Re: [jdt-debug-dev] Issue relating to wrappering jdt debug model

["Darin Wright" <Darin_Wright@xxxxxxx>]

Please file a bug for this. We should be able to change the internal implementation to use a more efficient method lookup (for 2.1).

Darin

David Sanders <dsanders@xxxxxxxxxxx> Sent by: jdt-debug-dev-admin@xxxxxxxxxxx 10/11/2002 01:22 PM Please respond to jdt-debug-dev

To:        jdt-debug-dev@xxxxxxxxxxx         cc:                 Subject:        Re: [jdt-debug-dev] Issue relating to wrappering jdt debug model

Unfortunately, we do not know where we will end up.  One item on our
list is to reduce the number of java steps for each of our steps.
I can tell you that the findMethod(JdwpMethodID methodID) routine in 
org.eclipse.jdi.internal.ReferenceTypeImpl is called 
2,670 times in just two of our step into's.  It takes 46% of the 
cumulative 'step into' time.  It loops 259,260 times through an 
iterator rather than doing a hash lookup or something more efficient.  
The method.getMethodID().equals(methodID) is the expensive call in
that method (short of the number of iterations).  

Thanks,

David


Darin Wright wrote:
> 
> If you must "step and check" then there is not much you can do to
> improve performance. If you know where you want to end up, setting an
> "invisible" breakpoint would be more efficient.
> 
> Darin
> 
>  David Sanders
>  <dsanders@xxxxxxxxxxx>                      To:
>  Sent by:                             jdt-debug-dev@xxxxxxxxxxx
>  jdt-debug-dev-admin@xxxxxxxxxxx             cc:
>                                              Subject:
>  10/10/2002 05:11 PM                  [jdt-debug-dev] Issue relating
>  Please respond to jdt-debug-dev     to wrappering jdt debug model
> 
> Hello,
> 
> We are having some performance issues with our step into functionality
> of our wrapper for JDIThread.  The problem is that for every "step
> into" of one of our statements, we require several "step into" calls
> in the underlying JDIThread.  After each of these JDIThread step into
> calls, we must get the top stack frame and check to see if we should
> continue stepping into the JDIThread.  The problem is that each of our
> getTopStackFrame calls results in a call to
> org.eclipse.jdi.internal.ThreadReferenceImpl's frames(int,int) method,
> 
> which can result in a significant time penalty when having to do this
> many
> times.  Any suggestions?
> 
> Regards,
> 
> David
> _______________________________________________
> jdt-debug-dev mailing list
> jdt-debug-dev@xxxxxxxxxxx
> http://dev.eclipse.org/mailman/listinfo/jdt-debug-dev
_______________________________________________
jdt-debug-dev mailing list
jdt-debug-dev@xxxxxxxxxxx
http://dev.eclipse.org/mailman/listinfo/jdt-debug-dev