3 Replies Latest reply on May 22, 2009 11:19 PM by gonorrhea

    Seam 2.1.1 performance with debug="true"

    paultillotson

      With Seam 2.1.1.GA, I set debug="true" in the components.xml.


      <core:init debug="true" jndi-pattern="@jndiPattern@"/>


      I find that every HTTP request, whether for seam pages, seam pages with no components, static html pages, css and javascript files all take about 1.4 seconds.


      If I set debug="false", requests for static content take 10 or 20 milliseconds (nearly instant).


      What kind of performance difference do other people see when enabling debug? 


      Is this just a cost that must be paid for having debug="true", or is there something I can do to avoid this?

        • 1. Re: Seam 2.1.1 performance with debug="true"
          amitev

          Seam scans the WEB-INF/dev classpath for hot redeployment. To avoid this, simply use debug=false.

          • 2. Re: Seam 2.1.1 performance with debug="true"
            paultillotson

            Thanks Adrian!  If I do that, then the application must be restarted for every change.  If possible I'd like to have hot deployment enabled during development.


            I was hoping to hear from people who use debug/hot deployment during development about how fast it works for them. 


            If they say that they notice a small performance difference, clearly I'm doing something wrong and I should attempt to figure out whether I'm using the wrong JRE, version of JBoss, have two many components, etc....



            • 3. Re: Seam 2.1.1 performance with debug="true"
              gonorrhea

              JBoss scans the server/xxx/deploy directory for hot deployment of EAR, WAR and foo-ds.xml files.  IIRC you can tweak the interval for the scan (default <= 5000ms IIRC). 


              So perhaps you can tweak this setting as well if it is available in the Seam config/API.