Re: Small performance change

Anthony,

Yes, the performance hit is pretty big. My profiling timings put the running 
of that method at around 1ms. This on its own is not huge, but does result in 
some cases in OFBiz ecommerce pages the increase of about 10ms of unnecessary 
time.

I was looking for this sort of thing last night in the OFBiz code and noticed 
this one here in JPublish. I'll be looking in more detail at JPublish and 
other projects over the next few days to find other similar things. A few 
milliseconds here and there could result in dropping the total page load time 
by a couple of hundred milliseconds, which is what I am going for.

Later,
-David


On Friday 07 March 2003 06:50, Anthony Eden wrote:
> The idea of having a map of classes was to support the addition of new URI
> types without having to change the code.  However if you are showing a
> significant performance hit because of it I will modify the code to match
> your patch.
>
> -Anthony
>
> David E Jones wrote:
> > This method is called quite a few times, to the point where it is a
> > noticeable amount. It doesn't look like the dynamic new instance is
> > necessary, and a new is much faster, so attached is an alternative.
> >
> > Later,
> > -David Jones
> >
> >
> > ------------------------------------------------------------------------
> >
> > package org.jpublish.util;
> >
> > import java.util.HashMap;
> >
> > public class InternalURIParser{
> >
> >     private static final InternalURIParser INSTANCE = new
> > InternalURIParser(); //private HashMap uriClasses;
> >
> >     protected InternalURIParser(){
> >         //uriClasses = new HashMap();
> >         //uriClasses.put("template", InternalURI.class);
> >         //uriClasses.put("repository", RepositoryURI.class);
> >     }
> >
> >     public static InternalURIParser getInstance(){
> >         return INSTANCE;
> >     }
> >
> >     public InternalURI parse(String uriString) throws Exception{
> >         int protocolTerminatorIndex = uriString.indexOf(":");
> >         if(protocolTerminatorIndex <= 0){
> >             throw new IllegalArgumentException("URI string is not a valid
> > URI"); }
> >
> >         String protocol = uriString.substring(0,
> > protocolTerminatorIndex); //Class uriClass =
> > (Class)uriClasses.get(protocol);
> >         //InternalURI uri = (InternalURI)uriClass.newInstance();
> >         InternalURI uri = null;
> >         if ("template".equals(protocol)) {
> >             uri = new InternalURI();
> >         } else if ("repository".equals(protocol)) {
> >             uri = new RepositoryURI();
> >         } else {
> >             return null;
> >         }
> >         uri.setURI(uriString);
> >         return uri;
> >     }
> >
> > }
>
> -------------------------------------------------------
> This SF.net email is sponsored by: Etnus, makers of TotalView, The debugger
> for complex code. Debugging C/C++ programs can leave you feeling lost and
> disoriented. TotalView can help you find your way. Available on major UNIX
> and Linux platforms. Try it free. www.etnus.com
> _______________________________________________
> Jpublish-developer mailing list
> [email protected]
> https://lists.sourceforge.net/lists/listinfo/jpublish-developer



-------------------------------------------------------
This SF.net email is sponsored by: Etnus, makers of TotalView, The debugger 
for complex code. Debugging C/C++ programs can leave you feeling lost and 
disoriented. TotalView can help you find your way. Available on major UNIX 
and Linux platforms. Try it free. www.etnus.com