Benchmarking a gui improvement


Recently I have been making some speed improvements to our own version of bui. We had found out that there is a part of the code that was especially slow by profiling. But now that I have made the improvements I can't verify that they actually help or how much.

In 3d/map making I usually just make a automated process of running the map and moving the camera through it while logging the fps, to see where performance is worst. But I have no idea how to do this with a gui and java, I'm thinking I want some fake mouse that moves over the screen the same way every time, maybe a series of use cases so the output is more presis.

Does any body have any experience with any thing in this genre?

No one has any idea's  :cry:

To start out with I had narrowed down the code in question to just one function, that way I could run both the old and the new function and via profiling, see witch was the fastest, but this time it is more of a structure thing and i would like to have some way of comparing the code revisions i.e. a benchmark, where the exact same actions are madeā€¦

If any one has the smallest idea or area to research. All the improvements are of course fully open source.

You can

a) trigger the same events that would be caused by the user by the unit test (not sure if that's possible in bui, though)

b) use java.awt.Robot to do what you described: moving the mouse the same way for each test run