[jboss-dev-forums] [Design of JBossCache] - Re: JBoss Cache Public API

[genman]

"manik.surtani at jboss.com" wrote : "genman" wrote : 
  |   | The problem with "putAll" (and "put" and "remove") is that they require a storage read for every call. If you can get them to perform okay, then I don't really care.
  |   | 
  | 
  | Why?  I agree that such calls should be optimised by the Interceptors not to do a load if the state needed is already in mem (for the return value of a put or remove), but beyond that, why would putAll, etc. do a storage read?
  | 

I don't see why they do, but they do.

For example, for FileCacheLoader, if erase is "false", which is called by "put(Fqn, Map)", then a read is made:

  |    @Override
  |    public void put(Fqn fqn, Map attributes, boolean erase) throws Exception
  |    {
  |       lock.acquireLock(fqn, true);
  |       try
  |       {
  |          Map m = erase ? new HashMap() : loadAttributes(fqn);
  |          if (m == null) m = new HashMap();
  |          if (attributes != null)
  |          {
  |             m.putAll(attributes);
  |          }
  |          storeAttributes(fqn, m);
  |       }
  |       finally
  |       {
  |          lock.releaseLock(fqn);
  |       }
  |    }
  | 

JDBC and all the other cache loaders are similar... I don't see why they should necessarily do a read for this operation.

View the original post : http://www.jboss.com/index.html?module=bb&op=viewtopic&p=4156881#4156881

Reply to the post : http://www.jboss.com/index.html?module=bb&op=posting&mode=reply&p=4156881