List:Eventum Development« Previous MessageNext Message »
From:Joao Prado Maia Date:March 11 2005 8:24pm
Subject:RE: Why eventum may be slow...
View as plain text  
Jeff, 

> The pie chart/graphical summary is also excellent, although I 
> found it 
> overwhelming on a lot of pages so (here it comes!) I made a couple of 
> changes:
> 
> * sort the list from "most time" to "least time"
> * escape % in labels because the piechart implementation 
> calls sprintf() 
> on them
> * replace tab/cr/lf in labels with space to keep SQL queries from 
> mangling legend
> * collapse whitespace in labels so we can see more detail
> * limit label length (before adding timings) to 40 chars
> * limit the total number of labels to 25 + an aggregate "Other"
> 
> Based on this, I have found "Included Validation Class" to be 
> the single 
> largest time sink item on virtually every page (~0.055) that 
> isn't a query.
> 

I think your changes make sense (and the graphical benchmark does look
better). However, I think we should scrap that pie chart and just create
a table with information, sorted by the order in the array. That way we
can visualize the code paths and see the difference in time between one
point and the next.

I'll accept patches for 1.5.2 on this :)

--Joao

Thread
Why eventum may be slow...Paul Mach10 Mar
  • RE: Why eventum may be slow...Joao Prado Maia10 Mar
    • Re: Why eventum may be slow...Paul Mach11 Mar
      • Re: Why eventum may be slow...Jeffrey D. Wheelhouse11 Mar
        • RE: Why eventum may be slow...Joao Prado Maia11 Mar
          • Re: Why eventum may be slow...Jeffrey D. Wheelhouse11 Mar
      • RE: Why eventum may be slow...Joao Prado Maia11 Mar
        • Re: Why eventum may be slow...Jeffrey D. Wheelhouse11 Mar
          • RE: Why eventum may be slow...Joao Prado Maia11 Mar
            • Re: Why eventum may be slow...Jeffrey D. Wheelhouse11 Mar
              • RE: Why eventum may be slow...Joao Prado Maia11 Mar
  • Re: Why eventum may be slow...Phillip Steinbachs10 Mar