is not very specialized. It is built under the assumption that all
124nodes from the source and output document can fit in the virtual memory of
125the system. There is a big trade-off there. It is fine for reasonably sized
126documents but may not be suitable for large sets of data. The gain is that it
127can be used in a relatively versatile way. The input or output may never be
128serialized, but the size of documents it can handle are limited by the size
129of the memory available.</p>
130<p>More specialized memory handling approaches are possible, like building
131the input tree from a serialization progressively as it is consumed,
132factoring repetitive patterns, or even on-the-fly generation of the output as
133the input is parsed but it is possible only for a limited subset of the
134stylesheets. In general the implementation of libxslt follows the following
135pattern:</p>
136<ul>
137<li>KISS (keep it simple stupid)</li>
138  <li>when there is a clear bottleneck optimize on top of this simple
139    framework and refine only as much as is needed to reach the expected
140    result</li>
141</ul>
142<p>The result is not that bad, clearly one can do a better job but more
143specialized too. Most optimization like building the tree on-demand would
144need serious changes to the libxml XPath framework. An easy step would be to
145serialize the output directly (or call a set of SAX-like output handler to
146keep this a flexible interface) and hence avoid the memory consumption of the
147result.</p>
148<h3><a name="libxml">The libxml nodes</a></h3>
149<p>DOM-like trees, as used and generated by libxml and libxslt, are
150relatively complex. Most node types follow the given structure except a few
151variations depending on the node type:</p>
152<p align="center"><img src="node.gif" alt="description of a libxml node"></p>
153<p>Nodes carry a <strong>name</strong> and the node <strong>type</strong>
154indicates the kind of node it represents, the most common ones are:</p>
155<ul>
156<li>document nodes</li>
157  <li>element nodes</li>
158  <li>text nodes</li>
159</ul>
160<p>For the XSLT processing, entity nodes should not be generated (i.e. they
161should be replaced by their content). Most nodes also contains the following
162&quot;navigation&quot; informations:</p>
163<ul>
164<li>the containing <strong>doc</strong>ument</li>
165  <li>the <strong>parent</strong> node</li>
166  <li>the first <strong>children</strong> node</li>
167  <li>the <strong>last</strong> children node</li>
168  <li>the <strong>prev</strong>ious sibling</li>
169  <li>the following sibling (<strong>next</strong>)</li>
170</ul>
171<p>Elements nodes carries the list of attributes in the properties, an
172attribute itself holds the navigation pointers and the children list (the
173attribute value is not represented as a simple string to allow usage of
174entities references).</p>
175<p>The <strong>ns</strong> points to the namespace declaration for the
176namespace associated to the node, <strong>nsDef</strong> is the linked list
177of namespace declaration present on element nodes.</p>
178<p>Most nodes also carry an <strong>_private</strong> pointer which can be
179used by the application to hold specific data on this node.</p>
180<h3><a name="XSLT">The XSLT processing steps</a></h3>
181<p>There are a few steps which are clearly decoupled at the interface
182level:</p>
183<ol>
184<li>parse the stylesheet and generate a DOM tree</li>
185  <li>take the stylesheet tree and build a compiled version of it (the
186    compilation phase)</li>
187  <li>take the input and generate a DOM tree</li>
188  <li>process the stylesheet against the input tree and generate an output
189    tree</li>
190  <li>serialize the output tree</li>
191</ol>
192<p>A few things should be noted here:</p>
193<ul>
194<li>the steps 1/ 3/ and 5/ are optional</li>
195  <li>the stylesheet obtained at 2/ can be reused by multiple processing 4/
196    (and this should also work in threaded programs)</li>
197  <li>the tree provided in 2/ should never be freed using xmlFreeDoc, but by
198    freeing the stylesheet.</li>
199  <li>the input tree 4/ is not modified except the _private field which may
200    be used for labelling keys if used by the stylesheet</li>
201</ul>
202<h3><a name="XSLT1">The XSLT stylesheet compilation</a></h3>
203<p>This is the second step described. It takes a stylesheet tree, and
204&quot;compiles&quot; it. This associates to each node a structure stored in the
205_private field and containing information computed in the stylesheet:</p>
206<p align="center"><img src="stylesheet.gif" alt="a compiled XSLT stylesheet"></p>
207<p>One xsltStylesheet structure is generated per document parsed for the
208stylesheet. XSLT documents allow includes and imports of other documents,
209imports are stored in the <strong>imports</strong> list (hence keeping the
210tree hierarchy of includes which is very important for a proper XSLT
211processing model) and includes are stored in the <strong>doclist</strong>
212list. An imported stylesheet has a parent link to allow browsing of the
213tree.</p>
214<p>The DOM tree associated to the document is stored in <strong>doc</strong>.
215It is preprocessed to remove ignorable empty nodes and all the nodes in the
216XSLT namespace are subject to precomputing. This usually consist of
217extracting all the context information from the context tree (attributes,
218namespaces, XPath expressions), and storing them in an xsltStylePreComp
219structure associated to the <strong>_private</strong> field of the node.</p>
220<p>A couple of notable exceptions to this are XSLT template nodes (more on
221this later) and attribute value templates. If they are actually templates,
222the value cannot be computed at compilation time. (Some preprocessing could
223be done like isolation and preparsing of the XPath subexpressions but it's
224not done, yet.)</p>
225<p>The xsltStylePreComp structure also allows storing of the precompiled form
226of an XPath expression that can be associated to an XSLT element (more on
227this later).</p>
228<h3><a name="XSLT2">The XSLT template compilation</a></h3>
229<p>A proper handling of templates lookup is one of the keys of fast XSLT
230processing. (Given a node in the source document this is the process of
231finding which templates should be applied to this node.) Libxslt follows the
232hint suggested in the <a href="http://www.w3.org/TR/xslt#patterns">5.2
233Patterns</a> section of the XSLT Recommendation, i.e. it doesn't evaluate it
234as an XPath expression but tokenizes it and compiles it as a set of rules to
235be evaluated on a candidate node. There usually is an indication of the node
236name in the last step of this evaluation and this is used as a key check for
237the match. As a result libxslt builds a relatively more complex set of
238structures for the templates:</p>
239<p align="center"><img src="templates.gif" alt="The templates related structure"></p>
240<p>Let's describe a bit more closely what is built. First the xsltStylesheet
241structure holds a pointer to the template hash table. All the XSLT patterns
242compiled in this stylesheet are indexed by the value of the the target
243element (or attribute, pi ...) name, so when a element or an attribute &quot;foo&quot;
244needs to be processed the lookup is done using the name as a key.</p>
245<p>Each of the patterns is compiled into an xsltCompMatch structure. It holds
246the set of rules based on the tokenization of the pattern stored in reverse
247order (matching is easier this way). It also holds some information about the
248previous matches used to speed up the process when one iterates over a set of
249siblings. (This optimization may be defeated by trashing when running
250threaded computation, it's unclear that this is a big deal in practice.)
251Predicate expressions are not compiled at this stage, they may be at run-time
252if needed, but in this case they are compiled as full XPath expressions (the
253use of some fixed predicate can probably be optimized, they are not yet).</p>
254<p>The xsltCompMatch are then stored in the hash table, the clash list is
255itself sorted by priority of the template to implement &quot;naturally&quot; the XSLT
256priority rules.</p>
257<p>Associated to the compiled pattern is the xsltTemplate itself containing
258the information required for the processing of the pattern including, of
259course, a pointer to the list of elements used for building the pattern
260result.</p>
261<p>Last but not least a number of patterns do not fit in the hash table
262because they are not associated to a name, this is the case for patterns
263applying to the root, any element, any attributes, text nodes, pi nodes, keys
264etc. Those are stored independently in the stylesheet structure as separate
265linked lists of xsltCompMatch.</p>
266<h3><a name="processing">The processing itself</a></h3>
267<p>The processing is defined by the XSLT specification (the basis of the
268algorithm is explained in <a href="http://www.w3.org/TR/xslt#section-Introduction">the Introduction</a>
269section). Basically it works by taking the root of the input document and
270applying the following algorithm:</p>
271<ol>
272<li>Finding the template applying to it. This is a lookup in the template
273    hash table, walking the hash list until the node satisfies all the steps
274    of the pattern, then checking the appropriate(s) global templates to see
275    if there isn't a higher priority rule to apply</li>
276  <li>If there is no template, apply the default rule (recurse on the
277    children)</li>
278  <li>else walk the content list of the selected templates, for each of them:
279    <ul>
280<li>if the node is in the XSLT namespace then the node has a _private
281        field pointing to the preprocessed values, jump to the specific
282      code</li>
283      <li>if the node is in an extension namespace, look up the associated
284        behavior</li>
285      <li>otherwise copy the node.</li>
286    </ul>
287<p>The closure is usually done through the XSLT
288    <strong>apply-templates</strong> construct recursing by applying the
289    adequate template on the input node children or on the result of an
290    associated XPath selection lookup.</p>
291  </li>
292</ol>
293<p>Note that large parts of the input tree may not be processed by a given
294stylesheet and that on the opposite some may be processed multiple times.
295(This often is the case when a Table of Contents is built).</p>
296<p>The module <code>transform.c</code> is the one implementing most of this
297logic. <strong>xsltApplyStylesheet()</strong> is the entry point, it
298allocates an xsltTransformContext containing the following:</p>
299<ul>
300<li>a pointer to the stylesheet being processed</li>
301  <li>a stack of templates</li>
302  <li>a stack of variables and parameters</li>
303  <li>an XPath context</li>
304  <li>the template mode</li>
305  <li>current document</li>
306  <li>current input node</li>
307  <li>current selected node list</li>
308  <li>the current insertion points in the output document</li>
309  <li>a couple of hash tables for extension elements and functions</li>
310</ul>
311<p>Then a new document gets allocated (HTML or XML depending on the type of
312output), the user parameters and global variables and parameters are
313evaluated. Then <strong>xsltProcessOneNode()</strong> which implements the
3141-2-3 algorithm is called on the root element of the input. Step 1/ is
315implemented by calling <strong>xsltGetTemplate()</strong>, step 2/ is
316implemented by <strong>xsltDefaultProcessOneNode()</strong> and step 3/ is
317implemented by <strong>xsltApplyOneTemplate()</strong>.</p>
318<h3><a name="XPath">XPath expression compilation</a></h3>
319<p>The XPath support is actually implemented in the libxml module (where it
320is reused by the XPointer implementation). XPath is a relatively classic
321expression language. The only uncommon feature is that it is working on XML
322trees and hence has specific syntax and types to handle them.</p>
323<p>XPath expressions are compiled using <strong>xmlXPathCompile()</strong>.
324It will take an expression string in input and generate a structure
325containing the parsed expression tree, for example the expression:</p>
326<pre>/doc/chapter[title='Introduction']</pre>
327<p>will be compiled as</p>
328<pre>Compiled Expression : 10 elements
329  SORT
330    COLLECT  'child' 'name' 'node' chapter
331      COLLECT  'child' 'name' 'node' doc
332        ROOT
333      PREDICATE
334        SORT
335          EQUAL =
336            COLLECT  'child' 'name' 'node' title
337              NODE
338            ELEM Object is a string : Introduction
339              COLLECT  'child' 'name' 'node' title
340                NODE</pre>
341<p>This can be tested using the <code>testXPath</code>  command (in the
342libxml codebase) using the <code>--tree</code> option.</p>
343<p>Again, the KISS approach is used. No optimization is done. This could be
344an interesting thing to add. <a href="http://www-106.ibm.com/developerworks/library/x-xslt2/?dwzone=x?open&amp;l=132%2ct=gr%2c%2Bp=saxon">Michael
345Kay describes</a> a lot of possible and interesting optimizations done in
346Saxon which would be possible at this level. I'm unsure they would provide
347much gain since the expressions tends to be relatively simple in general and
348stylesheets are still hand generated. Optimizations at the interpretation
349sounds likely to be more efficient.</p>
350<h3><a name="XPath1">XPath interpretation</a></h3>
351<p>The interpreter is implemented by <strong>xmlXPathCompiledEval()</strong>
352which is the front-end to <strong>xmlXPathCompOpEval()</strong> the function
353implementing the evaluation of the expression tree. This evaluation follows
354the KISS approach again. It's recursive and calls
355<strong>xmlXPathNodeCollectAndTest()</strong> to collect nodes set when
356evaluating a <code>COLLECT</code> node.</p>
357<p>An evaluation is done within the framework of an XPath context stored in
358an <strong>xmlXPathContext</strong> structure, in the framework of a
359transformation the context is maintained within the XSLT context. Its content
360follows the requirements from the XPath specification:</p>
361<ul>
362<li>the current document</li>
363  <li>the current node</li>
364  <li>a hash table of defined variables (but not used by XSLT)</li>
365  <li>a hash table of defined functions</li>
366  <li>the proximity position (the place of the node in the current node
367  list)</li>
368  <li>the context size (the size of the current node list)</li>
369  <li>the array of namespace declarations in scope (there also is a namespace
370    hash table but it is not used in the XSLT transformation).</li>
371</ul>
372<p>For the purpose of XSLT an <strong>extra</strong> pointer has been added
373allowing to retrieve the XSLT transformation context. When an XPath
374evaluation is about to be performed, an XPath parser context is allocated
375containing and XPath object stack (this is actually an XPath evaluation
376context, this is a remain of the time where there was no separate parsing and
377evaluation phase in the XPath implementation). Here is an overview of the set
378of contexts associated to an XPath evaluation within an XSLT
379transformation:</p>
380<p align="center"><img src="contexts.gif" alt="The set of contexts associated "></p>
381<p>Clearly this is a bit too complex and confusing and should be refactored
382at the next set of binary incompatible releases of libxml. For example the
383xmlXPathCtxt has a lot of unused parts and should probably be merged with
384xmlXPathParserCtxt.</p>
385<h3><a name="Descriptio">Description of XPath Objects</a></h3>
386<p>An XPath expression manipulates XPath objects. XPath defines the default
387types boolean, numbers, strings and node sets. XSLT adds the result tree
388fragment type which is basically an unmodifiable node set.</p>
389<p>Implementation-wise, libxml follows again a KISS approach, the
390xmlXPathObject is a structure containing a type description and the various
391possibilities. (Using an enum could have gained some bytes.) In the case of
392node sets (or result tree fragments), it points to a separate xmlNodeSet
393object which contains the list of pointers to the document nodes:</p>
394<p align="center"><img src="object.gif" alt="An Node set object pointing to "></p>
395<p>The <a href="http://xmlsoft.org/html/libxml-xpath.html">XPath API</a> (and
396its <a href="http://xmlsoft.org/html/libxml-xpathinternals.html">'internal'
397part</a>) includes a number of functions to create, copy, compare, convert or
398free XPath objects.</p>
399<h3><a name="XPath3">XPath functions</a></h3>
400<p>All the XPath functions available to the interpreter are registered in the
401function hash table linked from the XPath context. They all share the same
402signature:</p>
403<pre>void xmlXPathFunc (xmlXPathParserContextPtr ctxt, int nargs);</pre>
404<p>The first argument is the XPath interpretation context, holding the
405interpretation stack. The second argument defines the number of objects
406passed on the stack for the function to consume (last argument is on top of
407the stack).</p>
408<p>Basically an XPath function does the following:</p>
409<ul>
410<li>check <code>nargs</code> for proper handling of errors or functions
411    with variable numbers of parameters</li>
412  <li>pop the parameters from the stack using <code>obj =
413    valuePop(ctxt);</code>
414</li>
415  <li>do the function specific computation</li>
416  <li>push the result parameter on the stack using <code>valuePush(ctxt,
417    res);</code>
418</li>
419  <li>free up the input parameters with
420  <code>xmlXPathFreeObject(obj);</code>
421</li>
422  <li>return</li>
423</ul>
424<p>Sometime the work can be done directly by modifying in-situ the top object
425on the stack <code>ctxt-&gt;value</code>.</p>
426<h3><a name="stack">The XSLT variables stack frame</a></h3>
427<p>Not to be confused with XPath object stack, this stack holds the XSLT
428variables and parameters as they are defined through the recursive calls of
429call-template, apply-templates and default templates. This is used to define
430the scope of variables being called.</p>
431<p>This part seems to be the most urgent attention right now, first it is
432done in a very inefficient way since the location of the variables and
433parameters within the stylesheet tree is still done at run time (it really
434should be done statically at compile time), and I am still unsure that my
435understanding of the template variables and parameter scope is actually
436right.</p>
437<p>This part of the documentation is still to be written once this part of
438the code will be stable. <span style="background-color: #FF0000">TODO</span>
439</p>
440<h3><a name="Extension">Extension support</a></h3>
441<p>There is a separate document explaining <a href="extensions.html">how the
442extension support works</a>.</p>
443<h3><a name="Futher">Further reading</a></h3>
444<p>Michael Kay wrote <a href="http://www-106.ibm.com/developerworks/library/x-xslt2/?dwzone=x?open&amp;l=132%2ct=gr%2c%2Bp=saxon">a
445really interesting article on Saxon internals</a> and the work he did on
446performance issues. I wishes I had read it before starting libxslt design (I
447would probably have avoided a few mistakes and progressed faster). A lot of
448the ideas in his papers should be implemented or at least tried in
449libxslt.</p>
450<p>The <a href="http://xmlsoft.org/">libxml documentation</a>, especially <a href="http://xmlsoft.org/xmlio.html">the I/O interfaces</a> and the <a href="http://xmlsoft.org/xmlmem.html">memory management</a>.</p>
451<h3><a name="TODOs">TODOs</a></h3>
452<p>redesign the XSLT stack frame handling. Far too much work is done at
453execution time. Similarly for the attribute value templates handling, at
454least the embedded subexpressions ought to be precompiled.</p>
455<p>Allow output to be saved to a SAX like output (this notion of SAX like API
456for output should be added directly to libxml).</p>
457<p>Implement and test some of the optimization explained by Michael Kay
458especially:</p>
459<ul>
460<li>static slot allocation on the stack frame</li>
461  <li>specific boolean interpretation of an XPath expression</li>
462  <li>some of the sorting optimization</li>
463  <li>Lazy evaluation of location path. (this may require more changes but
464    sounds really interesting. XT does this too.)</li>
465  <li>Optimization of an expression tree (This could be done as a completely
466    independent module.)</li>
467</ul>
468<p></p>
469<p>Error reporting, there is a lot of case where the XSLT specification
470specify that a given construct is an error are not checked adequately by
471libxslt. Basically one should do a complete pass on the XSLT spec again and
472add all tests to the stylesheet compilation. Using the DTD provided in the
473appendix and making direct checks using the libxml validation API sounds a
474good idea too (though one should take care of not raising errors for
475elements/attributes in different namespaces).</p>
476<p>Double check all the places where the stylesheet compiled form might be
477modified at run time (extra removal of blanks nodes, hint on the
478xsltCompMatch).</p>
479<p></p>
480<p><a href="bugs.html">Daniel Veillard</a></p>
481</td></tr></table></td></tr></table></td></tr></table></td>
482</tr></table></td></tr></table>
483</body>
484</html>
Note: See TracBrowser for help on using the repository browser.