[Cosmo-dev] Dojo widget/event performance benchmarking

[Matthew Eernisse]

Inline, below ...

Mikeal Rogers wrote:
> I'll add this to the list of requirements -- which is already fairly 
> long. It's something that's easier to get in early rather than later.

Agreed. This is a case where 'a stitch in time saves a pound of cure,' 
or whatever.

> One issue I see right way -- automated tests run much slower than 
> regular app use so we'll need to find a way of off setting targeted 
> speed goals with automation speed results.

Maybe we should focus on the speed of individual operations (e.g., the 
time it takes a dialog box to appear, the time it takes to render an 
event lozenge, the time it takes for the calendar UI to load and 
render). Those should not really differ between actual use and automated 
testing, should they?

A few other things to keep in mind as we implement this:

1. As much as possible, all the tests should work the same way in all 
our supported browsers, so we can get accurate per-browser perf numbers.

2. Windows apparently has crappy system-clock resolution, which will 
affect timing of operations that take a very short time. We may have to 
figure out how to automate the process of invoking some operations 
repeatedly so we can take the average and get an accurate performance 
number for IE.


Matthew