Re: Make'em suffer!

[Manu Cornet <Manu Cornet GMail com>]

Hi Carl!

> I'm trying to make sense of the 3 different "expose" values reported
> by this tool. Can someone help me understand how they should be
> interpreted?

Here is what the torturer does:

1) Boot: this is a cycle where a widget is allocated, mapped, exposed,
and then immediately destroyed. This cycle is then repeated: the widget
is created/destroyed as many times as it is exposed.

2) Expose: the widget is created only once, then exposed many times, and
then destroyed only once.

3) Expose w/resize: just like 2) (only one creation/destruction), but
the widget is resized each time it is exposed again (from small sizes to
larger ones).

So the 3 different expose time values (let's say t_boot, t_expose and
t_resize) correspond to these 3 testing behaviors. If everything was
very logical, I guess we should have

	t_boot > t_expose
and	t_resize > t_expose

(I'm not sure about t_boot and t_resize) for most themes, but I can't
guarantee it will always be like this :)

> Meanwhile, it looks like both the XML and non-XML reports are
> incorrectly using the final "resize" expose value when trying to print
> the "boot/expose" value, (the non-XML report has the bug only in the
> average).
> 
> I've attached below the patch I used to fix this. The easiest way to
> get the patch is to just pull from my git repository with:

Thanks a lot! I've updated my own tree as well.

Cheers,
Manu