FreeLists / haiku-appserver / [haiku-appserver] Re: getting a clone of the current Accelerant

["Rudolf" <drivers.be-hold@xxxxxxxxxxxx>]

Stephan,

Please update the ATI driver!
Or at least recompile it with a small fix for cloning to work: all 
drivers failed back some time because of a fault in the R4 graphics 
driver kit:
the trailing '/' character for '/dev/' in CLONE_ACCELERANT:

        /* the data is the device name */
        /* Note: the R4 graphics driver kit is in error here (missing 
trailing '/') */
        strcpy(path, "/dev/");

was missing.

Rudolf.


> 
> On 2005-12-06 at 10:09:38 [+0100], Stefano Ceccherini <
> burton666@xxxxxxxxxxx> 
> wrote:
> > 
> > >I'm happy that we have the code in
> > >BWindowScreen for reference, but I have not looked into >it yet.
> > 
> > 
> > Note that it's a bit messy, I never found the chance to clean it 
> > up, and it 
> > involves using a private R5 class (_BAppServerLink_) to get the 
> > name of the 
> > accelerant and the driver used by the app server.
> 
> I found it to be readable! The code to get the accelerant path from 
> the 
> driver is nicely contained in AccelerantHWInterface (probably 
> initiating from 
> test_harness). I tried to combine the cloning in BWindowScreen and 
> AccelerantHWInterface, but didn't get it to work. With enough error 
> checking 
> I can prevent the system from crashing. But I'm uncertain about the 
> code in 
> BWindowScreen. It passes the path to the accelerant to the clone 
> function. I 
> would have guessed from the other functions (get_clone_info_size and 
> get_clone_info), that you first ask the accelerant for the clone info 
> size, 
> then create a buffer of that size, pass that to get_clone_info, and 
> than pass 
> that data without knowing what it contains to the clone function. I 
> tried 
> that too, but didn't get very far. I have to admit though that I'm 
> using a 
> very old version of the ATI driver, because of some particularities 
> with my 
> laptop (with that driver, I can just plugin a monitor and have the 
> desktop 
> cloned with no worries). Maybe this version doesn't support cloning 
> so well. 
> I will try with a current version and keep you posted.
> 
> I intend to keep at it, and further experiment with the "simple" 
> app_server 
> represented by my clipping prototype. I will try to make the thing 
> work with 
> transparent overlays (not video overlays...), to finally support drag 
> messages too. Maybe we ought to support the actual messages in the 
> real 
> app_server sooner than that though.
> 
> Best regards,
> -Stephan
> 
>