Mmm seem to me we need a process to inform the VM folks about bugs in  
the VM, versus thinking we carefully read all
the hundreds of messages on the mailing list. (well some of us do).  
However I was waiting a bit for someone to clarify
what the problem is and it's also nice to have a set of test cases too.

I'll roll the bug fix in primitiveDirectoryLookup into the pending  
3.5.1,

PS don't forget to test with aliases, those require special code to  
resolve which *I  think* is mostly bug free. Multiple alias
directory chains was the most painful, along with 255 byte names.  Ha,  
anyone care about maximum path lengths, one might get surprised about  
how long or short they really can be...

On Tuesday, May 20, 2003, at 08:15  PM, Stephen Pair wrote:

> That's correct...this test passes on Unix and Windows systems, but  
> fails on Mac OSX (and apparently 9.1 as well).  There is a bug in the  
> Mac implementation that should be corrected.  That was why I wrote the  
> test and posted the message.  The problem is that the Mac  
> implementation of the primitive FilePlugin>>primitiveDirectoryLookup  
> (dir_lookup() is the platform specific C function) is simply checking  
> to see whether there is an entry named like the FileDirectory without  
> regard to whether or not the entry is actually a directory.  As I  
> mentioned earlier, I think the behavior on Windows and Unix is  
> correct, and the Mac implementation needs to be corrected to conform  
> to the Windows and Unix implementations.
>
--
======================================================================== 
===
John M. McIntosh <johnmci at smalltalkconsulting.com> 1-800-477-2659
Corporate Smalltalk Consulting Ltd.  http://www.smalltalkconsulting.com
======================================================================== 
===