Hi,

Thanks very much!  This is now in place at:

ftp://st.cs.uiuc.edu/Smalltalk/Squeak/3.5/mac/squeak3.5.1b4-mac-vm.sit.

cheers

bruce

John M McIntosh <johnmci at smalltalkconsulting.com> wrote:
> I've push a mac vm 3.5.1b4 into the update stream which should appear  
> on the servers in a
> day or two.  Early adopters can email for a copy.
> 
> 3.5.1b4 Please note the major changes to network code in 3.5.1b2
> 
> a)Added options SqueakWindowType & SqueakWIndowAttributes to  
> info.plist. The SqueakWindowType of 6 is the standard document window,  
> a type of 8 is the floating utility window. The attributes control with  
> window attributes and behavior according to
> http://developer.apple.com/techpubs/macosx/Carbon/ 
> HumanInterfaceToolbox/WindowManager/Window_Manager/wind_mgr_ref/ 
> constant_2.html#//apple_ref/c/tdef/WindowAttributes
> &
> http://developer.apple.com/techpubs/macosx/Carbon/ 
> HumanInterfaceToolbox/WindowManager/Window_Manager/wind_mgr_ref/ 
> constant_1.html#//apple_ref/doc/c_ref/WindowClass
> b) Sockets are not valid if the network is not initialized.
> c) The os-9 vm can return the open file error in the vm attribute data  
> 1202
> d) MyWindowEventMouseHandler was leaking region handles.
> e) Support for primitiveClosureValue, primitiveClosureValueWithArgs,  
> primitiveExecuteMethod
> f) when rootTableCount  goes > 2400 then allocationCount gets  
> allocationsBetweenGCs + 1
> g) BitBlt Alpha Blending has changed
> h) Added primitiveWrite24BmpLine
> i) The guard if statement for primitiveSerialPortRead was flawed, noted  
> the buffer overflow, but then proceeded, versus returning.
> 
> 3.5.1b3
> a) Fix, Stephane Ducasse <ducasse at iam.unibe.ch> pointed out that using  
> a virtual desktop we had a crash in ioSetFullScreen. This
> was due to a pthread race condition between the UI thread and interp.c.  
> Fixed by using UI callback to handle fullscreen window/move/resize.
> 
> 3.5.1b2
> a) Switch to aio.h logic and unix socket logic! This is a MAJOR  
> change!!!!
> 
> b) Call aioPoll  (select()) to cause process wait versus old pthread  
> logic. This follows in the footsteps of the unix port. You should  
> verify that
> cpu time when squeak is idle is not affected in a dreadful manner.
> c) Fix, if you call does this directory exists on a file now you will  
> get an error which will signal that the directory does not exist. This  
> ensures
> the FileDirectory test suite works.
> d) Fix, if you quit the browser plugin, check the status of the  
> pthread_cancel before attempting the pthread_join
> 
> 3.5.1b1 limited distribution
> --
> ======================================================================== 
> ===
> John M. McIntosh <johnmci at smalltalkconsulting.com> 1-800-477-2659
> Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
> ======================================================================== 
> ===