It was a mistake, not done on purpose.
Fixed

On Sun, Mar 3, 2013 at 4:05 PM, sebb <[email protected]> wrote:

> On 2 March 2013 19:45,  <[email protected]> wrote:
> > Author: pmouawad
> > Date: Sat Mar  2 19:45:22 2013
> > New Revision: 1451926
> >
> > URL: http://svn.apache.org/r1451926
> > Log:
> > Bug 54268 - Improve CPU and memory usage
> > Bugzilla Id: 54268
> >
> > Modified:
> >
> jmeter/trunk/src/protocol/http/org/apache/jmeter/protocol/http/parser/RegexpHTMLParser.java
> >
> > Modified:
> jmeter/trunk/src/protocol/http/org/apache/jmeter/protocol/http/parser/RegexpHTMLParser.java
> > URL:
> http://svn.apache.org/viewvc/jmeter/trunk/src/protocol/http/org/apache/jmeter/protocol/http/parser/RegexpHTMLParser.java?rev=1451926&r1=1451925&r2=1451926&view=diff
> >
> ==============================================================================
> > ---
> jmeter/trunk/src/protocol/http/org/apache/jmeter/protocol/http/parser/RegexpHTMLParser.java
> (original)
> > +++
> jmeter/trunk/src/protocol/http/org/apache/jmeter/protocol/http/parser/RegexpHTMLParser.java
> Sat Mar  2 19:45:22 2013
> > @@ -27,11 +27,6 @@ import org.apache.jmeter.protocol.http.u
> >  import org.apache.jmeter.util.JMeterUtils;
> >  import org.apache.jorphan.logging.LoggingManager;
> >  import org.apache.log.Logger;
> > -
> > -// NOTE: Also looked at using Java 1.4 regexp instead of ORO. The
> change was
> > -// trivial. Performance did not improve -- at least not significantly.
> > -// Finally decided for ORO following advise from Stefan Bodewig (message
> > -// to jmeter-dev dated 25 Nov 2003 8:52 CET) [Jordi]
>
> Why did you remove the comments?
> I think they are still potentially useful (and they don't affect
> performance!).
>
> Please re-instate.
>
> >  import org.apache.oro.text.MalformedCachePatternException;
> >  import org.apache.oro.text.regex.MatchResult;
> >  import org.apache.oro.text.regex.Pattern;
> > @@ -146,15 +141,16 @@ class RegexpHTMLParser extends HTMLParse
> >       */
> >      @Override
> >      public Iterator<URL> getEmbeddedResourceURLs(byte[] html, URL
> baseUrl, URLCollection urls, String encoding) throws HTMLParseException {
> > -
> > +        Pattern pattern= null;
> > +        Perl5Matcher matcher = null;
> >          try {
> > -                       Perl5Matcher matcher = JMeterUtils.getMatcher();
> > +            matcher = JMeterUtils.getMatcher();
> >                         PatternMatcherInput input = localInput.get();
> >                         // TODO: find a way to avoid the cost of
> creating a String here --
> >                         // probably a new PatternMatcherInput working on
> a byte[] would do
> >                         // better.
> >                         input.setInput(new String(html, encoding));
> > -                       Pattern
> pattern=JMeterUtils.getPatternCache().getPattern(
> > +                       pattern=JMeterUtils.getPatternCache().getPattern(
> >                                 REGEXP,
> >                                 Perl5Compiler.CASE_INSENSITIVE_MASK
> >                                 | Perl5Compiler.SINGLELINE_MASK
> > @@ -199,6 +195,8 @@ class RegexpHTMLParser extends HTMLParse
> >                         throw new HTMLParseException(e.getMessage(), e);
> >                 } catch (MalformedCachePatternException e) {
> >                         throw new HTMLParseException(e.getMessage(), e);
> > -               }
> > +               } finally {
> > +            JMeterUtils.clearMatcherMemory(matcher, pattern);
> > +        }
> >      }
> >  }
> >
> >
>



-- 
Cordialement.
Philippe Mouawad.

Reply via email to