AppleTalk address collision with default settings

[rakslice]

In BII & SS, with OS versions 7.5, 7.6.1, 8.1 running OT 1.3.x Appletalk, with multiple instances connected with tap, the default state that the systems get into is to be all sitting on AppleTalk address 41201.173 (hex a0f1.ad) all broadcasting NBP lookups to the net for their own address (are these the EtherTalk equivalent of the LLAP Enquiry that is part of the dynamic node ID assignment process? Inside AppleTalk, Ch. 1) but no one acking anyone elses' requests, and of course the AppleTalk network is not usable for anything else because it is one giant address conflict.

Getting systems out of this state:

• Toggling Appletalk Inactive/Active in Chooser has no effect
• Shutting down and relaunching the individual systems has no effect
• Throwing away the AppleTalk prefs and shutting down and relaunching does cause that system to choose a new address
• Testing with 8.6, it chooses a new (randomized?) address on boot, so doesn't get into this state.

What's going on here? Something missing from the ether Mac driver implementation? Some XPRAM or other configuration problem? Other?

[rakslice]

• Reproduce an Appletalk address collision situation with a real Mac and see what it does
• If it works, figure out why this the same situation isn't working in BII/SS