On Wed, Dec 07, 2005 at 11:09:02AM +0100, Lukas Renggli wrote:
> I am usually using the Unix VM on OS X using the Quarz display driver,
> however most tests of OSProcess seem to fail using this configuration,
> whereas they perfectly run when using the X11 display driver.

The #forkHeadlessSqueak tests should succeed, and #forkSqueak tests
should succeed if you are running headless Squeak. I'm not sure what
the #forkSqueak tests do with the Quartz driver, but hopefully
they fail gracefully (let me know if this is not true).

Please use a recent version of OSProcess (4.0 or newer), as I added
some fixes to the SUnit tests to make them behave better on a Mac.

> I am wondering what the constraints are using OSProcess together with
> the Quarz display driver? Are #forkSqueak and #forkHeadlessSqueak
> supposed to work in this configuration?

I don't have a Mac, so I can't tell you what is required to
implement control of the display on Quartz. You can see how
it works on X11 by looking at the XDisplayControlPlugin.

With the display running on X11, #forkSqueak works by having
the child Squeak VM clean up all its references to the original
X display connection, then open a new one as if it were a new
image starting up. I'm sure a similar thing could be done with
Quartz or other external display media.

You can also look at Areithfa Ffenestri <http://minnow.cc.gatech.edu/squeak/3862>.
I think there is a Mac plugin, and perhaps it already contains the
required display control primitives.

Dave