On Jan 29, 2011, at 3:59 AM, sdumitriu (SVN) wrote:
  Author: sdumitriu
 Date: 2011-01-29 03:59:38 +0100 (Sat, 29 Jan 2011)
 New Revision: 34247
 Modified:
   platform/core/trunk/xwiki-core/src/main/java/com/xpn/xwiki/pdf/impl/PdfExportImpl.java
 Log:
 [refactoring] Better performance by avoiding repeated lookup of components
 Modified:
platform/core/trunk/xwiki-core/src/main/java/com/xpn/xwiki/pdf/impl/PdfExportImpl.java
 ===================================================================
 ---
platform/core/trunk/xwiki-core/src/main/java/com/xpn/xwiki/pdf/impl/PdfExportImpl.java
2011-01-29 02:52:13 UTC (rev 34246)
 +++
platform/core/trunk/xwiki-core/src/main/java/com/xpn/xwiki/pdf/impl/PdfExportImpl.java
2011-01-29 02:59:38 UTC (rev 34247)
 @@ -124,6 +124,12 @@
     /** Logging helper object. */
     private static final Log LOG = LogFactory.getLog(PdfExportImpl.class);
 +    /** Velocity engine manager, used for interpreting velocity. */
 +    private static VelocityManager velocityManager =
Utils.getComponent(VelocityManager.class);
 +
 +    /** The OpenOffice manager used for generating RTF. */
 +    private static OpenOfficeManager oooManager =
Utils.getComponent(OpenOfficeManager.class); 
This is not very since it will prevent the Extension Manager and the dynamic component
mechanism to work (if a new component impl overrides the role/hint).
We shouldn't use any static as much as possible. Note that a lookup doesn't cost
anything since there's no synchronize and it's only a concurrent hashmap get, so I
doubt it's going to improve perf by much (unless there's something I don't
understand).
Note that in the future an idea we've had with Thomas is that all @Requirement will
actually be proxies to the component impl so that if the component is replaced you'll
always get the current one.
Thanks
-Vincent
  +
     /** DOM parser factory. */
     private static DocumentBuilderFactory dbFactory =
DocumentBuilderFactory.newInstance();
 @@ -242,11 +248,9 @@
             LOG.debug("Final XHTML for export: " + xhtml);
         }
 -        OpenOfficeManager OpenOfficeManager =
Utils.getComponent(OpenOfficeManager.class);
 -
         // If OpenOffice server is connected use it instead of FOP which does not support
RTF very well
         // Only switch to openoffice server for RTF because FOP is supposedly a lot more
powerful for PDF
 -        if (type != PDF && OpenOfficeManager.getState() ==
ManagerState.CONNECTED) {
 +        if (type != PDF && oooManager.getState() == ManagerState.CONNECTED) {
             exportOffice(xhtml, out, type, context);
         } else {
             // XSL Transformation to XML-FO
 @@ -285,10 +289,8 @@
         // id attribute on body element makes openoffice converter to fail
         html =
html.replaceFirst("(<body[^>]+)id=\"body\"([^>]*>)",
"$1$2");
 -        OpenOfficeManager OpenOfficeManager =
Utils.getComponent(OpenOfficeManager.class);
 +        OpenOfficeConverter documentConverter = oooManager.getConverter();
 -        OpenOfficeConverter documentConverter = OpenOfficeManager.getConverter();
 -
         String inputFileName = "export_input.html";
         String outputFileName = "export_output" + (type == PdfExportImpl.RTF ?
".rtf" : ".pdf");
 @@ -605,7 +607,6 @@
                     Utils.getComponent(EntityReferenceSerializer.class);
                 try {
                     StringWriter writer = new StringWriter();
 -                    VelocityManager velocityManager =
Utils.getComponent(VelocityManager.class);
                     VelocityEngine engine = velocityManager.getVelocityEngine();
                     try {
                         VelocityContext vcontext = velocityManager.getVelocityContext();