Viren a �crit :

> > Honestly, you will find errors using Pose long before you do on a
> physical
> > device.  Pose is designed to be harder to trick and since it uses
> an actual
> > device ROM, you know you are testing against the real thing.
> > I've found few exceptions as to where Pose doesn't behave exactly
> like a
> > Palm unit.
>
> I do agree with you over there...and I am not under-estimating the
> value of pose in palm-development. However, due to stringent QA rules
> in our company, it was made mandatory to do regression on the device
> before giving any intermediate/final release to clients. It was more
> due to the warning that palm folks have put on there site that before
> making any final release, device testing is a must.
>
> One thing that gremlin cannt do for you is to simulate pen-move
> events or other such non-UI-originated events....( atleast I am not
> aware of any such things! )...the product we were developing was all
> about vector map display and zooming,dragging it on palm...dragging
> the stuff via gremlin was impossible and hence the need of in-house
> tool that I talked about earlier.

I have the same problem with a cave map and I figured this could be
partly solved if I could first run POSE in recording mode, where pen
events would be stored for future playback. This sequence of events could
then replace the random-generated Gremlin events every time the same test
needs to be perform on a new version. It would also be very convenient to
trap a hard-to-reproduce bug: once you've reached the crash, just let the
Gremlin minimizing process find the smallest error sequence and then
debug your app. If you have spare time, POSE is open source...


--
Luc Le Blanc



-- 
For information on using the Palm Developer Forums, or to unsubscribe, please see 
http://www.palmos.com/dev/support/forums/

Reply via email to