I've searched the list archives and googled the last couple of
days, but haven't found an answer to this problem which I however
assume is rather trivial. I'm porting a win32 application which has
the following basic structure:
mainloop:
poll & handle events
draw changes to window backbuffer
draw changes in backbuffer to screen
Actually the port has been done for a while, it was started on
Jaguar and finished on Panther, but because my own benchmarks gave
the result that QuickDraw was the most efficient way to draw
graphics I was simply using an NSQuickDrawView and then porting was
rather straight-forward. Now in Tiger it however seems that Quartz
has surpassed QuickDraw in speed (is this correct?).
For many operations Quartz may be faster than QuickDraw. For others,
it is not. It depends on what you are trying to draw.
For this application, however, the speed of your engine may be
eclipsed by the triple buffering scheme. The application would have
to move main memory from your back buffer to the windows back buffer
and then from the window's back buffer to the video card. That seems
like an awful lot of memory copying.
My problem with switching to Quartz (this problem was there also in
the port using QuickDraw, but I'd like to correct it now) is that I
would like to avoid having an "extra backbuffer", as each window
already has it's own backbuffer it feels unnecessary to have
another backbuffer allocated by the application into which all
drawing is first done which is then copied to the window's backbuffer.
Exactly.
Basically my question is: Is there any way to draw directly to the
backbuffer of the window, outside of the drawRect method in my
NSView? The problem is that while handling events the application
will need to update graphical elements, and I want to avoid having
an extra backbuffer.
You can use the "lockFocus" method of an NSView to activate its
graphics context (i.e. [myView lockFocus]. You can then draw into
the view and when you are done you should call [myView unlockFocus].
This is generally frowned upon because the system can, potentially,
synchronize the calls to drawRect with whatever else may be happening
in the system.
It is much more efficient to do ALL of your drawing in the drawRect
method. If something outside of that routine needs the display to be
updated, you should make a note of what needs to be drawn and call
[myView setNeedsDisplayInRect: someRect] (or any of the
setNeedsDisplay routines). At that point the system will make note
of the fact that your view needs drawing and will ensure that
drawRect is called at the most appropriate time.
Since your application seems to be doing some animation... another
option is to let your engine run and return to the event loop without
drawing or calling setNeedsDisplay. You can then set up an event
timer that fires at some reasonable time (say 30 times a second)
which calls the setNeedsDisplay on your view. Only when you get that
message, then, would you collect your graphics and draw them.
Scott
_______________________________________________
Do not post admin requests to the list. They will be ignored.
Quartz-dev mailing list (email@hidden)
Help/Unsubscribe/Update your Subscription:
http://lists.apple.com/mailman/options/quartz-dev/email@hidden
