[Openmcl-devel] differences from os10.5.... to 10.6...

Gary Byers gb at clozure.com
Mon Aug 23 17:46:07 UTC 2010

Apple's 10.6 release notes:


don't seem to describe any changes in Core Animation, but they do mention
some under-the-hood changes to NSImage/CGImage/CGImageRef.

I don't know enough about this stuff to do anything but wave my hands around,
but I've seen cases where an OS release appears to break something that had
appeared to work before and on closer inspection it turns out that it had
"appeared to work before" almost by accident (e.g., that code was depending
on undocumented behavior or something like that.)

The code that I'd want to look at carefully in that context is the stuff
that sets the CALayer's content from an NSImage (if I understand correctly)
by way of a CGImageRef.  I don't know if that suspicion is well-founded, but
my first reaction is to wonder whether that code is correct (whether it might
depend on some external state that was set as expected in 10.5 but isn't 
in 10.6; the fact that these classes changed under-the-hood in 10.6 makes this
seem like a plausible theory, as does the fact that the CALayers don't seem
to all have valid contents in 10.6.

(Plausible theories aren't necessarily correct ones, and I don't know of any
concrete problem in that code, so this vague handwavy suspicion may turn out
to be completely unfounded.)

On Mon, 23 Aug 2010, Arthur Cater wrote:

> A month ago I was having trouble with demos (my own and Neil Baylis's) that were not running properly
> for some folk. I wrote to the list as copied below.
> Then my Darwin-64's disk failed. It was replaced for me while I was on holiday.
> So I took the opportunity of having the Leopard OS installed first, and tested the demos.
> Then I installed the Snow Leopard OS, and tested again. I used ccl 1.5-r13651  (DarwinX8664)
> in both cases. The demos did work properly under Leopard, and they do not work properly
> under Snow Leopard. Therefore I conclude that the problem is something to do with versions
> of MacOS and not the graphics card or other hardware - though conceivably with the driving of them.
> If somebody with insight into Leopard/SnowLeopard differences could have a look at these demos,
> which I attach, I would be most grateful. (I first made a small change to Neil Baylis's demo to make it
> use cg-floats).
> Arthur
> On 26 Jul 2010, at 09:40, Arthur Cater wrote:
>> Thank you for the feedback and suggestions.
>> I wonder if anybody is plannng anyway to upgrade any mac from 10.5.... to 10.6.... and onward to 10.6.4, would
>> they be so kind as to try out this demo and also Neil Baylis's CA-DEMO (in contrib directory), before and after
>> upgrade, in an effort to determine whether it is an OS or a hardware issue?
>> It is best to try these tests when not doing anything else of importance with your CCL, since
>> you may have to force quit if things go wrong.
>> Arthur
>> On Jul 23, 2010, at 5:23 PM, Raffael Cavallaro wrote:
>>> On Jul 23, 2010, at 12:01 PM, Arthur Cater wrote:
>>>> I've been told the demo of layered-view.lisp I contributed a few days back does not
>>>> work properly on a Darwin64 machine with Mac OS 10.6.3; it does work on my
>>>> own Darwin64 with 10.5.8.
>>> Just to add a data point or two, it fails in the same way in both the 32 and 64 bit IDE under 10.6.4 running on an Intel Core 2 Duo. This may also be a graphics card/driver issue since Core Animation is GPU accelerated. This machine has an NVIDIA GeForce 9400M.
>>> warmest regards,
>>> Ralph
>>> Raffael Cavallaro
>>> raffaelcavallaro at me.com
>>> _______________________________________________
>>> Openmcl-devel mailing list
>>> Openmcl-devel at clozure.com
>>> http://clozure.com/mailman/listinfo/openmcl-devel
>> _______________________________________________
>> Openmcl-devel mailing list
>> Openmcl-devel at clozure.com
>> http://clozure.com/mailman/listinfo/openmcl-devel
> _______________________________________________
> Openmcl-devel mailing list
> Openmcl-devel at clozure.com
> http://clozure.com/mailman/listinfo/openmcl-devel

More information about the Openmcl-devel mailing list