On Thursday 01 December 2005 19:16, Stephan Aßmus wrote: > On 2005-12-02 at 00:32:39 [+0100], Axel Dörfler <axeld@xxxxxxxxxxxxxxxx> > > wrote: > > Stephan Aßmus <superstippi@xxxxxx> wrote: > > > > Ah, so what about changing the app_server test environment to use > > > > the > > > > same tactics you wanted to implement for your clipping test app? > > > > This would benefit the whole app_server testing. > > > > > > The plan was to implement this in a way to use it there too. I think > > > we > > > would need only a "ClonedAccelerantHWInterface" and a BDirectWindow > > > that > > > handles some additional stuff. > > > > Sounds good :-) > > I hope it's doable. But since the ATI driver works with BWindowScreen, I > suppose it's accelerant is clonable in general. I just need to narrow down > that crash in "harness" and go from there. Since I worked out scrolling The harness makes it very easy to "freeze" your app_server. The issue deals with acquiring the engine and not releasing it (I forget the exact terminology since it has been so long since I've looked at the stuff). The harness has a command that acquires the engine. Whenever the app has the engine, the app_server cannot draw since it can't acquire the engine for itself. Apparently this somehow also interferes with the app_server processing input, so trying to tell the harness to release the engine doesn't work. Anyway, if you put together some code that acquires the engine, uses it, and then releases it without requiring user input in the middle, you should be able to use that sort of code just fine without freezing the app_server.